Wing Chan is co-founder of Sourceful and Riverflow, where he leads research and development for image models and a visual-production platform. His guides focus on keeping product details consistent in generated ecommerce images.[1][2][3]
Start with How to Keep Products Consistent in AI Images →
Career
Chan studied computer science at Cambridge and formerly served as CMO and CTO of The Hut Group. At Sourceful and Riverflow, his work combines model research with the practical requirements of consumer-product imagery. Riverflow is built by Sourceful as a platform for creating, reviewing and editing brand-specific visual assets.[1][3]
Contributions
Chan’s product-consistency guide separates two layers. Product identity stays fixed: package shape, material, label, color, scale and variant. The creative layer can change lighting, props, composition and setting. An approved reference image anchors each generated variation so a visually appealing result does not silently become a different product.[2]
Ideas
The guide makes review part of the workflow. A label can mutate or a scene can imply the wrong pack size even when the image looks convincing. Editing corrects local errors before export. The standard is whether an ecommerce team can publish an accurate representation of the item.[2]
Tools, writing & talks
Writing, research & talks
Start here
How to Keep Products Consistent in AI Imagesriverflow.aiA reference-led workflow separating fixed product identity from flexible campaign visuals.
Start with the fixed-product and flexible-scene distinction, then use the review checklist on one existing image. A finished product reference matters more than prompting experience.
Author[2]