How to get back and side views of clothes right with AI
AI gets back views wrong because the model never saw the back, so it draws a typical one. Photograph the back of every garment and give that photo to every back and three-quarter shot. Keep backdrop, light, lens and framing identical to the front shot so the carousel lines up, and check pockets, yokes, zips, labels and prints against the real garment before you publish.
Why back views come out wrong
An image model draws every view as a new picture, built from the prompt and the references it was given. Shown only the front, it fills in the back from what backs usually look like: a yoke on a shirt that has none, patch pockets on jeans that have welt pockets (slits with a narrow finished edge), a center-back zip on a dress that closes at the side. Side views fail the same way, because front and back flat lays never show the side seam or how the garment turns at the hip.
Shoppers look for the back. Amazon’s 2021 clothing imaging guide, citing its customer studies, asked for five to seven images per product, including a clear front and back, and marked front and back on-figure images as "extremely important". Its current product image guide has variant codes for BACK and SIDE shots and recommends at least six images beyond the main one, plus a video.
What each view needs as a reference
| Shot | References to send, in this order | What to name in the prompt |
|---|---|---|
| Front | Character sheet, front flat lay | Full body front view and the garment detail list |
| Back | Character sheet, back flat lay, back detail close-ups | Full body back view and each back detail by name: pockets, yoke, vents, labels, print |
| Side | Character sheet, front and back flat lays, a side photo if you have one | Where the side seam sits, how the hem changes from front to back, pocket depth |
| Three-quarter | Character sheet, front flat lay, back flat lay | Which side faces the camera, and which back details should show |
| Back detail | Back flat lay and the detail close-up | Macro framing on the detail, in the same light as the store photos |
Send what a view needs and nothing else. On our surf apparel job, the kept character sheet used 5 references, after attempts with up to 8 were not kept. Leave the front flat lay off a straight back shot, where it only gives the model front details to copy onto the back. To shoot the back flat lay, see how to photograph your product for AI.
Keep the carousel lined up
A store carousel reads as one garment turning when every view shares the backdrop, light, lens, camera height and crop. On our surf apparel job, the back shot kept the same cyclorama (a curved studio wall with no visible corner), light and framing as the front shot, so the two could swap in the carousel without a jump. On our trousers job, the backdrop and lighting sentences were copied word for word into every store-photo prompt, and only the framing and pose sentences changed.
Backdrop: [color name and hex], a seamless [paper or cyclorama] that meets the floor with no visible seam, no props. Light: [a large soft key from the upper left at 45 degrees], even fill, a soft contact shadow under the shoes. Camera: [85mm] lens at [f/8], camera at [chest] height, level, no tilt. Framing: the frame begins above the head and ends below the shoes, the whole figure inside the frame with clean margin. Color: colors stay true to the flat lay, never shifted, never saturated. Changes per view: [full body back view, standing upright, arms relaxed at the sides]
After rendering, crop every view to the same aspect ratio and the same head and foot margins. If the figure’s size drifts between views, fix it in the crop rather than paying for another render.
Details that go wrong on the back and side
| Detail | Common error | Fix |
|---|---|---|
| Back pockets on jeans and trousers | Wrong count, patch instead of welt, invented stitching designs | Send a close-up of the back pocket and name its type and count |
| Yoke, the shaped panel across the upper back or seat | Added where there is none, or the wrong shape | Write "no back yoke", or describe its shape and seam |
| Center-back zip or seam | Added to dresses that close at the side | Name where the garment opens |
| Jacket vents and pleats | A single vent becomes two, a box pleat disappears | Name the vent count and pleat type |
| Neck labels and back prints | Letters redrawn, print moved or resized | Send a close-up, and paste the real print in afterwards if it must be exact |
| Prints that wrap around | The pattern stops at the side seam | Send a side reference and count the repeats |
| Hoods, collars and waistbands | Hood drawn flat, collar stand missing, drawcord added | Lay the hood out in the back flat lay |
When a back print or label must be exact, place the real one onto the photo in an editor instead of re-rendering until the model gets it right. More on this in why AI changes your product.
Check the views as a set
- Front, side and back placed side by side at the same size
- Same person in every view: face, hair, skin tone, build
- Same garment length in every view
- Every back detail matches the back flat lay, and no front detail has moved to the back
- The side seam sits where it does on the garment, and the hem line runs on from front to back
- Same backdrop color, light direction and shadow in every view
- Same crop, with the same head and foot margins
- The pattern crosses the side seam the way it does on the garment
Then run the full AI image QA checklist on each photo.
Questions people also ask
- Can I skip the back view if the back is plain?
- Shoppers still look for it, and a plain back is where AI adds seams and pockets that are not there. If you skip the on-model back, show the back flat lay instead. It is accurate and costs nothing extra.
- Do AI tools have back-view presets?
- Some do. Photoroom’s Virtual Model API, for example, lists a back pose preset. A preset sets the pose; the tool still needs your back photo to know what the back looks like.
- Should the back ever be the main image?
- Amazon’s 2021 clothing imaging guide suggested using the back as the main image for athletic wear, such as sports bras and athletic tops, when the back detail is the main selling point. Check Amazon’s current rules first; see Amazon product image requirements.
Where Overs fits
In Overs, the shot plan lists the reference pictures each photo needs, in order, such as the character sheet and then the back of the garment, and only those are sent to the image model. If a needed reference does not exist, Overs skips it instead of guessing with the wrong picture, so photograph the back before you plan the set.
Free for 40 photos a month. The AI that makes the photos is billed separately, on your own key, with no markup from Overs.