Walmart 电商图片生成与编辑

Walmart Ecommerce Image Generation and Editing

Create retail images that make item, variant, package and scale unambiguous. Online visuals must match what a shopper receives and must not imply a store placement, certification, price, rollback or bundle that the merchant did not supply.

Retail item record

Collect item ID/SKU, UPC-linked variant, product angles, retail package, included pieces, dimensions, materials, approved claims, warnings and omnichannel brand rules. Use this record as the sole source of product facts.

Scenarios and commands

1. Item-page primary image

python3 "$SKILL_PATH/scripts/imagegen.py" generate \
  --prompt 'Walmart Marketplace primary image for the supplied retail item. Preserve exact product, retail package, label, color, variant and included pieces. Show the complete item large on a clean background with realistic shadow and clear edges. No price, rollback, pickup badge, rating, certification or unprovided text.' \
  --image /path/to/item-front.png \
  --image /path/to/retail-package.png

2. Package-contents image

python3 "$SKILL_PATH/scripts/imagegen.py" generate \
  --prompt 'Walmart package-content image showing exactly the approved main item, two attachments, power cable and guide, each once and with correct connector and scale. Clean top-down arrangement; do not add a bonus, replacement part, multipack or warranty card.' \
  --image /path/to/package-contents.jpg

3. Dimension and scale base

python3 "$SKILL_PATH/scripts/imagegen.py" generate \
  --prompt 'Retail dimension graphic base: accurate product shown front and side plus one realistic in-room scale view, with blank measurement arrows for merchant-approved numbers. Keep proportions consistent and do not generate dimensions, capacity, room size or performance.' \
  --image /path/to/product.png

4. Variant gallery

python3 "$SKILL_PATH/scripts/imagegen.py" generate \
  --prompt 'Create consistent Walmart listing images for the verified single, 2-pack and 4-pack variants. Lock product appearance, package design, camera, background and shadow; show the correct quantity for each variant and do not mix accessories or invent bundle savings.' \
  --image /path/to/item-and-packages.jpg \
  --batch 3

5. Omnichannel lifestyle image

python3 "$SKILL_PATH/scripts/imagegen.py" generate \
  --prompt 'Walmart product lifestyle image in a credible family home, preserving exact item and use method. Show one practical everyday benefit with realistic scale and safe caption space. Do not imply in-store availability, pickup time, price, guarantee or customer endorsement.' \
  --image /path/to/product.png

Retail QA

• Item, variant, package, label and included quantity match the catalog record.
• Pack-size images cannot be mistaken for a different offer.
• Dimension callouts use approved values and units after generation.
• No generated price, rollback, pickup, rating, certification or retailer badge.
• Lifestyle scene does not imply unsupported use or store availability.
• Verify current Walmart Marketplace image and category requirements.

Run

pip3 install requests
python3 "$SKILL_PATH/scripts/imagegen.py" init --skill-name walmart-ecommerce-image-generation-editing
python3 "$SKILL_PATH/scripts/imagegen.py" task --task-id 

The fixed image endpoint is public_model_nano_banana_pro. Use multiple references, batch, live parameters, routing, output directory and submit-only mode as needed.