# Brandmachine Help Documentation This file contains all documentation for the Brandmachine Shopify app. ================================================================================ # AI Metadata on Downloads When you download the **JPEG (sRGB)** version of an image from Brandmachine, it carries a small piece of **metadata that declares it as AI-generated**. It's written into the file automatically, so images you hand to marketplaces, retailers, or platforms that require AI disclosure are compliant the moment you download that file. You don't have to do anything beyond picking that download option. ## Why this matters A growing number of marketplaces and platforms now require AI-generated product and campaign images to be labelled as such **in the file itself**, not just declared in a form. An image that arrives without that label can be rejected on upload. Brandmachine adds it for you automatically on the JPEG (sRGB) download, so your catalog pipeline keeps moving. ## What gets written We use the open **IPTC "Digital Source Type"** standard, the same vocabulary the press and stock-photo industry uses. The value marks the image as fully AI-generated: ``` http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia ``` It sits in the image's XMP metadata and is written **last**, after all resizing and colour conversion, so it survives the export intact. Your downloads are also tagged with a standard **sRGB colour profile** so colours stay correct in print and design tools. ## Which images are covered Right now, the label is added only to the **JPEG (sRGB)** download, generated at download time (for example, from Workbench). If you need the AI-labelled file for a marketplace, choose **JPEG (sRGB)** as your download format. Other download options, including the lossless PNG master, don't carry the label yet. If you need a lossless copy for your own archive, keep the PNG master separately and use the JPEG (sRGB) file for any submission that requires the AI disclosure. ## How to verify it yourself If a marketplace asks you to confirm the label, or you just want to check, you can read it back with the free [ExifTool](https://exiftool.org) utility: ``` exiftool -XMP-iptcExt:DigitalSourceType your-image.jpg ``` You should see: ``` Digital Source Type : http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia ``` That confirms the image is correctly labelled as AI-generated. ## Good to know - The label is a statement about **how the image was made**. It does not change how the image looks. - It is machine-readable, so a marketplace's automated intake can detect it without anyone reviewing the image by hand. ================================================================================ # Billing & Credits Brandmachine bills usage in one of two ways, depending on your account. **If your shop is on a paid Shopify subscription plan** (Free, Test Drive / Starter, Launch Promotion / Growth, Enterprise, or an internal plan), that plan covers your usage. Your generations and releases are tracked for audit purposes only, they are never debited from a credit balance or blocked for lack of one. The credit system described on this page doesn't apply to you. **If you're a standalone team that signed up directly, or a Shopify shop without a paid subscription plan**, Brandmachine runs on **pay as you go credits**. You top up a credit balance and each generation or release deducts from it. The rest of this page covers that system. ## Where to Find It Open **Settings → Billing**. You can also tap the credit balance pill near the bottom of the sidebar to jump there directly. ## What the Billing Page Shows **Usage this month.** A stacked bar chart breaking down spend into Generation and Release. Expand "By feature" to see how much went to Campaign, Product Studio, and so on. **Credit balance.** Your current balance in euros. **Top up.** Preset amounts (€50, €100, €150, €200) plus a custom amount field. Allowed range is €5 to €500. **Transaction history.** A paginated ledger of every charge and top-up, newest first. Columns: date, description, cost, amount (signed delta to balance), balance after. ## Topping Up There are two top-up paths depending on how you signed up. ### Shopify App Users If you installed Brandmachine through the Shopify App Store, top-ups go through Shopify's one-time charge flow: 1. Pick a preset or enter a custom amount. 2. Click **Continue with [amount]**. 3. You're redirected to Shopify Admin to approve the charge. 4. After approval, you land back in Brandmachine and your balance updates. ### Direct (Stripe) Users If you signed up directly outside of Shopify, top-ups use Stripe Checkout: 1. Pick a preset or enter a custom amount. 2. Click **Continue with [amount]**. 3. Stripe Checkout opens, with Stripe Tax and optional VAT-ID collection. 4. After payment, Brandmachine processes the webhook. You'll see "Processing your payment" briefly while the balance settles, then a **Top-up successful** banner. If a payment doesn't complete, the page shows a warning banner with a **Try again** button. The system re-checks your balance before re-enabling the retry, since the webhook sometimes lands a few seconds late. ## Low Credit Banner When your balance drops below **€2.00**, a banner appears at the top of every page (except the Billing page itself) telling you credits are running low, with a **Top up now** button. You can dismiss it for the session. The banner never appears for teams on an active paid Shopify plan, or for invoiced teams, since neither is billed through the credit balance. ## Price List Click **Price list** in the top-right of the Billing page to see what each operation costs. Charges are split into two tiers: - **Generation.** Charged when you generate an image, video, design revision, etc. The rate is multiplied by the number of images actually delivered, so an operation that returns several output images costs more than one that returns a single image. - **Release.** One-off fees applied when you release a final image to high-res or publish it. The Price list page stamps the current pricing version (for example `v20260611`) at the top. Historical ledger entries show the price stamped at the time of the charge, so old transactions stay accurate even when prices change. ## Invoices and Payment Methods (Stripe) Stripe-billed teams see a **Manage billing & invoices** button in the top-right of the transaction history. Clicking it opens the Stripe Customer Portal where you can: - Download invoices - Update payment methods - Update your VAT ID and billing address - Review your full payment history Shopify-billed teams manage payments through Shopify Admin instead. ## FAQs **Do credits expire?** No. Your balance carries forward indefinitely. **What happens if I run out mid-generation?** Each operation gates against your balance independently. If a single charge would put you below zero, that operation is blocked until you top up. Already-running tasks finish. **Can I get an invoice?** Yes. Stripe issues an invoice for every top-up and emails it to your billing contact. You can also download invoices any time from the Stripe Customer Portal via **Manage billing & invoices**. **What about VAT?** Stripe handles VAT automatically at checkout. EU customers can add a VAT ID either at checkout or later via the Stripe portal to switch to reverse charge. **I need volume pricing or a dedicated plan.** Reach out to [support@brandmachine.shop](mailto:support@brandmachine.shop) and we'll work out a custom arrangement. **I think I was charged in error.** Contact [support@brandmachine.shop](mailto:support@brandmachine.shop) with the date and description from the ledger. We'll look into it. ================================================================================ # Brand Memory Brand Memory lets you define a consistent brand persona that Brandmachine uses across AI-generated media and prompt suggestions. Once it's set up, you don't need to re-describe your brand every time you generate an image or video, Brandmachine references it automatically. In the sidebar, this lives under **Shared Resources → Brand Memory**. You may still see it called **"Brand Identity"** in a few places, most notably the tile label on the Settings page. How much it helps depends almost entirely on how it's written. For what to actually put in the box, see [Writing a Good Brand Identity](/brand-identity-writing-guide). --- ## Creating Your Brand Memory 1. Go to **Settings**. If you haven't set one up yet, you'll see a **"Setup Brand Identity"** tile (once a profile exists, this becomes **"Edit Brand Identity"**). 2. Click it. A **"Discovering Brand"** popup appears, prompting for your brand or store name (Brandmachine tries to suggest one from your store name automatically). 3. Confirm with **"Yes, create Brand Identity"**. Brandmachine drafts a profile describing your brand's style and tone. 4. You'll land on the edit page, where the generated description appears under **"Core Values & Identity"**, a multiline text box. The generated draft is written from your store name alone, without ever having seen your products, so treat it as a starting point rather than a finished profile. ## Editing Your Brand Memory Come back to this page any time from **Settings** or from **Brand Memory** in the sidebar. There are three things to edit: - **The brand name**, using the pencil icon next to it. This is for your own reference. It's never printed into a generated image. - **Core Values & Identity**, the main description. This is the text that shapes generated media. Be specific about visual style, colour, casting, and setting rather than abstract about values. - **Fashion Model Casting Instructions**, rules that apply whenever Brandmachine casts a model for you: age range, grooming, makeup, hair, accessories. Optional. Leave it empty and a sensible default standard applies. A save bar appears as soon as you change something, with **Save** and **Discard**. There's currently no "Regenerate" button, if you want a different starting point, edit the text box directly. You can also ask the agent in chat to update your Brand Memory for you, in plain language. A store keeps one Brand Memory in practice, Brandmachine uses your most recently created profile. It isn't a hard limit enforced by the app, but you shouldn't rely on creating multiple profiles and switching between them. ## Where Brand Memory Is Used Brand Memory is read whenever Brandmachine has to make a creative decision you haven't made yourself. In practice that means: - **The agent's proposals in chat.** When you ask for a campaign and don't spell out the concept, the agent grounds what it suggests in your brand rather than in generic fashion photography. - **[Fashion Models](/fashion-models-getting-started).** Casting a new model and refining an existing casting both draw on it for styling, hair, makeup, and mood. - **[Campaign Composer](/campaign-composer-getting-started).** It shapes the mood, atmosphere, and styling direction of a campaign image, and the pose and expression notes when you refine one. - **[Set Locations](/set-locations-getting-started).** When you refine a location, your brand steers the lighting, mood, and overall feel of the scene. - **[Shot Planner](/shot-planner).** Used mainly to complete bare looks in your brand's spirit, for example pairing a lone skirt with your staple top. - **Product Spotlight.** The scene ideas suggested for a product are proposed on brand rather than at random. - **Reference images you upload.** When Brandmachine analyses an image you've provided, your brand gives it context for what to pay attention to. A few places deliberately don't use it. **[Product Studio](/product-studio-getting-started)** perspectives are driven by your Product Studio settings and instructions instead, so catalogue styling rules belong there. **[Designer](/designer)** and **[Patterns](/patterns)** work from the inspiration and references you attach to each design. ## Good to Know - Generation reads your Brand Memory live, at the moment you generate something. Editing your Brand Memory does not retroactively update media you've already generated, only new generations pick up the change. - You don't need to re-enter brand details in each feature. Once Brand Memory is set, it's applied automatically wherever it's used. - If you write a prompt or copy manually, your input is respected, Brand Memory only shapes auto-generated suggestions. - Brand Memory sits below what you type in chat and below your saved Skills. It fills the gaps when nothing more specific applies, so it never fights a direct instruction. [Shot Planner](/shot-planner) has the full priority order. - Your brand name and logo never appear inside a generated image. Scenes are kept free of brand marks and signage on purpose. ================================================================================ # Writing a Good Brand Identity Your [Brand Memory](/brand-identity-getting-started) is the text Brandmachine reads before it proposes a scene, casts a model, or completes a look. A vague one produces generic photography. A specific one produces photography that looks like yours. This page is about what to actually write in the two text boxes on the Brand Identity page. --- ## The one rule **Write things a photographer could act on.** Everything that reads your Brand Memory is building an image. It can act on a colour, a fabric, a backdrop, a kind of face, a quality of light. It cannot act on a value. Compare these two lines. Both describe the same brand: > We believe in authenticity, timeless quality, and craftsmanship that lasts. > Garments are shot with the weave visible and the retouching light, on models with unstyled hair and natural skin, against worn concrete and painted brick rather than seamless studio backdrops. The first one is true and says nothing an image can be built from. The second one produces a photograph. Write the second kind. You don't have to drop your values, but every value should be followed by what it looks like. "Sustainable" becomes "undyed and low-wash fabrics, visible mending, muted natural colours." "Premium" becomes "deep shadow, single-source light, no props competing with the garment." --- ## What to cover Aim for roughly 250 to 350 words in **Core Values & Identity**, organised under a few headings. Cover these four areas: **Who the brand is.** Two or three sentences on where it comes from and what it stands for. Keep it short. This is the part that grounds the rest, not the part that does the work. **The clothes.** Name your actual silhouettes and fabrics. "The boxy chore jacket, the wide-leg trouser, the ribbed tank." "Heavyweight organic cotton, washed linen, undyed merino." Named garments are what let Brandmachine style a complete look that looks like your catalogue instead of a generic one. **Casting and setting.** Who wears the clothes and where they're shot. Not demographics, but a picture: "women in their thirties with unstyled hair and minimal makeup, more artist than model." For setting: "coastal light, weathered timber, empty interiors with big windows." These lines carry the most weight in campaign and set-location work. **Colour and light.** Name colours, not moods. "Bone, oatmeal, deep navy, a rare rust accent" beats "a warm, earthy palette." Say how it's lit: overcast daylight, hard midday sun, a single warm lamp. Say how it's graded: cool and slightly desaturated, or warm with lifted blacks. --- ## Name your staple basics This is the single highest-value sentence in your Brand Memory, and most brands leave it out. When you shoot a skirt with no top, or trousers with no shirt, Brandmachine has to complete the outfit so the model isn't half-dressed. The complement it adds is always neutral and unbranded so your real product stays the hero. Left to itself it will pick something sensible. Told what you'd pick, it picks yours. So write the line: > Bare bottoms are completed with a plain white ribbed tank. Bare tops go with straight-leg indigo denim. Footwear is a low leather sneaker in bone, never a dress shoe. That one sentence changes what comes out of [Shot Planner](/shot-planner) on every product you upload with a single photo. It's used as a fallback, so anything you say in chat or save as a Skill still overrides it. See the priority order in [Shot Planner](/shot-planner). --- ## A worked example A short, complete Brand Memory for a made-up label: > **Nordlys** > > **The brand.** Nordlys makes everyday wardrobe pieces in Copenhagen, cut for wear rather than for the season. The clothes are meant to be kept for years, so nothing is trend-loud and nothing is disposable. > > **The clothes.** The core is the oversized poplin shirt, the pleated wide-leg trouser, the ribbed knit tank and the unstructured wool coat. Fabrics are washed linen, heavy poplin, undyed merino and organic cotton with visible texture. Fits are generous and hang away from the body. > > **Casting and setting.** Models read as real people rather than fashion mannequins: women and men from mid-twenties to mid-forties, unstyled or loosely tied hair, minimal makeup, hands and skin left natural. Locations are quiet and unpeopled: bare Nordic interiors with tall windows, pale wood floors, coastal dunes, concrete stairwells. No studio seamless, no props that compete with the garment. > > **Colour and light.** The palette is bone, oatmeal, stone grey, deep navy and black, with rust or moss as the only accents. Light is soft overcast daylight from one side, with long shadows rather than fill. Grading is cool and slightly desaturated, blacks left soft, skin kept warm. > > **Staple basics.** Complete a bare bottom with a plain bone ribbed tank. Complete a bare top with straight-leg indigo denim. Shoes are a low bone leather sneaker or a bare foot, never a heel. That's about 240 words and every line of it can be photographed. --- ## Fashion Model Casting Instructions The second box on the page is narrower. It applies whenever Brandmachine casts a model for you, and it's the right home for rules that should hold on every model regardless of the shoot. Good things to put there: age range, grooming and beard rules, makeup level, hair, jewellery and accessories, body types you want represented, anything you never want to see. Write it as short direct rules, and split by gender where the rules differ. > For all models: skin left natural with visible texture, no heavy smoothing. Hair loose or simply tied, no elaborate styling. > Men: clean-shaven or short stubble, no full beards. No jewellery. > Women: minimal makeup, no strong lip. Small studs only. Leave it empty and Brandmachine applies a sensible default grooming standard, so this box is optional. Fill it in if you have opinions about faces. --- ## Common mistakes **Marketing copy.** Your About page, your brand deck, and your tagline are written to persuade a customer. They tend to be all values and no surface. Rewrite them into what the camera sees before pasting them in. **Too long.** Past roughly 400 words the important lines start competing with each other. If you catch yourself writing a fourth paragraph about heritage, cut it and spend those words on colour and casting. **Contradictions.** "Minimal and clean" in one paragraph and "maximalist, layered, chaotic" in another leaves Brandmachine to pick one, and it may not pick the same one twice. Read the whole thing through and resolve the fights. **Naming the brand as a visual.** Your brand name is stored for your reference but it's deliberately kept out of generated scenes: no logos, signage, or labels are invented into a shot. Describing your logo or your packaging won't put them in an image. **Copying a reference brand.** Naming another label as shorthand ("like Toteme but cheaper") gives Brandmachine nothing to work with, since it can only use what's actually written down. Describe the look directly instead. --- ## Keeping it current Brand Memory is read live, at the moment you generate. Editing it changes everything you make from that point on and leaves what you already generated untouched. That makes it worth revisiting. If a season shifts your palette, or you keep correcting the same thing in chat, put the correction into your Brand Memory so you stop repeating it. A rule you've typed into chat three times belongs either here or in a Skill. You can also just ask the agent in chat to update your Brand Memory. Describe the change in plain language ("our palette moved to warmer neutrals this season") and it will revise the text for you. --- ## FAQs **How long should it be?** Roughly 250 to 350 words for Core Values & Identity. Long enough for four short sections, short enough that no line gets lost. **Can I include hex codes, fonts, or my logo?** You can write them, but they won't be used. Brand Memory drives photography, not graphic design, so colours are best given as names ("bone, deep navy, rust") rather than hex values, and generated images never contain your logo or brand name. **Does it override what I ask for in chat?** No, it's the opposite. What you type in chat wins, then your saved Skills, then Brand Memory, then Brandmachine's defaults. Brand Memory is what fills the gaps when nothing more specific applies. See [Shot Planner](/shot-planner) for the full order. **Why doesn't my Product Studio shot follow my Brand Memory?** Product Studio perspectives are driven by their own settings and instructions rather than by Brand Memory, so styling rules for catalogue shots belong in your Product Studio settings. See [Product Studio](/product-studio-getting-started). **Do I have to write it myself?** No. Brandmachine drafts one from your store name when you first set it up, and that draft is a reasonable starting point. It's written without ever having seen your product, though, so it will be generic until you replace the abstract lines with your real garments, colours, and casting. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Campaign Composer Campaign Composer is your AI photo studio for campaign-quality imagery. Each **campaign image** combines a shooting location, a model, products, and styling, just like a real photo shoot. Multiple campaign images belong to a **campaign**. For multi-person scenes that combine separate single shots into one shared scene, see [Group Shots](/group-shots). --- ## Anatomy A campaign has: - A title (editable inline) - **Single shots** (one model per shot) - **Group shots** (multiple models composited into a shared scene) - **Unassigned single shots** (single shots not yet grouped into a campaign view) Home shows a 7-day **Recent campaigns** strip. The full list of every campaign lives on the separate **All Campaigns** page. --- ## 1. Prerequisites Before generating, you'll typically want: - At least one library **fashion model** - At least one **set location** - Products you can pull from your Shopify catalog You can also create models and locations on the fly during the workflow. --- ## 2. Create a Campaign Image 1. Click **New Campaign** to start a campaign. 2. Inside the campaign, click **Create Single Shot** to start a new single shot, or **Add Existing** to assign an unassigned single shot to this campaign. Once you have at least two completed single shots, **Create Group Shot** also becomes available (see [Group Shots](/group-shots)). 3. A new single shot opens with a persistent **Inspector** panel on the right, where you set up the shot in two steps: - **Step 1 (required):** - **Shooting location.** Pick a saved set location, or create one inline. - **Fashion model.** Pick a library model. - **Products.** Choose products from your Shopify **Catalog**, or **Upload** files directly. The Catalog picker is disabled when the app isn't running inside a Shopify context. - **Step 2 (optional):** - **Styling reference.** Add narrative direction as text (pose, accessories, expression, additional garments, mood, e.g. "Walking forward, hands in pockets, slight smile") or an image (drop a reference image and let Brandmachine extract the styling cues). - **Aspect ratio.** Pick from the aspect ratio selector based on where the shot will run (1:1, 4:3, 3:2, 16:9, 21:9, 5:4, 4:5, 3:4, 2:3, 9:16). 4. Click **Generate Preview** to check composition first, or go straight to **Generate Images**. You can rename a campaign image's title inline at any time. --- ## 3. Choosing Your AI Models Before generating, you can control which AI model (or models) render your images from **Settings > Campaign** (`/settings/campaign`): - **Gemini 3 Pro**: best for editorial, atmospheric shots with strong narrative direction; weaker on print and fabric detail. - **GPT Image 2**: best for apparel, prints, and fabric detail; less reliable for narrative or atmospheric scenes. At least one model must stay on. With both enabled ("mixed mode"), a generation produces 4 images, 2 from each model. With a single model enabled, a generation also produces 4 images, all from that model. Your current selection appears as a read-only "Generating with" badge row wherever you generate images. --- ## 4. Generate Preview vs. Generate Images - **Generate Preview** is a free, low-res (1K) preflight so you can sanity-check composition, location, and styling before committing to a full generation. Preview results are marked with a badge and cannot be released or downloaded in high-res. - **Generate Images** runs a full-quality generation batch. Once you have a batch, use **Generate More** to add another batch with the same inputs. --- ## 5. Curate Through Selection Campaign Composer uses **selection by elimination**, like culling shots from a real shoot. 1. Review the batch. 2. Click images you don't want, then **Remove Selected**. 3. **Generate More** to add another batch. 4. Favorite the ones you like to mark them. Switch between **All Images**, **Favorites**, and **Rejected** tabs at the top of the gallery. --- ## 6. Card Actions Clicking a tile opens **Zoom**, a full-size view with a toggle between full-width (scrollable) and fit-to-viewport display. The two icons visible on hover are: - **Compare**: split-screen the result against the original product photos. - **Favorite**: mark the result as a favorite. Everything else lives in the **"..." (more actions)** menu on each tile: - **Reject**: move to Rejected (recoverable). - **Download / Release**: release and download the high-res version. Preview results show "not releasable" here instead. - **Edit with AI**: open the [Photo Editor](/photo-editor) for color correction, spot fixes, or lasso-driven generative edits. This option may not be available to everyone, it's feature-flag gated and may not appear on your account. --- ## 7. Group Shots To composite two or three single shots into a shared multi-person scene, see [Group Shots](/group-shots). They live nested inside a campaign, and become available once the campaign has at least two completed single shots. --- ## 8. Publishing Once a campaign image is final, release and download the high-res version from the card's more-actions menu. From there you can use the file however you publish product imagery today, for example uploading it into your theme editor or a marketing tool. --- ## 9. Best Practices - **Build a library first.** Prepared models, locations, and brand identity make every shoot faster and more consistent. - **Be specific with styling.** Pose, accessories, expression, mood. Vague prompts produce vague results. - **Use Generate Preview first** when you're unsure about a location, model, or product combination, then move to Generate Images once it looks right. - **Cull aggressively.** Generate more batches and remove what doesn't work. The good shots stand out fast. - **Use the same model and location** across products in one campaign for a cohesive look; vary styling and aspect ratio per placement. --- ## 10. FAQs **Can I use Campaign Composer for non-fashion products?** It's optimized for fashion. For e-commerce catalog shots, use [Product Studio](/product-studio-getting-started). For pure product videos, use [Product Videos](/product-shorts-getting-started). **What if generation fails?** Rarely, a generation may fail due to content filters or a transient error. Retry or adjust inputs. Contact support if it persists. **Can I save without publishing?** Yes. Campaign images live in your campaign indefinitely, and you can download the high-res version whenever you're ready. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Group Shots Group Shots composite two or three single-shot portraits into a single multi-person scene while preserving each model's identity. Use it when you want models who exist on separate campaign images to appear together in the same final shot. --- ## When to Use a Group Shot - You've generated single shots for several models in the same campaign and want a "together" version. - You want a hero image with a small ensemble that wouldn't be feasible to shoot as one scene. - You need consistent styling across multiple models from one campaign. --- ## 1. Create a Group Shot 1. Open a campaign in **Campaign Composer**. 2. In the **Group Shots** section, click **Create Group Shot** (or open an existing tile to edit it). 3. You're taken to the group shot detail view, with the inspector panel on the right. --- ## 2. Configure the Composite ### Pick Single Shots (2 to 3) The inspector lists the single shots already in this group shot and the available shots in the campaign (completed single shots not yet in this group). Click a thumbnail to add it. Click the **X** on an added shot to remove. You need at least 2 and at most 3. ### Shooting Location (optional) By default, the group shot inherits the parent campaign's location. Click **Select** (or **Edit**) to pick a different one. The location determines the shared environment in the composite. ### Styling Instructions (optional) Free-text guidance for poses and interaction. For example: "Walking together through the city, laughing, relaxed poses." ### Aspect Ratio Pick the output dimensions for the composite (1:1, 4:5, 16:9, etc.). --- ## 3. Generate Click **Generate Group Shot**. The button is disabled until you have 2 or 3 single shots selected. Generation shows a spinner; results appear in a grid when ready. To make more variants with the same inputs, click **Generate More**. --- ## 4. Review and Refine Each composite has the standard image card actions. **Favorite** is available directly on the tile. Everything else lives in the **"..." (more actions)** menu: - **Reject**: move to Rejected (recoverable from the Rejected tab). - **Download / Release**: upscale to high-res and download. - **Edit with AI**: open the [Photo Editor](/photo-editor) for fine-tuning. This option is feature-flag gated and may not be available on your account. Switch between **All Images**, **Favorites**, and **Rejected** tabs at the top of the gallery. --- ## How Identity Consistency Works Each selected single shot acts as a portrait reference for its model. The compositor uses those references to place every model into the shared scene while keeping facial identity, hair, and styling consistent with the original single shots. You don't have to do anything special to enable this; it's built into the flow. **AI models.** Group Shots follow the Campaign's AI model setting from **Settings > Campaign**, with one exception: the initial per-person clean render (Phase 1) always uses GPT Image 2, regardless of which model or models are enabled. The composite step (Phase 2), where models are placed into the shared scene, respects whichever model or models are enabled for the campaign. --- ## FAQs **Can I use models from different campaigns?** A group shot pulls from single shots in the same campaign. To mix across campaigns, recreate the needed single shots in one campaign first. **Why not 4+ models?** The current limit is 3 for compositional consistency. We may expand this later. **What if a model's identity drifts?** Reject the result, regenerate, or pick a single shot with a clearer face for that model. The clearer the reference, the better the composite. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Designer Designer is where you create new garment designs in Brandmachine. You assign a fashion model, give the AI agent inspiration (mood shots, fabric photos, written notes), and optionally attach a fabric pattern. Each request to the agent produces a revision that shows the garment on the model from the front and back. From there you can add technical drawings (flat sketches) and take garment measurements off them. --- ## Anatomy of a Design A **design** consists of: - A **title** (editable inline) - One assigned **fashion model** (locked once set) - **Inspiration resources** (images and notes) - Optional **attached patterns** (see [Patterns](/patterns)) - One or more **revisions**. A revision starts with two photo tiles: - **Front**: garment on the model, front view - **Back**: garment on the model, back view - Generating **technical drawings** adds two more tiles, **Front flat** and **Back flat**, for a four-tile grid. - **Measurements** taken from the flat sketches (optional, available after technical drawings exist) --- ## 1. Create a Design 1. Open **Designer** in the sidebar and click **New Design**. 2. On the right panel, work through the setup steps. ### Pick a Fashion Model (required) On the **Fashion Model** card, click **Pick a Fashion Model**. A modal shows all library models. Select one and click **Select model**. The chip on the right panel then shows the model's thumbnail, name, and gender. The fashion model is **locked once it's set.** The **Change** control is disabled, with the tooltip "The fashion model can't be changed once it's set. Start a new design to use a different model." If you want a different model, start a new design. ### Add Inspiration Resources (optional) Click **+ Add** on the **Inspiration Resources** card. In the **Add Inspiration Resource** modal, pick a **Type**: - **Image**: an inspiration image such as a worn outfit, a flat, a mood shot, or a fabric photo. Brandmachine describes it automatically, so a label is optional. - **Note**: a short written note covering a fabric, colorway, or styling cue (for example "heavy silk" or "deep emerald"). A note requires both a **Label** and a **Description**. Click **Add resource**. Resources appear in the panel, and you can delete any with the trash icon. Brandmachine analyzes each image resource with Gemini Vision so the agent has a structured understanding of what makes it visually distinct. ### Attach Patterns (optional) If you've created any in [Patterns](/patterns), click **+ Add** on the **Patterns** card. A two-step picker opens: pick a pattern, then pick a specific revision of that pattern. Attached patterns show their name and repeatability (for example "Tileable both axes"). Remove with the trash icon. ### Add Outfit Context (optional) The **Outfit context** card describes the rest of the outfit around the hero garment (for example wide-leg trousers and loafers under a top). Leave it empty for dresses or full-length pieces. Click **Edit** to write it, then **Save**. ### Edit the Description (optional) Below the setup, the Description card starts blank. After the first agent turn, the agent co-authors a description. Click **Edit** to refine it. --- ## 2. Start Designing When you have at least a fashion model, click **Start designing** at the top right. The inspector opens with the agent chat, and your first message kicks off generation. Once a design has revisions, the same button reads **Continue designing**. Each request from you produces a new revision. The revision strip at the top shows your revisions horizontally, newest first, with a pin badge on favorites. --- ## 3. The Revision Each new revision starts with two photo tiles, **Front** and **Back**, showing the garment on the model. Tiles fill in as they generate, with a spinner labeled "Generating" until done. Revision actions (top right of the revision card): - **Generate technical drawings**: adds the flat sketches (see below). - **Take measurements / Re-run measurements**: opens the measurement flow, enabled once technical drawings exist. - **Pin revision**: toggles favorite state. - **Delete revision**: removes it. ### Paint and Regenerate To fix one area without regenerating the whole look, use the paint tool on a Front or Back tile ("Paint on Front to regenerate"). Brush over the region, add an optional note like "wider collar", and click **Regenerate** to redo just that area. A **Painted on front** or **Painted on back** pill marks the result. --- ## 4. Generate Technical Drawings Photo tiles show the garment worn. Technical drawings are the flat sketches used for production. This is a separate step, and it is billed separately from the initial revision. Click **Generate technical drawings** on a revision. Once the front and back photos are ready, this adds a **Front flat** and a **Back flat**, turning the revision into a four-tile grid. On flats that have measurements, a **Plain** / **Annotated** toggle appears: **Plain** is the raw sketch, **Annotated** overlays the measured points and labels. --- ## 5. Take Measurements Measurements give you accurate garment dimensions off the flat sketches. **Take measurements** is only enabled after technical drawings exist. Before then it is disabled with the reason "Generate technical drawings first". The flow runs in two steps (front, then back): 1. Click **Take measurements** on a revision that has flats. 2. **Front.** Drag the two pins on the front flat to mark a known-length reference (for example a seam or a stripe). Enter the **Reference length (cm)** and an optional label. Click **Next**. 3. **Back.** Repeat on the back flat. 4. Click **Generate measurements.** The modal closes, and processing runs as an async job (Gemini Vision), usually a minute or two. When it's done, a **Measurements** card appears below the grid with two columns, **Front** and **Back**. Each row has a **Label** and a **Value**, both editable. Values are in centimeters, and new rows default to cm. You can: - Add rows for anything the AI missed. - Edit values to refine the AI's reading. - Click **Save and re-annotate** to push your edits back through annotation, or **Discard changes** to revert. If the reference was off, click **Re-run measurements** to redo the two-step flow. --- ## 6. Iterate Use the agent chat to request changes: "make the sleeves a bit shorter", "try a contrast collar", "use the moth wing pattern". Each request generates a fresh revision. Pin the ones you want to keep visible. You can also use the **Patterns** card to swap fabric prints in and out of the agent's context between turns. --- ## 7. Collections Designs can be grouped into **Collections**. A collection carries a **Season** (for example "FW26") and a **Brief** (mood, customer, references) that is injected into every generation for its member designs. Create one with **New Collection**, or add an existing design with **Add to Collection**. --- ## Good to Know Designs created in an earlier beta may be flagged as **Legacy design** and can't be continued here. To pick one back up, contact the Brandmachine team to migrate it. --- ## FAQs **Do I need a pattern to design?** No. Patterns are optional. A design works fine without one. **Can I design without inspiration resources?** Yes, but the agent has more to work with when you give it images and notes. **Why can't I change the fashion model?** The model locks once it's set. Start a new design to use a different one. **Why is "Take measurements" greyed out?** Measurements need technical drawings first. Generate the flats, then take measurements. **What happens to old revisions when I delete them?** They're gone. Pin anything you want to keep before cleaning up. **Can I download a flat sketch?** Yes, from the slot itself once technical drawings have been generated. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Patterns Patterns is where you create reusable fabric prints and tileable textures. A pattern lives independently of any design and can be attached to multiple [Designer](/designer) instances. Unlike a design (whose revisions carry photo and technical-drawing tiles), a pattern revision is a single image: the print itself. --- ## Two Ways to Build a Pattern You can either let the agent generate one from inspiration, or upload a finished image and iterate from there. ### A. Agent-Generated 1. Open **Patterns** in the sidebar and click **New Pattern**. 2. In the right panel, click **+ Add** under **Inspiration References**. 3. Enter a **Label** (e.g. "dalmatian dog", "moth wing") and upload an image. Click **Add reference**. 4. Add more references as needed. 5. (Optional) Configure **Repeatability** in the right panel. There are two toggle switches: - **Repeats horizontally**: tileable left-to-right when placed on fabric. - **Repeats vertically**: tileable top-to-bottom when placed on fabric. Both, neither, or one are all valid. 6. Click **Start pattern**. The agent generates the first revision. Once revisions exist, the same button reads **Continue iterating**. 7. Use the chat to iterate ("more contrast", "softer edges", "tighter spacing"). Each turn produces a new revision. ### B. Upload a Finished Pattern 1. Open **Patterns** and click **New Pattern**. 2. Scroll to **Upload a finished pattern** and click **Upload pattern image**. 3. Provide an optional label and the image. Click **Upload**. 4. The uploaded image becomes the first revision. You can still iterate from there via chat. --- ## Repeatability Set repeatability any time, even after revisions exist, using the two toggle switches. The setting is stored on the pattern itself (not per revision), and it informs how the print behaves when applied to a garment downstream. Wherever an existing pattern's repeatability is shown, the label reflects the toggles: - **Tileable both axes**: both toggles on. Works as a seamless all-over print. - **Tileable horizontally**: only the horizontal toggle is on. Works as a horizontal stripe or border. - **Tileable vertically**: only the vertical toggle is on. Works as a vertical stripe. - When neither toggle is on, no repeatability label is shown. The print is treated as a placed motif. --- ## Revisions Each turn produces a new revision. The revision strip shows them horizontally, newest first, with a pin badge on favorites. For each revision: - The pattern image fills the main card. An optional **Inputs** strip below it lists the inspiration references used. - **Pin revision** toggles favorite state. - **Delete revision** removes it. --- ## Attaching to a Design Once a pattern has at least one revision (with an image), you can attach it to a design: 1. Open the design in [Designer](/designer). 2. In the right panel's **Patterns** card, click **+ Add**. A two-step picker opens. 3. First, pick a pattern from the list of eligible patterns. 4. Then pick a revision of that pattern. If the pattern has any pinned revisions, only those are shown; otherwise all image-bearing revisions are listed. 5. Click **Select pattern**. The attached pattern revision now appears on the design with its name and repeatability. The agent uses it when generating the next design revision. Remove with the trash icon. --- ## FAQs **Can I attach the same pattern to multiple designs?** Yes. Patterns are designed for reuse. **Can I edit a pattern's name?** Click the title at the top of the pattern page; inline edit. **What's the difference between an inspiration reference and an input?** Inspiration references are the source images you upload to the pattern. The "Inputs" strip on a revision shows which of those references were used for that specific generation. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Fashion Models Fashion Models is where you create and manage the AI models that appear in your [Campaign Composer](/campaign-composer-getting-started) photos, [Product Studio](/product-studio-getting-started) shots, [Group Shots](/group-shots), and [Designer](/designer) revisions. You can have Brandmachine cast a brand new model, or upload a real model you already have the rights to. --- ## How It Works Every model goes through the same pipeline: 1. **Casting.** Describe the look you want. Brandmachine generates a batch of candidate images to choose from. 2. **Name a candidate.** Naming a candidate turns it into a **draft** model and pulls structured **Resources** out of the image automatically. 3. **Resources.** The model's editable attributes (Description, Makeup, Hairstyle, Beard, Body Shape). Adjust them any time. 4. **Looks.** Generate styled snapshots of the model. Each **Look** is a saved revision you can pin, rename, and reuse. 5. **Library.** Add the model to your library so it becomes selectable everywhere else in the app. You can also skip casting and bring your own images, either a real model (**Upload Model**) or a style reference that seeds a fresh casting (**Upload Reference**). --- ## The Fashion Models Page Open **Fashion Models** in the sidebar. The list has two tabs: - **In Use**: models you've added to your library. These are the ones available across the app. - **Draft**: models you've started but not yet added to the library. Drafts are private to your team. Across the top you'll find four actions: - **New Model Casting** (primary): start a casting from a text prompt. - **Castings**: browse your past castings and their candidate images. - **Upload Reference**: upload a style photo to seed a new casting. - **Upload Model**: upload a real model you have rights to. --- ## 1. Cast a New Model 1. Click **New Model Casting**. A dialog opens. 2. Pick a **Gender**: Female, Male, or Other. 3. Write a **Prompt** describing the look you're casting for. More detail usually gives better results. Example: "25-year-old female, shoulder-length wavy brown hair, warm complexion, natural makeup, friendly approachable expression." 4. Leave **Use brand identity** on (the default) to fold your brand profile's style cues into the results, or turn it off to cast without them. See [Brand Identity](/brand-identity-getting-started). 5. Click **Start Casting.** Brandmachine generates a batch of candidate images on the casting page. Each candidate is tagged with the AI model that produced it, so you can compare different renderings of the same brief side by side. Castings draw from your credit balance (see [Billing & Credits](/billing-credits)). --- ## 2. Turn a Candidate Into a Model 1. On the casting page, click the candidate you like. 2. Choose **Name your model**, enter a model name, and confirm with **Create model.** Naming the candidate is what creates the model. There's no separate "add to drafts" step. The new model lands in your **Draft** tab, and Brandmachine automatically extracts its **Resources** from the image (Description and Hairstyle always, plus Makeup and Beard when they're relevant). --- ## 3. Resources Resources are the model's editable attributes. They're available on both drafts and library models, and you can change them at any time: - **Description** - **Makeup** - **Hairstyle** - **Beard** - **Body Shape** Edit each one as free text. You can also drop an image into the analysis zone at the top of the Resources panel and let AI read it. The analysis fills in **Makeup**, **Hairstyle**, **Beard**, and **Body Shape** for you, which you can then fine-tune. --- ## 4. Looks A **Look** is an immutable snapshot of the model: the Resources at that moment, the styling instruction you gave, and the resulting portrait or set-card image. - Click **Generate New Look** to create one. Iterate as many times as you want, before or after you add the model to your library. - **Pin Look** flags a Look as a keeper (it shows as **Pinned**). Pinned Looks are the ones offered first when you pick this model elsewhere. - Rename a Look by clicking its title. - Remove one with **Delete Look.** Generating a Look draws from your credit balance. --- ## 5. Add to Library While a model is a draft, it only lives in Fashion Models. Click **Add to Library** to move it into your library, where it becomes selectable in Campaign Composer, Product Studio, Group Shots, and Designer. Adding to the library is free. --- ## 6. Upload a Real Model If you work with a real model and have the rights to their likeness, skip casting and upload directly. From the Fashion Models page, click **Upload Model**, then provide: - **Model Name** - **Gender** (Female, Male, or Other) - **Portrait Image**, which must be exactly **2048 x 2048 px** - **Set Card Image** (optional), also **2048 x 2048 px** - The consent checkbox: "I confirm I have the legal rights to use this image for AI fashion model creation." The upload won't proceed without it. This creates the model directly as a **draft**, ready for Resources, Looks, and the library. Uploading is free. --- ## 7. Upload a Reference **Upload Reference** is different from Upload Model. A reference is style inspiration, not a real person. Use it when you have a photo whose look you want to reproduce, but you don't have the rights to the individual. 1. Click **Upload Reference.** 2. Pick a **Gender** and add one photo (any size up to 15 MB, JPG, PNG, or WEBP). 3. Submit. Brandmachine reads the style from your photo and starts a **new casting** inspired by it, which you then work through like any other casting. The reference photo itself isn't saved as a model or shown in your library. --- ## 8. Using a Model Elsewhere Only library models are selectable across the app. - In **Campaign Composer**, **Product Studio**, and **Designer**, the model picker is two steps: first choose the model, then choose a specific **Look**. If you've pinned Looks for that model, only the pinned Looks are offered; otherwise you see all of them. - **Group Shots** don't have their own picker. A group shot is composited from existing single shots, so each model (and its Look) is chosen upstream in Campaign Composer. Reusing the same model and Look across projects keeps your brand visually consistent. --- ## 9. Managing Models You can permanently **Delete** a model from its detail page. Deleting also removes the model's Looks. Images you already generated keep their pixels, but they lose their reference back to the model and Look. There's no deactivate step. A model is either a draft, in your library, or deleted. --- ## Billing Fashion Models runs on pay-as-you-go credits: - **Costs credits:** generating castings, Looks, and set-card images. - **Free:** uploading a model or reference, and adding a model to your library. See [Billing & Credits](/billing-credits) for balances, top-ups, and per-operation pricing. --- ## FAQs **Can I use a real person instead of casting one?** Yes. Use **Upload Model** and confirm you have the rights to their likeness. That's different from **Upload Reference**, which uses a photo only as style inspiration to seed a new casting. **Can I change a model's attributes after casting?** Yes. A model's attributes live in its **Resources** and stay editable at any time, on drafts and library models alike. Edit them as text, or drop in an image and let AI fill them. **How do I make a model available in Campaign Composer and Product Studio?** Add it to your library. Only library models show up in the pickers. **What happens to existing images if I delete a model?** The images you already generated stay intact. They just lose their reference to the deleted model and its Looks. **How do I control which Look gets used when I pick a model?** Pin the Looks you prefer. When a model has pinned Looks, only those are offered in the picker. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Getting Started Welcome to Brandmachine. This guide will help you get up and running. ## Quick Setup 1. **Get access**. Install the app from the Shopify App Store. If you're using Brandmachine outside of Shopify, request access at the `/signup` page. You'll see a "Check your inbox" confirmation, then an email invitation to set your password and finish creating your account. 2. **Create a Brand Identity**. Give Brandmachine your brand name and it will draft a brand profile you can refine. You'll find this later under **Shared Resources → Brand Memory** in the sidebar. 3. **Top up credits**. Brandmachine runs on pay as you go credits. Add a starting balance from **Settings → Billing**. That's the bare minimum to start generating. ## The Home Page Once you're signed in, the Home page shows: - **Recent campaigns** you've been working on - A **"Needs your attention"** board flagging items that need a decision or action - **Recent designs** if you've used Brandmachine Designer - A collapsible **"Learn Brandmachine"** band with onboarding videos at the top The credit balance pill in the bottom of the sidebar shows your current balance and links to Billing. ## The Sidebar The sidebar is organized as a set of doors, top-level areas that open into related tools. **Home**: your dashboard, described above. **Product Studio** - **Product Studio**: clean catalog photography with reusable perspectives, hero focus, and exact brand background color matching. - **Workbench**: the Kanban view (To Do / In Progress / Done) of your product studios, plus the finalize-and-download flow. - **Studio Setup**: default settings for your Product Studio shoots. **Campaigns** - **Campaigns**: campaign-quality photography combining models, locations, products, and styling. Includes Group Shots for multi-person scenes. - **Set Locations**: reusable background environments built from location, lighting, and mood references. - **Campaign Setup**: default settings for your campaigns. **Designer** - **Designer**: design garments with inspiration resources, generate Front / Back / Front flat / Back flat sketches, and take measurements off the flat patterns. - **Collections**: organize your designs. - **Patterns**: reusable fabric prints and tileable textures. Attach pattern revisions to a design when you're ready. **Automations** - **Shot Planner**: bulk-plan campaign images and product studios from a batch of input images. An agent drafts the plan, you review and execute. - **Workflows**: a newer area for building multi-step automations. See the Workflows section for details. **Shared Resources** - **Fashion Models**: create AI models, or upload existing models with rights consent. Active models show up in Campaigns, Product Studio, Designer, and Group Shots. - **Castings**: generate fashion model options based on your Brand Memory. - **Product Videos**: short promotional videos from a single product image. - **Brand Memory**: your brand profile, used for prompt suggestions and tone across every workflow. See the Brand Memory guide for details. **Settings**: credit balance, top-ups, transaction history, and API tokens live here, alongside the setup pages for Product Studio, Campaigns, and Brand Memory. While you're working inside a Product Studio, Campaign, or Group Shot image, you can also open the built-in Photo Editor for color correction, spot fixes, or lasso-driven generative edits. ## What Got Renamed If you've been using Brandmachine for a while: - **Review** is now **Workbench**. - **Batch Editor** is now **Shot Planner**. - **Campaign Composer** is now **Campaigns**. - **Brand Identity** is now **Brand Memory** in the sidebar (you may still see "Brand Identity" used on the Settings page itself). - **Store Content** is no longer part of the app. - **Workflows** is a new feature, not a rename, for building multi-step automations. - The old subscription plans are now **PAYG credits**, billed through Stripe or Shopify. ## Need Help? Email **[support@brandmachine.shop](mailto:support@brandmachine.shop)** any time. ================================================================================ # Photo Editor The Photo Editor is Brandmachine's in-browser tool for fine-tuning generated images. It opens from any image card's **"..." (More actions)** menu. Three modes cover most retouching needs: color correction, spot correction, and lasso-driven generative edits. > **Note:** "Edit with AI" is a gated feature and may not be available in every workspace yet. If you don't see it in the "..." menu, contact [support@brandmachine.shop](mailto:support@brandmachine.shop) to ask about access. --- ## Opening the Editor From any image's card in **All Images**, **Favorites**, or **Rejected**, click the **"..." (More actions)** icon in the corner of the card, then choose **Edit with AI**. The editor opens in a full-screen modal titled "AI Image Editor". The three modes are selectable as toggle buttons at the top of the sidebar: 1. **Color Correction** 2. **Spot Correction** 3. **Image Edit** --- ## Color Correction Useful for region-specific color and exposure tweaks. 1. Type what you want to segment (e.g. "background", "jacket", "skin"). 2. Click **Segment.** Brandmachine generates a mask for that region. 3. Adjust: - **Exposure**. ±2.0 EV - **Hue Shift**. ±180° - **Saturation**. 0 to 2.0x 4. Fine-tune the mask: - **Grow / Shrink**. ±20 px to expand or contract the selection - **Blur**. 0 to 25 px to feather the edge 5. Click **Save & Start New Adjustment** to lock in this layer and start another region. Layers are stackable, each toggleable on/off, each deletable. --- ## Spot Correction For fixing blemishes, damage, or inconsistencies using a clean reference crop. 1. Upload a **reference image**: a clean crop showing what the area should look like. You'll need to crop this yourself before uploading. 2. Use the **brush tool** to paint over the area you want to fix. 3. Click **Apply Fix.** Brandmachine returns one or more variants. 4. The editor enters **blend refinement** (see below). 5. Browse variants with arrow controls and **Apply** the one you want, or **Discard** to start over. --- ## Image Edit (Lasso) For generative edits: replace, modify, or augment regions using a prompt. 1. Use the **lasso tool** to outline the area to change (freehand outline, not a brush). 2. Write a prompt describing the edit, e.g. "Replace jacket with leather, keep pose and lighting". 3. (Optional) Click the **+** icon to attach reference images that guide the AI. 4. Click **Generate.** Brandmachine returns one or more variants at the image's native resolution. 5. The editor enters **blend refinement**. 6. Browse variants and **Apply**, or **Discard**. --- ## Blend Refinement After Spot Correction or Image Edit returns variants, the editor opens the blend controls so you can seat the AI patch cleanly into the original image: - **Grow / Shrink**. ±20 px. Spreads or contracts the patch into the surrounding image. - **Feather**. 10 to 150 px. Softens the transition edge. - **Edge Sharpness**. 1 to 20. How abruptly the falloff happens. - **Rotation**. ±200°. Rotate the patch if the content sits at an angle. - **Offset X / Y**. ±256 px. Nudge the patch position. - **Hue**. ±180°. Color-correct just the patch. - **Brightness**. ±2.0 EV. Match exposure to the surrounding image. Internally, the editor erodes and feathers the boundary inward before compositing, which prevents the dark halos you'd otherwise see where an AI patch meets original pixels. --- ## Viewing and Comparing Outside the editor, the card-level controls give you a viewer and, where applicable, a comparison view: - **Viewer.** Click a card's thumbnail to open it in a full-screen viewer. There's no dedicated zoom control, just a toggle between **full-width** (scrollable) and **fit-to-viewport** display. (Pinch-to-zoom works on the canvas inside the AI Image Editor itself, but the card-level viewer doesn't have a separate zoom icon.) - **Compare modal.** Columns icon, only shown when the image has input images attached (for example product references). Split-screen view: inputs on the left, generated image on the right. Useful for color or pose checks. If an image has no inputs, this option isn't available. --- ## FAQs **Will edits cost credits?** Yes. See [Billing & Credits](/billing-credits) for the price list. Color correction layers are cheaper than generative edits. **Can I edit a Group Shot or Product Studio output?** Yes. The Photo Editor works on any generated image's card. **Where do edits get saved?** Applied edits replace the active image. The original is kept under the hood; if you want it back, contact support. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Product Videos Product Videos turns a product image into a short animated video you can add straight to your Shopify catalog. Pick a product or upload an image, choose an animation prompt (or write your own), generate, preview, then push the finished clip to your product as new Shopify media. --- ## 1. What You Need - **Input**: any single product image works, garments, accessories, furniture, electronics, and everything in between. There's no restriction on product type. - A note on licensing: make sure you're allowed to use the source image for further processing. --- ## 2. Workflow Steps 1. ### Choose Your Product Image - Click **Catalog** to pick an existing product using Shopify's product picker, or click **Upload** to add a new image instead. 2. ### Pick or Write a Prompt - Brandmachine suggests exactly 4 animation prompts tailored to your product. Click one to use it as is. - Want something different? Edit the prompt directly in the text field (placeholder: "Describe the video animation you want to create..."). There's no character limit, write as much detail as you need. 3. ### Generate Video - Click **Generate Video**. Generation takes a few minutes. If you regenerate later, this button reads **Generate New Version** instead. 4. ### Preview - Once rendering finishes, the video appears in the preview area so you can watch it before deciding to use it. 5. ### Upload to Product Catalog - Click **Upload to Product Catalog** to attach the video to the product in Shopify. The video is added as a new media item, your original product image is kept, not replaced. - A success banner confirms the upload. - If the upload fails (e.g. API permissions revoked), reach out to [support@brandmachine.shop](mailto:support@brandmachine.shop). --- ## 3. FAQs 1. **Can I upload any product image, or does it need a specific background?** - Any clear product image works, studio shots, lifestyle images, or flat lays. 2. **Which AI model is used for generation?** - We use best-in-class external video models and keep them updated, so you don't have to think about it. 3. **What if I don't like the suggested prompts?** - Edit the prompt directly in the text field before generating, or generate a version and try a different prompt afterward. 4. **How long does it take to generate a video?** - Typically a few minutes per video. 5. **Can I regenerate until I like the result?** - Yes, generating and regenerating both use credits, so this depends on your plan's quota. See [Billing & Credits](/billing-credits). 6. **Will the video replace my original product image?** - No. The video is added as a new media item alongside your existing images on the Shopify product page. Your original image remains intact. 7. **Is the generated video optimized, so it won't negatively affect my store's performance?** - Yes. Shopify handles the upload and optimizes the video, which is also why it can take a few minutes for the video to be ready within Shopify after the upload itself completes. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Product Studio Product Studio generates professional e-commerce product photography from your product images. Unlike Campaign Composer (which creates lifestyle campaign imagery), Product Studio focuses on clean, catalog-ready shots driven by reusable **perspectives**. --- ## How It Works A **studio** is one product (or group of products) shot from multiple **perspectives**. Each perspective is a saved combination of AI model, styling instructions, aspect ratio, background (a color or a scene), accepted input view types, and toggles that decide whether the shot includes a model or just the product. Once your perspectives are configured, generating new studios for new products is fast: upload the product images, pick a model (if needed), generate a preview, generate the full shots, pick favorites, finalize. --- ## 1. Set Up Perspectives (One-Time) Go to **Settings → Product Studio** and open the **Perspectives** section. You'll see two tabs: **Active** and **Archived**. Click **Add Perspective** (or the three-dot menu on an existing one to **Edit**). In the modal, configure, in order: - **Name**: e.g. "Front Studio", "Pack Shot Detail", "Color Swatch". - **AI Model**: choose **Gemini 3 Pro**, **GPT Image 2**, or **Nano Banana 2 (fast)**. Nano Banana is fastest, GPT Image 2 is best for apparel and print detail, Gemini 3 Pro is the most stylistic. - **Instructions**: per-perspective text guidance (lighting, composition, mood, accessories). - **Accepted View Types**: which input view types feed this perspective (Front, Back, Other, Cropped). - **Aspect Ratio**: a standard preset (3:4 default, 1:1, 4:5, 2:3, 9:16, 5:4, 4:3, 3:2, 16:9, 21:9, ten in total) or a **custom ratio** like 7:9. See [Aspect Ratios & Custom Crops](/product-studio-aspect-ratios). - **Brandmachine Color Control**: an optional 6-character hex (e.g. `ff0077`). Brandmachine grounds the model on a real color swatch, measures the output, then applies an algorithmic correction so the final image matches your exact brand color. - **Scene (Shooting Location)**: optionally render the product into a saved shooting-location scene instead of a flat backdrop color. A scene takes priority over the backdrop color, so selecting one disables the hex color field above, a perspective uses either a scene or a color, not both. - **Product Only**: when checked, the shot generates the product without a model or scene. Use for pack shots. - **Use Hero Inputs Only**: when checked, this perspective only generates from product inputs you've marked as **hero**. Useful for fabric swatches and focus shots. You can **archive** perspectives you don't want cluttering the picker without deleting them. The default perspective can't be archived. Restore from the **Archived** tab any time. --- ## 2. Create a Studio 1. **New Studio** from the Product Studio page. 2. **Upload product images.** Drag-and-drop for a single product, or use [Shot Planner](/shot-planner) to bulk-create studios across many products at once. Supported: 3D renders, pack shots, isolated product photos, existing catalog images. 3. **Assign view types.** Click each uploaded image to set its view type (front, back, other, cropped). The view type determines which perspectives will use it. 4. **Mark hero products.** On the edit panel, each product image has a **MAKE HERO** button. Hero products are the main focus when a perspective has "Use Hero Inputs Only" enabled. A star icon appears on hero images. 5. **Select a fashion model.** Required for perspectives that aren't Product Only. 6. **Edit the studio title** by clicking it (the pencil icon appears on hover). 7. **Generate Preview (optional).** Click **Generate Preview** for a free, low-res preflight of the current perspective, a quick way to sanity-check styling before spending a full generation. 8. **Generate.** Click **Generate Studio Images** (or **Generate More**, once that perspective already has outputs) to create full-resolution outputs for the perspective. See also: [Shot Planner](/shot-planner) for bulk creation across a whole catalog at once. --- ## 3. Generation Eligibility The Generate buttons stay disabled until a perspective's requirements are met: - **A matching input photo.** At least one uploaded image must have a view type the perspective accepts. - **A fashion model**, unless the perspective is set to **Product Only**. - **A hero-marked input**, if the perspective has **Use Hero Inputs Only** enabled. Hover the disabled button to see exactly what's missing. --- ## 4. Background Color Correction When a perspective has a hex color set (and no scene attached), Brandmachine doesn't just prompt the AI to "use this background". It: 1. Generates the shot. 2. Removes the background. 3. Measures the actual rendered backdrop. 4. Applies an algorithmic correction so the final image hits the exact hex. You don't have to do anything to enable this beyond setting the hex in the perspective. When you zoom on a generated image in the comparison view, a small color swatch appears in the corner showing the target color. If the perspective has a **Scene** set instead, the product renders into that shooting location and this correction step is skipped. Scenes and backdrop color are mutually exclusive per perspective, setting one clears the other. --- ## 5. Review and Finalize Generated outputs land in the studio detail view. You can: - **Favorite** an output (heart icon) to mark it as the keeper for that perspective. - **Generate More** to produce additional variants for a specific perspective. - **Edit** an output in the [Photo Editor](/photo-editor). When you're ready to lock in selections and download, jump to the [Workbench](/workbench) and open **Quick review**. There you pick one favorite per perspective, then click **Finalize selection** to commit and download to export everything. --- ## 6. The "No Match" Warning In the [Shot Planner](/shot-planner) you may see a "No match" badge on an input image. That means the image's view type (e.g. `back`) doesn't match the accepted view types of any active perspective. Either: - Change the input's view type so it matches an existing perspective, or - Add a perspective that accepts that view type. --- ## 7. Best Practices - **Invest time in perspectives.** They're the leverage point. Get lighting, background, and styling right once; reuse forever. - **Pick the right AI Model per perspective.** Nano Banana 2 (fast) for quick iteration, GPT Image 2 for apparel and print detail, Gemini 3 Pro for the most stylistic results. - **Use Generate Preview before a full run**, especially while you're still dialing in a new perspective's instructions. - **Use Brandmachine Color Control** for any catalog where exact color matters. The algorithmic correction is what makes this work end to end. - **Mark heroes carefully.** A "Use Hero Inputs Only" perspective is great for fabric swatch shots; a generic perspective should usually leave hero off so all inputs feed it. - **Bulk creation.** For large catalogs, use [Shot Planner](/shot-planner) to plan and generate many studios at once. --- ## 8. FAQs **What's the difference between Product Studio and Campaign Composer?** Product Studio makes catalog and product-page shots. Campaign Composer makes lifestyle campaign images. Different purposes, different outputs. **Can I use Product Studio for non-fashion products?** Yes, especially with Product Only perspectives (pack shots). The system handles 3D renders well too. **Which AI Model should I pick?** Nano Banana 2 (fast) is fastest, GPT Image 2 is best for apparel and print detail, Gemini 3 Pro is the most stylistic. You can set a different model on every perspective. **Can I export directly to Shopify?** The Workbench's download flow gives you ready-to-upload assets. Direct push to product pages isn't built yet. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Aspect Ratios & Custom Crops Every Product Studio perspective is generated at a specific **aspect ratio**, the shape of the final image, like 3:4 or 1:1. You can pick a standard ratio, or set a **custom ratio**: any width-to-height you need, such as **7:9** for a marketplace that requires it, even though the AI models don't produce that shape on their own. Brandmachine bridges the gap and delivers the exact ratio you asked for. --- ## Standard vs. Custom Ratios When you set up a perspective (**Settings → Product Studio → Perspectives**), the **Aspect Ratio** control offers: - **Standard presets**: 3:4 (default), 1:1, 4:5, 2:3, 9:16, 5:4, 4:3, 3:2, 16:9, 21:9. These are shapes the AI models produce directly, so there's no cropping at all. - **Custom W:H**: enter any width and height (e.g. `7` and `9`). Brandmachine generates at the closest standard shape and then crops the result down to your exact ratio. A perspective uses **one or the other**: choosing a standard preset clears any custom ratio, and entering a custom ratio takes over. Whatever you choose, every place the image appears in Brandmachine (the results grid, zoom, review, and the high-res download) shows that exact ratio. **What you see is what you download.** --- ## What Happens Behind the Scenes The AI models can only generate a fixed set of image shapes. When you ask for a shape they don't make (like 7:9), Brandmachine: 1. **Generates** at the closest shape the model *does* make (for 7:9, that's 4:5). 2. **Crops** the result down to your exact 7:9. This happens **once, at generation time**. The stored image is already your ratio, so everything downstream (background color correction, the photo editor, releasing, and downloading) works on the final cropped image. Nothing later has to know a crop happened. --- ## Reading the Crop Hint When you enter a custom ratio, the modal shows a hint like: > *Generates at 4:5, trims ~2.8% to reach 7:9.* It tells you (a) which standard shape is actually generated, and (b) roughly how much of the image is trimmed to reach your target. The closer your custom ratio is to a standard one, the smaller the trim: 7:9 sits very close to 4:5, so only about 2.8% is cropped away. --- ## What You Can Rely On - **It never stretches or upscales.** Brandmachine only ever crops *inward* from a larger image; it never enlarges pixels to fill a shape. Your product is never distorted. - **The crop is centered**, trimming evenly from the edges of the longer dimension. - **The trim is kept minimal** by always generating at the closest available shape. - **It's consistent everywhere**: grid, zoom, review, and download all reflect the same ratio. --- ## Under the Hood (Technical) For the technically curious, here's how the crop stays exact and why the model you pick matters a little. **The models have different native shapes.** Product Studio runs on three image models, and they don't all support the same shapes: - **Nano Banana 2** and **Gemini 3 Pro** each produce a fixed set of 10 "native" ratios, exactly the standard presets listed above. - **GPT Image 2** is more flexible: it accepts explicit pixel dimensions, so it isn't limited to those 10 shapes. The standard presets are deliberately the **common set** every Product Studio model supports, so a custom target always maps to a shape the chosen model can render. **Choosing the shape to generate.** For a custom target, Brandmachine picks the native shape that loses the least to cropping, the one whose proportions are closest to your target. For 7:9 that's 4:5 (~2.8% trimmed), which beats the next-closest 3:4 (~3.6%). **Why the result is always exact, whatever the model does.** The crop is computed from the **actual pixels the model returns**, not from the shape we requested. So even if a model interprets the request loosely and returns a slightly different shape, the crop still trims to your exact target. The only thing that changes is *how much* is trimmed, never the final ratio. And because the crop box is always **inscribed inside** the returned image, it can only ever cut, never enlarge. **GPT Image 2 nuance.** Because GPT Image 2 accepts arbitrary sizes, it could in principle be asked to generate your exact ratio directly, with almost no crop. Today it goes through the same closest-shape-then-crop path as the other models for consistency, so it may trim a little more than strictly necessary. The final image is exactly your ratio either way. **Adding new models.** Because the standard presets are the shared, model-agnostic set, a future model that supports *more* shapes would still work, it just wouldn't yet take advantage of its extra shapes (it would crop from the nearest shared one). A model that supports *fewer* shapes would need its supported set checked before being added. In short: aspect-ratio handling is currently shared across models rather than tailored per model. **Reliable even when a model is sparse with metadata.** Some models don't report the generated image's dimensions in their response. Brandmachine measures the stored image itself in that case, so the crop always has accurate dimensions to work from. --- ## FAQs **Can I set any ratio I want?** Yes, within reason: enter any width and height as whole numbers. Extreme ratios work but trim more of the image, since they sit farther from the nearest standard shape. **Does a custom ratio reduce image quality?** There's no stretching or upscaling, so nothing is degraded that way. You do lose the pixels trimmed off the edges (the larger the crop, the fewer pixels remain), but the kept area stays at full quality. **Why can't the AI just generate 7:9 directly?** Most image models only output a fixed menu of shapes. 7:9 isn't on that menu, so Brandmachine generates the closest menu item and crops to your exact ratio. **Does the model I choose change the result?** The final ratio is always exactly what you set. The model can affect how much is trimmed (each model's native shapes differ slightly), but not the outcome you receive. **Does background color correction still work with a custom ratio?** Yes. Correction and the photo editor run on the already-cropped image and preserve its ratio. **What happens to the original, uncropped image?** The cropped image becomes the final and only image, and the uncropped original isn't kept. (If the crop step ever fails for a render, Brandmachine keeps that render's uncropped image rather than losing it.) **If I change a perspective's ratio, do my existing images change?** No. The ratio applies to **new** generations. Images you've already generated keep the ratio they were made with. **How do I switch a perspective back to a standard ratio?** Just pick a standard preset in the Aspect Ratio control, and the custom ratio clears automatically. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Breuninger Studio Set The **Breuninger Studio Set** is a ready-made group of three Product Studio perspectives that produce images in the shape and style the Breuninger marketplace expects: a model on a seamless off-white studio backdrop, standing small in the frame with floor visible below the shoes, delivered at 3:4. The set is **managed by Brandmachine**. You add it once, and the prompt, backdrop, framing, and format behind it are maintained for you. When we improve them, every team that has the set improves with them, with nothing for you to re-apply. --- ## What You Get Adding the set creates three perspectives in your list: | Perspective | Uses input images marked | Format | Editable | |---|---|---|---| | **Breuninger Front** | Front | 3:4 | AI model only | | **Breuninger Back** | Back | 3:4 | AI model only | | **Breuninger Detail** | Cropped | 3:4 | Fully yours | **Front** and **Back** are the marketplace-conform full-body shots. Both need a fashion model selected on the studio, since they show the garments worn. **Detail** is a starting point for close-up texture and stitching shots on the same backdrop. It is an ordinary perspective after it is created: rename it, rewrite its instructions, change its format, archive it. It behaves like any perspective you made yourself. --- ## Adding the Set 1. Go to **Settings → Product Studio** and open the **Perspectives** section. 2. At the top of the list you'll see a card for the **Breuninger Studio Set**. 3. Click **Add them**. The three perspectives appear in your list right away, each carrying a **Breuninger preset** badge so you can tell them apart from your own. The card disappears once all three exist. Clicking **Add them** more than once is safe: it only creates what's missing and never duplicates anything. If you archive or delete the **Detail** perspective later, it stays gone; that one is your decision to make. --- ## What "Managed by Brandmachine" Means On the **Front** and **Back** perspectives: - **The AI model is yours to choose.** Everything else is fixed. - **Instructions, backdrop, accepted view types, and aspect ratio are not editable.** They're what makes the output marketplace-conform, so the editor hides them rather than letting a change slip through. - **They can't be archived or deleted.** The set is meant to stay intact and ready. - **Brandmachine Color Control doesn't apply.** These shots use a fixed off-white studio backdrop, so there's no hex color to set and no color-correction step. - **Your team-wide Product Studio instructions don't apply to them either.** Global instructions can pull a shot away from the marketplace spec, so the managed perspectives ignore them. Your own perspectives still use them as usual. The **Detail** perspective has none of these restrictions. --- ## Shooting a Product for Breuninger 1. **Create a studio** and upload your product images as usual. 2. **Assign view types carefully.** This is what decides which perspective picks up which image. An image marked **front** feeds Breuninger Front; **back** feeds Breuninger Back; **cropped** feeds Breuninger Detail. An image with no matching view type is simply skipped, so a missing "back" tag is the usual reason a back shot never appears. 3. **Select a fashion model.** Required for Front and Back. 4. **Generate a preview** if you want to sanity-check the styling first. Previews go through the same framing step as full shots, so what you see is what you'll get. 5. **Generate.** The Front and Back outputs arrive already framed to the marketplace spec. --- ## The Framing Step Front and Back shots go through one extra automatic step after the AI render: 1. Brandmachine renders the model on the studio backdrop. 2. The image is then measured and cropped so the figure sits where Breuninger wants it: centered, with headroom above and a consistent band of floor below the shoes, at exactly 3:4. This runs on its own, on both previews and full-resolution shots. It's why a Breuninger generation takes slightly longer than a regular perspective and why the task briefly shows a framing stage. Detail shots skip it. **If a render doesn't leave enough floor to crop**, Brandmachine delivers the image uncropped rather than cutting into the model. You get a usable shot either way, but it won't match the marketplace framing, so it's worth regenerating that one. --- ## Tips - **Fewer garments per studio gives more reliable framing.** The more product images you feed into one shot, the larger the model tends to render in the frame, which leaves less floor for the crop. If you're seeing uncropped results, splitting a large outfit across studios usually fixes it. - **Tag your inputs before generating.** View types are the whole routing mechanism here. See [Product Studio](/product-studio-getting-started) for the full walkthrough. - **Use the Detail perspective as a base, not a fixed rule.** It ships with a generic close-up instruction; adjust it to whatever your category needs. --- ## FAQs **Do I need a separate Breuninger account or connection?** No. The set is just a group of perspectives inside Product Studio. You download the finished images from the [Workbench](/workbench) and upload them wherever you need them. **Why can't I edit the Front and Back instructions?** Because the exact wording is what keeps the output conform: figure size, backdrop, and shadow behavior all come from it, and each was tested. If the instructions were editable, an ordinary-looking tweak could quietly break the framing. Use the Detail perspective, or one of your own, when you want full control. **Can I use these perspectives for images that aren't going to Breuninger?** Yes. Nothing stops you. They produce a clean, consistent full-body studio look that works well in most catalogs. **What if I click "Add them" twice?** Nothing breaks. The set is created once; a second click only fills in anything missing. **I archived the Detail perspective. How do I get it back?** Restore it from the **Archived** tab in the Perspectives section. Adding the set again won't bring it back, since archiving it was a deliberate choice. **Can I change the aspect ratio to something else?** Not on Front and Back. 3:4 is part of the marketplace spec. The Detail perspective can use any ratio, including a custom one; see [Aspect Ratios & Custom Crops](/product-studio-aspect-ratios). **Can I edit a Breuninger image afterwards?** Yes, in the [Photo Editor](/photo-editor), same as any other output. Your edit becomes the current version of that image and is what gets downloaded. **Does the set change over time?** Yes, and that's the point of it being managed. When we improve the prompt or the framing, your existing perspectives use the improvement on the next generation. Images you've already generated stay exactly as they are. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # ZIP Upload > **Use [Shot Planner](/shot-planner) instead.** Shot Planner is now the recommended way to bulk-import products into Product Studio: upload your images, let an AI agent draft a plan across your perspectives, review it, and execute. For most catalogs it replaces the ZIP workflow described below. ZIP upload still works under the hood, but there's currently no **Upload ZIP** button in the app, the entry point is switched off in production. Treat it as a legacy, secondary method rather than something you can pick up self-serve today. The mechanics below remain accurate if your account still has this method available. --- ## Preparing the ZIP Archive ### Folder Structure Your ZIP must follow this hierarchy: ``` {product-group}/ fashion-model.txt {product}/ {perspective}/ image.[jpg|jpeg|png|webp] ``` ### What Each Level Means **Product Group** - The top-level folder. Give it any name you like (e.g. `summer-collection`). Each product group becomes its own Product Studio session. You can have multiple product groups in a single ZIP to process different collections at once. **`fashion-model.txt`** - A plain text file placed directly inside each product group folder. It must contain the UUID of the fashion model you want to use for that group. See [Finding the Fashion Model UUID](#finding-the-fashion-model-uuid) below. **Product** - A subfolder within the product group, one per product. The folder name is used as the product title (hyphens are replaced with spaces and the name is capitalized, so `red-sneakers` becomes "Red Sneakers"). **Perspective** - A subfolder within each product. The name must be one of: - `front` - `back` - `other` - `cropped` The name is case-insensitive, so `Front` and `front` both work. **Image Files** - Place your product images inside the perspective folders. Supported formats are `.jpg`, `.jpeg`, `.png`, and `.webp`. --- ## Limits - **Maximum ZIP size:** 20 MiB - **Maximum 6 images per product** (across all perspectives combined) --- ## Example ``` my-import.zip ├── summer-collection/ │ ├── fashion-model.txt │ ├── red-sneakers/ │ │ ├── front/ │ │ │ └── sneaker-front.jpg │ │ └── back/ │ │ └── sneaker-back.jpg │ └── blue-jacket/ │ ├── front/ │ │ └── jacket-01.png │ └── other/ │ └── jacket-detail.webp └── winter-collection/ ├── fashion-model.txt └── wool-scarf/ └── front/ └── scarf.jpg ``` In this example, `summer-collection` and `winter-collection` are two separate product groups. Each can use a different fashion model. The summer collection contains two products (red sneakers and blue jacket), while the winter collection has one (wool scarf). --- ## Finding the Fashion Model UUID Each product group needs a `fashion-model.txt` file containing the UUID of the model to use. You can find this UUID in the URL when viewing a fashion model in the Brandmachine app: ``` https://admin.shopify.com/store/{your-store}/apps/brandmachine/fashion-models/{MODEL-UUID}/revision/{REVISION-UUID} ``` For example, in this URL: ``` https://admin.shopify.com/store/my-store/apps/brandmachine/fashion-models/88F9F63E-0773-4F40-944E-2B3853BC6A83/revision/45909C76-D54C-4F9F-A58B-B0B30A1B4E7C ``` The fashion model UUID is `88F9F63E-0773-4F40-944E-2B3853BC6A83`. Copy this value and paste it into your `fashion-model.txt` file. --- ## What Happens After Upload Once you upload the ZIP, the system processes it in the background: 1. **Image analysis** - Each product image is analyzed by AI to generate a description 2. **Generation** - Product Studio outputs are generated for every configured perspective that matches your uploaded view types 3. **Results** - Generated images appear in the Product Studio interface, grouped by the product groups from your ZIP All generations use 4K resolution, the same as studios created through the normal Product Studio flow or through [Shot Planner](/shot-planner). --- ## Troubleshooting **"File must be a ZIP archive"** - Make sure your file has a `.zip` extension. **"ZIP file must not exceed 20 MiB"** - Reduce image file sizes (e.g. use WebP format) or split into multiple uploads. **"Invalid perspective"** - Check that your perspective folder names are exactly `front`, `back`, `other`, or `cropped`. No other names are accepted. **"Too many images per product"** - A product can have at most 6 images total across all its perspective folders. **"Archive contains no valid images"** - The ZIP was parsed but no supported image files were found. Check that your images are `.jpg`, `.jpeg`, `.png`, or `.webp` and that the folder structure is correct. **"Fashion model not found"** - The UUID in `fashion-model.txt` doesn't match any fashion model. Double-check the UUID from the URL and make sure the model exists and is fully trained. --- For bulk imports today, start with [Shot Planner](/shot-planner). If you rely on an existing ZIP-based import and need help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Set Locations Set Locations lets you create and refine reusable background environments for your campaign photography. Think of it as building a virtual photo shoot location that you perfect once, then reuse across your campaigns. --- ## Overview Set Locations gives you fine-grained control over the scenery in your campaign images. Instead of describing a background every time you create a campaign photo, you create, refine, and save reusable locations that keep your look consistent. **Key Concept:** The images you generate in Set Locations are **placeholders** that capture the essence and description of a scenery, not the exact backgrounds used in your campaigns. When you use a set location in [Campaign Composer](/campaign-composer-getting-started), the system recreates the environment with the right perspective and composition for your product and model. --- ## 1. Understanding References Set Locations uses four types of **references** to define a scenery: 1. **Location** (required): The physical environment (e.g. "rugged mountain landscape", "urban rooftop"). 2. **Lighting** (optional): Light quality and direction (e.g. "golden hour sunset", "soft diffused studio lighting"). 3. **Mood** (optional): Emotional atmosphere (e.g. "dramatic and moody", "bright and energetic"). 4. **Extras** (optional): Background people and crowd characteristics (e.g. "a few passersby in the distance", "busy street crowd"). Only **Location** is required. Lighting, Mood, and Extras are optional and refine the result further when you add them. Each reference can be created in two ways: - **Text description**: type what you want. - **Image analysis**: drop in an image and let Brandmachine analyze it. --- ## 2. Workflow Steps ### Creating a New Set Location 1. **Navigate to Set Locations** - In the sidebar, click **Set Locations**. - Click **New Set Location** to start. 2. **Define References** Configure your references. Location is required; Lighting, Mood, and Extras are optional. **Text-Based Reference:** - Click a reference card (e.g. Location). - A modal opens with a description field. - Describe what you want (e.g. "Rocky mountain environment with pine trees and misty peaks"). - Click **Save**. **Image-Based Reference (AI Analysis):** - Click a reference card. - Drag and drop an image into the drop zone. - Brandmachine analyzes the image based on the reference type: - **Location**: environmental elements, setting, architecture. - **Lighting**: light quality, direction, color temperature. - **Mood**: emotional tone, atmosphere, energy. - **Extras**: background people and crowd characteristics. - A text description is extracted automatically. 3. **Generate a Scene** - Once at least your Location reference is set, click **Generate New Scene**. - Brandmachine creates a scene based on your references. This may take a few moments. 4. **Refine by Editing References** Refinement happens by adjusting your references and generating again: - Edit any reference card to change its description, swap its image, or add an optional reference you skipped earlier (Lighting, Mood, or Extras). - Click **Generate New Scene** to produce a new scene from the updated references. - Repeat until the scenery feels right. Every generation uses your current references, so each change you make is reflected in the next scene. 5. **Manage Scenes** - Each generation adds a new scene to the **Scenes** panel. - You can revisit any previous scene at any time, which is handy if a change goes in the wrong direction. - For each scene you can: - **Pin** it to mark it as a favorite. - **Rename** it (click the title to edit inline). - **Delete** it when you no longer need it. ### How Refinement Works - References influence **every** generation, not just the first. Whatever your Location, Lighting, Mood, and Extras cards say when you click **Generate New Scene** is what shapes that scene. - To nudge a scene in a new direction, edit the relevant reference and generate again. - To start from a completely different concept, clear or rewrite your references, or create a new set location from scratch. --- ## 3. Using Set Locations in Campaigns Once you have a set location with a scene you like, you can use it when building a campaign image: 1. Go to your [Campaigns](/campaign-composer-getting-started) area and start (or open) a campaign photo. 2. On the **Shooting Location** card, click **Select** (or **Edit** if one is already chosen). 3. A grid picker opens. First pick the set location you want, then pick a specific **scene** within it. 4. Your choice is applied to the campaign image. When used in a campaign: - The system uses the **description** of the scenery, not the placeholder image itself. - The environment is recreated with the right perspective and composition for your product and model. - This keeps the product as the focal point while preserving the scenic atmosphere. --- ## 4. Best Practices 1. **Start with Location, Layer the Rest** - Begin with a solid Location reference. - Add Lighting, Mood, and Extras as you discover what the scene needs. - You don't have to describe everything up front. 2. **Use Image References for Precision** - If you have an inspirational photo, use it as a reference. - Brandmachine extracts relevant attributes based on the reference type. - Combine image references with text for the best control. 3. **Refine One Thing at a Time** - Change a single reference, then generate a new scene to see its effect. - Small, deliberate edits make it easier to see what each reference contributes. 4. **Lean on the Scenes Panel** - Don't be afraid to experiment. You can always return to an earlier scene. - Pin the scenes you like so they're easy to find later. 5. **Build a Library** - Create set locations for different campaign types. - Keep a consistent look across seasonal campaigns. - Reuse successful locations across multiple products. --- ## 5. Technical Details ### Why Placeholder Images? You might wonder why Set Locations generates placeholder images instead of the exact backgrounds used in campaigns. **The reason:** - Inspirational environment photography focuses on composition, leading lines, and artistic framing. - Campaign photography focuses on the product, and needs different angles and composition. - Using the environment photo directly would create composition conflicts. - Instead, Brandmachine captures the **description** of the scenery and recreates it appropriately in your campaign image. ### Reference Usage - Your references (Location, Lighting, Mood, Extras) are active on **every** generation, not only the first. - Editing a reference and clicking **Generate New Scene** produces a fresh scene shaped by your current references. - This makes refinement a matter of tuning the references, rather than starting over. --- ## 6. FAQs 1. **Can I edit references after generating a scene?** Yes. Editing references and clicking **Generate New Scene** is exactly how you refine. References apply to every generation, so your latest edits always shape the next scene. 2. **How many scenes can I create?** There's no hard limit on scenes, though your plan may include generation quotas. 3. **Can I use the same set location in multiple campaigns?** Yes. Set locations are reusable across any number of [Campaign Composer](/campaign-composer-getting-started) projects. 4. **Which references do I actually need?** Only **Location** is required. Lighting, Mood, and Extras are optional and give you extra control when you add them. 5. **What is the Extras reference for?** Extras describes background people and crowd characteristics, so you can set how busy or empty the scene feels. 6. **Can I delete scenes or set locations?** Yes. You can pin, rename, and delete individual scenes, and you can remove whole set locations you no longer need. 7. **Do I need to create set locations before using Campaigns?** No. You can create set locations in advance, or on the fly while building a campaign image. 8. **What makes a good reference image?** - Clear representation of the environment (for Location). - Clear lighting characteristics (for Lighting). - Clear emotional atmosphere (for Mood). - High-quality images work best, though the analysis is forgiving. --- ## Conclusion Set Locations gives you professional-level control over campaign backgrounds by separating environment creation from campaign production. With this workflow you can: 1. Define scenery using text or image references. 2. Generate a scene from your references. 3. Refine by editing references and generating again. 4. Build a reusable library of brand-consistent locations. 5. Use those locations confidently in [Campaign Composer](/campaign-composer-getting-started). For further assistance, contact Support at [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Shot Planner Shot Planner is the bulk-planning view for campaign images and product studios, found under **Automations** at `/batch`. You start a shoot, add product photos, and let a planning agent draft what to create across all of your products at once. Once the plan looks right, you generate previews to check the results, then generate finals when you're ready to spend credits. If you've used Brandmachine before, Shot Planner is the renamed and restructured version of the old **Batch Editor**. --- ## Creating a Shoot From the Shot Planner list, click the **New shoot** split button and choose: - **Product Studio shoot**, for product photography through Product Studio. - **Campaign shoot**, for campaign images. This choice sets the shoot's journey. You can switch it later from the Inspector, and switching is non-destructive: it does not touch an existing plan. If a plan is already drafted for the old type, you'll see a reminder to redo the plan so the shots match the new type. Nothing is wiped automatically, but the plan won't reflect the new type until you redo it. --- ## Two Moments: Setup and Planning A shoot moves through two moments rather than a fixed set of steps. **Setup (no plan yet).** Drag product photos into the **Products** drop zone, or click it to browse (jpeg, png, webp). Once you have images in place, click **Draft the plan** in the header. **Planning (a plan exists).** The Products drop zone gives way to an inline plan board, with a collapsible "Show/Hide photos" strip if you want to see the source images again. The header button changes to **Redo plan**. --- ## How the Planning Agent Works When you click **Draft the plan**, a dedicated planning step takes over. This is a more deliberate pass than the regular chat: it reviews your products and their photos, the shoot type, your Skills and Instructions, your [Brand Memory](/brand-identity-getting-started), and your chosen fashion models, set locations, and perspectives, then writes out the plan. The summary the agent replies with **is** the plan, so it's worth reading closely. It tells you what it is going to create for each product, which looks it built, and it will call out anything it noticed (for example a photo whose view type looks wrong, covered in the Products, View Types, and Mismatches section below). You do not drive planning with a separate button. The chat is where planning happens: **Draft the plan** and **Redo plan** simply send the agent a starting message for you. --- ## Skills and Instructions Both live in the **Inspector** panel on the right, and both steer how the agent styles and completes your shots. They are not the same thing. - **Skills** are reusable, team-wide styling rules (a short title plus a description, for example "always use two models" or "shoes are accessories, never the hero"). Once saved, a Skill applies to every relevant shoot automatically. Skills are journey-specific: a Skill written for campaigns is ignored on a Product Studio shoot and vice versa, so you never have to detach one before switching types. The agent can also offer to save a good one-off decision as a Skill mid-conversation, which is how your Skill library grows over time. - **Instructions** are one-off, per-shoot notes. Use them for something that applies to this shoot only and shouldn't become a standing rule. The shoot-type switch also lives in the Inspector. ### How the agent decides styling When the agent decides how to style and complete a look, it follows this order of priority, strongest first: 1. **What you type in chat, and your batch Instructions.** These win over everything else. 2. **Your saved Skills.** 3. **Your Brand Memory.** Used mainly to complete bare looks in your brand's spirit (for example, your staple white tee). 4. **Brandmachine's built-in defaults.** Even with no Skills saved, sensible defaults already apply: every look is a complete, styled outfit rather than a lone garment (bare pieces are completed with neutral, unbranded basics); products are grouped by their filename convention; only the views your photos actually cover are shot (a missing view is never invented); accessories support the outfit unless the accessory is itself the key product; and each paired item stays consistent across every frame a product appears in. If a shot ever looks off-brand, check your Instructions and Skills first, since they override brand-level and default settings. --- ## Two Ways to Change a Plan There are two different ways to change a plan once one exists, and they behave very differently. Knowing which to use saves the most time. - **Ask for a specific change in chat.** Tell the agent exactly what you want ("drop the cap from the denim look", "swap the model on frame 3"). It makes a targeted edit, changing only what you asked about. Everything else, including previews you have already reviewed, stays exactly as it was. - **Redo plan.** This is a full reset. The agent discards every draft item and its previews, including any edits you made directly on the plan board, and drafts a fresh plan from your products, Skills, and Instructions. A confirm modal warns you first. Use it when you want to start the plan over, not for a small tweak. In both cases, anything you have already generated as finals is never touched. A couple of edits the agent will decline, with a plain explanation rather than an error: a look whose preview or generation is still running (wait a moment and retry), and a look that has already been generated as a final (that one is finished inventory, so edit it in Product Studio or start a fresh plan). --- ## Products, View Types, and Mismatches The upload section is called **Products** (previously "Product Images"). When you upload a photo, Brandmachine scans it and assigns a **view type**: Front, Back, Other, or Cropped. It also groups photos into products by their filename convention (for example `TEE-01_front.jpg` and `TEE-01_back.jpg` become one product). Perspectives only render the views your photos actually cover. If a product has no back photo, the agent skips the back rather than inventing one. **If a view looks wrong, fix the tag before you build.** If a front photo is scanned as "Other," the agent flags a view mismatch in its summary and will not pretend the view exists. The fix is quick: open that image, set the correct view type from the dropdown (Front, Back, Other, Cropped), then ask the agent to redo or adjust the plan. The agent cannot correct a mistagged view for you, it can only avoid faking it, so this is the first thing to check when a view you expected is missing. --- ## Hero Products For each look, the agent picks one **hero** product automatically at plan time, guided by your Skills and Instructions (for example "shoes are accessories, not heroes") or, if nothing applies, by choosing the look's key garment. Hero only matters for perspectives set to **Use Hero Inputs Only** (packshot-style shots), which render only the hero product's images. All other perspectives (outfit and lifestyle shots) include every product regardless of the hero flag. You can override the automatic choice afterward. **MAKE HERO** on the Product Studio detail page is a per-image correction, not the starting point. See [Product Studio](/product-studio-getting-started) for the full hero workflow. --- ## Reviewing the Plan The review board has a **Plan / Preview / Finals** lens toggle, so you can flip between the drafted plan, generated previews, and finished finals in one place. Product Studio shoots add an **Overview / Detailed** density toggle for scanning many items at once. A progress header keeps a running count of how many items are planned, previewed, and generated. --- ## Generating Previews and Finals Once you're happy with the plan, two actions turn it into images. Each one opens a confirm modal stating the shot count and cost before anything runs. - **Generate previews**: free, low-resolution, meant for review only, not releasable. - **Generate finals**: full-resolution and releasable, costs credits. Generate previews as many times as you like while you refine the plan, then generate finals only for the shots you're ready to pay for. --- ## What to Expect: Campaign vs Product Studio **Campaign shoots.** If you don't say how many frames you want, the agent proposes a varied set (around 8 to 12). If you do name a number, it produces exactly that many. Every frame is a distinct scene (a different setup, moment, or angle), never a repeat, and when you have more than one set location the frames spread across them. **Product Studio shoots.** The number of images isn't a target you set. It's simply your products multiplied by the perspectives you choose, so the useful question is which views and perspectives you want, not how many images. Perspectives are the camera-angle and framing presets you configure in Product Studio settings; the agent picks from the ones your team already has, filtered to the views your photos cover. One consistency rule applies to both: once the agent decides how to complete a bare garment (say, pairing a skirt with a plain neutral top), it reuses that exact pairing across every frame and perspective the product appears in, rather than varying it. --- ## What Shot Planner Creates Finals land in: - **Product Studio**, for Product Studio shoots, where you review and finalize them in [Workbench](/workbench). - **Campaigns**, for campaign shoots. The shoot itself keeps its full history (plan, previews, and finals), so you always have a record of what it produced. --- ## Inventory Everything you generate is permanent inventory, not a temporary draft. Generated items: - Survive a **Redo plan** on the same shoot, with favorites and releases preserved. - Appear under the **Finals** lens as dated history, so you can look back at earlier rounds. You do not need to start a new shoot to iterate. Redo the plan on the same shoot and keep working from there. --- ## Before You Plan The agent needs a few things in place before it can draft, and it will tell you what is missing rather than guessing: - **Product photos** uploaded to the shoot. - **A fashion model.** Shots are built around a model, so set one up in [Fashion Models](/fashion-models-getting-started) first. - **A set location**, for campaign shoots. Create one in [Set Locations](/set-locations-getting-started). - **Perspectives**, for Product Studio shoots. Configure these in Product Studio settings. --- ## FAQs **What's the difference between messaging the agent and clicking Redo plan?** A message asking for a specific change is a surgical edit: it touches only what you named and leaves everything else, including your previews, in place. **Redo plan** is a full reset that discards every draft and its previews (including edits you made on the board) and drafts a fresh plan. Your generated finals are safe either way. **A view I expected is missing. Why?** Usually a view-type mismatch. Open the image, set the correct view type from the dropdown, then ask the agent to redo or adjust the plan. The agent will not invent a view your photos don't cover. **Who decides which product is the hero?** The agent does, per look, at plan time, guided by your Skills and Instructions. You can override it per image later with MAKE HERO on the Product Studio detail page. Hero only affects "Use Hero Inputs Only" perspectives. **How long does planning take?** Usually under a couple of minutes, but it can take up to a few minutes for large shoots. If you ever see a planning error and then find a plan appeared anyway, that can happen when the result lands just after the app stopped waiting. Refresh and check before redoing. **Can I switch a shoot between Product Studio and Campaign?** Yes, from the shoot-type switch in the Inspector. Switching is non-destructive: it doesn't touch your existing plan. You'll be nudged to redo the plan so the shots match the new type, but nothing is wiped until you do. **Can I re-run a shoot, or do I need to start a new one?** You can re-run it. Generated items are permanent inventory, so a **Redo plan** on the same shoot keeps your existing favorites and releases and adds the new plan on top. You don't need a new shoot to iterate. **Does Shot Planner cost credits?** Drafting a plan and generating previews are both free. **Generate finals** costs credits based on the number of shots. See [Billing & Credits](/billing-credits) for pricing. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================ # Workbench Workbench is your review-and-finalize view for [Product Studios](/product-studio-getting-started). It shows your studios as a board organized by how much work is left on each one, and gives you the finalize-and-download flow. If you've used Brandmachine before, Workbench is the renamed and redesigned version of the old **Review** page. --- ## How Studios Move Across the Board Studios sort themselves. You don't drag cards between columns; they move as you generate and export. - **To Do.** Nothing generated yet. The card shows the studio title and when it was created. - **In Progress.** Something has been generated, but not every generated perspective has an exported image yet. The card shows how far along you are ("2 of 4 perspectives exported"). - **Finished.** Every perspective you generated has at least one exported image. There's nothing left to action, so the studio leaves the board and collects in a **finished** row underneath it. Click that row to expand it if you need to get back to one. Each column header restates its rule, so you never have to remember which is which. Each card shows a thumbnail (your favorite if you've picked one, otherwise the newest output) and the studio title. ### Only the perspectives you generated count A studio is finished once everything **you actually generated** has been exported, not once every perspective you have set up has been generated. So if your team has fifteen perspectives configured but for this product you only wanted a front and a back shot, that product is finished as soon as those two are exported. It won't sit on the board waiting for thirteen shots you never intended to take. ### One export per perspective is enough You don't have to export every image. Product Studio usually generates several variants per perspective and you keep one. Exporting that one is enough to count the perspective as done. --- ## Time Range Above the board you can switch between **30 days**, **90 days**, and **All**, which filters the board by recent activity. It defaults to the last 30 days. "Activity" means the newer of when the studio was created and when it last generated an image, so a studio you set up months ago but generated into yesterday still counts as recent and stays on the board. This keeps old, abandoned studios out of the way without deleting anything. Whenever the filter hides something, the board tells you how many ("41 older studios hidden"), and switching to **All** brings everything back. The range is part of the page address, so refreshing or sharing the link opens the same view. --- ## Two Ways to Open a Studio Each card has two click targets: - **Card body.** Opens the full studio detail view, where you can upload more inputs, edit styling, mark hero products, and regenerate. - **Quick review button** (two-column icon). Opens the **Finalize Selection** flow. --- ## Finalize Selection The Finalize Selection page walks you perspective by perspective. For each one: - If you already favorited something, the card shows that favorite with a heart and arrow controls to swap. - If nothing's favorited yet, the card shows all outputs in a carousel with a **Quick pick** tag. Navigate with the arrows to pick one. Each card displays the perspective name and (if set) a small color swatch chip showing the target background hex. When you've made your picks, click **Finalize selection** at the top. A tooltip tells you exactly what will change ("Will favorite X image(s) and unfavorite Y other(s)"). Finalizing records which image you picked. It's the **export** in the next step that finishes the studio and takes it off the board. --- ## Downloading Your Images The Finalize Selection page has two download buttons at the top, both disabled until you've finalized: - **Download high-res (PNG)**: the original lossless master, exactly as generated. It isn't tagged with AI-generated metadata, so it may not meet marketplace requirements on its own. - **Download JPEG (sRGB)**: a smaller, sRGB-tagged file. This is the version that carries the [AI-generated content label](/ai-metadata) some marketplaces require. Downloading is what exports an image, and exporting is what moves a studio to finished. Neither button runs an upscale. Product Studio generates at full resolution (4K) up front, so downloading just releases and exports the image you already generated, nothing is enlarged or re-rendered in between. Both buttons download every finalized selection at once, named after the perspective. Use PNG when handing off to retouching or a designer who wants the untouched master; use JPEG when uploading straight to Shopify or a marketplace that requires the AI-disclosure label. --- ## FAQs **Why does a studio disappear from the board?** Because it's finished: every perspective you generated has an exported image, so there's nothing left to do. It's in the **finished** row underneath the board, which expands when you click it. **A studio came back to In Progress. Why?** You generated a new perspective for it. That perspective doesn't have an export yet, so the studio has work outstanding again. **Do favorites decide which column a studio is in?** No. Favorites record which image you picked, and set the card thumbnail, but exporting is what moves a studio across the board. **What if a perspective hasn't generated yet?** It won't block Finalize, and it won't hold the studio back. You can finalize and export what's ready, and the studio counts as finished based on what you've generated so far. **I can't find an older studio.** Check the time range above the board. It shows the last 30 days by default; switch to **All** to see everything. **Where does my high-res download go?** Wherever your browser sends downloads. Files are named with the perspective so they're easy to drop into a catalog tool. For further help, contact [support@brandmachine.shop](mailto:support@brandmachine.shop). ================================================================================