When scale matters
The AI doesn’t inherently know how big your product is. A bedside table might render the size of a wardrobe, or a large sideboard could appear too small. If proportions and scale matter — and for product imagery, they usually do — you need to include real dimensions.
How image models understand dimensions
Our image models don’t “measure” objects the way humans do — they understand "proportional" dimensions.
So they understand that if a table is 100cm / 40" wide that a chair of 50cm / 20" width is half of that.
Adding the context of dimensions from the background, products you are adding in helps keep everything proportionally accurate.
How to include dimensions
Add the real height, width, and depth of your product in the prompt. Then explicitly instruct the AI not to alter the dimensions.
If you're using references for backgrounds, you can also add the dimensions of the back-wall for example
Example
✅ Good prompt: Create a lifestyle scene of this white chest of drawers in a modern bedroom scene. Keep the proportions the same, dimensions: height 77cm, width 60cm, depth 40cm. Do not change the dimensions of the chest of drawers |
This gives the AI exact measurements and explicitly says not to change them. The result will show the product at a realistic scale within the room.
❌ Bad prompt: Create a lifestyle scene of this white chest of drawers in a modern bedroom scene |
Without dimensions, the AI will estimate the product’s size, which often leads to products that look too large, too small, or out of proportion with the room.
Tips
Use either cm or inches — both work well
Always include all three measurements: height, width, and depth
Add “Do not change the proportions” or “Do not change the dimensions”
This is especially important for smaller items like bedside tables, shelves, and accessories that can easily be rendered at the wrong scale