Workflow test case study
Creating an Amazon-Ready T-Shirt Main Image Without Changing the Product
This case study examines how a ListingReady workflow converts an ordinary T-shirt photo into a clean main listing image while prioritising the accuracy of the original garment.
1. The objective
Create a clean main image without redesigning the garment
The objective was to create a professional front-facing main listing image from an ordinary T-shirt photograph. The product needed to remain recognisably the same, including its colour, print, collar, stitching, fabric appearance and proportions.
2. The original-image problem
The source photo was not ready for a marketplace listing
The source photograph was useful as a product reference, but its background, lighting, framing and presentation were not suitable for a polished main listing image. The required change was presentational—not a redesign of the product.
3. What commonly goes wrong
A cleaner image is not useful when the product changes
- ×The garment colour shifts after the background is replaced.
- ×The print, artwork or logo is redrawn instead of preserved.
- ×The collar shape or sleeve proportions change.
- ×The fabric becomes unnaturally smooth or plastic-looking.
- ×The product is not centred or framed appropriately.
- ×The AI adds a floating shadow, props or unrelated objects.
4. The ListingReady approach
Constrain the edit and define what must remain unchanged
Reference-only instruction
The uploaded T-shirt is treated as the only product reference, reducing the chance that the model invents a replacement garment.
Explicit preservation checklist
The prompt names the shape, proportions, stitching, collar, sleeves, folds, fabric texture, print, artwork and colours that must remain unchanged.
Controlled background
Only the background is replaced, using pure white #FFFFFF for a clean main listing presentation.
Restricted composition
The prompt requires a centred front view and excludes models, mannequins, hangers, props, reflections and unrelated objects.
Natural commercial lighting
Soft studio lighting and a restrained contact shadow are requested so the product remains readable without appearing to float.
5. Original and result
Compare the source product photo with the published workflow result
Original photo vs ListingsReady result
Compare the uploaded product reference with the image created using this workflow.
Original product photo
Uploaded reference
ListingsReady result
Click the image to zoom.
6. Result evaluation
Evaluation against the workflow objective
7. Corrections and limitations
Final review is still required
AI-generated product images should be compared closely with the original item before publication. Sellers should inspect colour accuracy, artwork, stitching, garment edges and background purity, then check the current rules of the marketplace where the image will be used.
This case study evaluates visual workflow performance. It does not measure customer conversion, sales, revenue or advertising performance.
8. Testing information
- Tested
- July 2026
- AI
- ChatGPT image generation
- Model
- GPT-5.5
- Typical attempts
- 1–2 generations
- Workflow version
- 1.0
9. Workflow used
Amazon Main Listing Image — T-Shirt
Open the complete prompt, recommended settings, common mistakes and fix prompts used for this test.