WeShop Canvas: Apparel Images & Video Workflow

You have a new outfit to launch. You need a clear front view for the product page, a different pose for a campaign and perhaps a short clip for social media. They all need to show the same garment. If the neckline changes between pictures, you have a problem to fix before posting.

You can organize this project in WeShop Canvas. The pink outfit below is a worked planning example. Adapt the prompts to your own item; we have not generated or tested this particular campaign. Interface examples are grounded in the official apparel tutorial, recorded with Canvas 0.1.10; check the interface and available settings in your installed version.

1. Decide which pictures you actually need

Imagine a pink two-piece outfit that needs a listing image, a three-quarter campaign image and an optional motion clip. Write down what would make a picture unusable, even if you like its lighting.

• Product: identify the exact item and colorway. List details that must stay unchanged, such as the neckline, sleeve length, seams and logo placement.
• References: choose a clear person image and a garment image. Add actual detail or back-view photos when those surfaces need to appear.
• Deliverables: name the required views, channel dimensions and whether video is needed. Confirm each channel's current requirements before final sizing.
• Creative direction: describe the backdrop, lighting and pose separately from the product details.
• Acceptance: decide who checks product accuracy and what would require a new attempt.

Do a quick reference check before you start. Pick a garment photo where you can see the areas you plan to show. A bag hiding the waistband or hair covering the neckline leaves less information to work with. If you only have a front photo, get a real back photo before asking for a back view you intend to use on a product page.

Keep a close-up of any detail that could decide a purchase, such as a fastening, knit texture or printed logo. You can use it during review even when it is not part of the generation request. If you are still choosing source pictures, the WeShop AI product photography guide covers the related preparation work.

A reusable brief might read: “Pink long-sleeve top and matching shorts. Neutral studio setting. Deliver a front view and a three-quarter view. Preserve the garment silhouette, color and visible stitching. No added text or accessories. Review the stills before deciding on video.”

2. Tell Canvas what each reference is for

A person reference answers who appears in the picture. A garment reference answers what they wear. A mood reference may suggest lighting, but should not silently replace the garment's color or construction.

The official tutorial imports two images and selects them as references for an outfit request. Before sending your own request, check which images are attached and describe their roles explicitly. Avoid a vague instruction such as “make it look like this” when several images are selected.

Suggested request: “Use the person reference for identity and the outfit reference for clothing. Create one full-body studio image. Keep the outfit's neckline, sleeves, shorts length and color faithful to the reference. Add no text or accessories. Show the plan before generating.”

3. Check the plan, then check the first image

In the recorded manual workflow, the Agent proposes a plan; manual creation adds connected Drafts. Generation is a separate action. The tutorial also notes that planning can use conversation-model points, while media generation requires its own paid confirmation. Review the current usage information before proceeding.

Before generating, check the attached images and output count. Ask for one image while you are still deciding on the direction. Once it returns, open it and compare it with your garment reference before making more pictures from it.

Approve the base image against the original garment first. Check the garment outline at full size, especially near hands, hair and the waist. If an important seam is missing, fix the base before multiplying the same mistake into more views.

4. Make each pose from the same base image

Use the same checked base image for each pose so you can compare them fairly. If you repeatedly edit the newest result, differences accumulate and it becomes harder to tell which change caused the garment to drift.

Suggested request: “Use the approved base as the reference. Plan three separate images: a relaxed front-facing stance, a three-quarter stance with one hand at the waist, and a gentle walking pose. Preserve identity, outfit construction, color and studio background. Keep the full person in frame. Show the plan before generating.”

Try one pose first and see whether the clothing holds up. If the base is sound but one pose fails, revise that branch's pose instruction. If every branch repeats a clothing error, return to the base or improve the source reference.

The official tutorial also demonstrates Skill-based edits and a short video derived from a selected pose. Choose an edit to solve a specific problem. A tone adjustment is useful for uneven lighting; it cannot establish a missing product detail.

5. Check the details before choosing a favorite

Compare each candidate with the real item and the brief. Use the same checks for every branch so that “best looking” does not become the only selection rule.

• Garment: inspect the neckline, cuffs, hem, seams, fastenings and logos. Reject invented text or altered construction.
• Color and texture: compare with the product reference. A filter that makes pink appear purple may be unsuitable for a listing.
• Body and occlusion: check fingers, limbs and the boundary between clothing and skin. Look for fabric merging into a hand or disappearing behind hair.
• Framing: confirm required product areas remain visible after the intended channel crop.
• Evidence: mark any view whose hidden details were inferred. A generated back view does not prove the actual back design; use a real back photograph for verification.

Write a short reason beside your chosen version in your working notes: “Approved for front listing view; hem and sleeve checked; three-quarter branch needs hand correction.” That tells whoever picks up the project exactly what to fix.

6. Start the video with a still you trust

A good still gives you a starting point for video, but you still need to check the motion. For this example, begin with restrained motion rather than asking for walking, turning and camera movement together.

Suggested request: “Use the approved three-quarter image as the first frame. Create a five-second silent showcase clip with a fixed camera and subtle breathing and weight shifts. Preserve the person, garment shape, color and background. Show the plan before generating.”

Watch the whole clip. Check garment edges during movement, inspect the first and last frames, and look for changes in the face, hands or logo. If movement alters the clothing, simplify the motion or keep the approved still as the deliverable. A successful generation is not the same as a publishable clip.

7. Fix the problem you can actually see

When a result is wrong, name the visible error before changing the prompt. “Make it better” gives you very little to compare. “The left cuff is missing; restore the cuff from the garment reference and keep the pose and background” gives you a specific check for the next attempt.

Suggested correction: “The shorts hem in this result is shorter than in the garment reference. Keep the person, pose and studio setting. Restore the shorts length and visible hem detail to match the reference. Create one revision and show the plan first.” This is a request to try, not a guarantee that the model will recover the detail.

Choose where to restart based on where the error first appeared:

• If the base already has the wrong neckline, return to the base and its references. More pose variations will give you more pictures with that mistake.
• If only the walking pose has a distorted hand, keep the other views and redo the walking branch. Consider a pose where the hand does not cover the garment.
• If a color edit changes the fabric or logo, return to the original pose image and narrow the edit. Do not feed that altered version into the video.
• If the video changes the outfit during a turn, reduce the movement. If it still fails review, deliver the stills and leave the clip out.

A Draft with no image may simply be waiting for generation in manual mode. Check its state before creating another copy. If the interface shows a running or failed task, read that status and the displayed recovery options. Repeating the request before you know what happened makes it harder to identify which output belongs to which attempt.

Keep a small review note for each attempt: the source you used, the change you requested and whether the error remains. After repeated failures, reconsider the reference or pose instead of adding more adjectives to the same prompt. A clearer source photo may be more useful than another round of “high quality, realistic, perfect detail.”

8. Save the files you want someone else to use

Keep the original photos and your chosen outputs easy to tell apart. A practical filename might be “pink-set_front_listing_approved_v02”; include the item, view, purpose and revision rather than relying on “final”.

Open the saved files before sending them. Check the dimensions, crop and playback, and send the chosen set with a short note about anything that still needs attention. Keep the original references available for the next campaign.

Before you call the set finished, look at the images together at the size they will be used. Check that the pink outfit stays the same color across views and that the crop still shows the details you need. Then inspect those details at full size. A thumbnail can hide a broken seam; a full-size view can hide the fact that the sleeve gets cut off in the final layout.

If another person will publish the files, include the intended use beside each filename. For example: “Front view: product listing. Three-quarter view: campaign placement. Video: reviewed silent clip.” Keep trials out of that delivery folder so nobody has to guess which version you chose.

For the exact recorded controls and screenshots, follow the official step-by-step tutorial. To explore the product, visit the WeShop Canvas introduction. Start with one item and one accepted base image; expand the project only when the garment remains faithful.