One kurti, four marketplaces, four different photos

The single most expensive assumption in Indian fashion e-commerce is that one set of product photos will work everywhere. It will not. Meesho wants a square image on a pure white background. Myntra rejects square outright and requires a vertical portrait with a human model in it. Selling the same kurti on both means producing two genuinely different photographs — not two crops of one.

Here is what each platform asks for, where they conflict, and the shortest path to satisfying all four.

The side-by-side

The conflict that costs sellers the most

Meesho and Myntra want incompatible images, and this is the one that catches people out.

• Meesho's primary image is square, on pure white, and does not need a person in it. A clean flat-lay or ghost-mannequin shot is ideal.
• Myntra's primary image is vertical, and the garment must be on a human model or an invisible ghost mannequin. A square flat-lay fails on both counts at once.

You cannot crop your way from one to the other. Cropping a 1:1 square to 3:4 throws away the top and bottom of the frame, which on a full-length kurti is the hem and the neckline — the two things the listing exists to show. And no crop puts a model into a photograph that never had one.

The practical answer is to shoot for the strictest platform first. A 3:4 portrait on a model at 1500×2000 can be cropped down to a clean 1:1 for Meesho, Flipkart and Amazon, provided the garment is centred with room around it. The reverse never works.

Why listings actually get rejected

Across all four platforms, the same handful of mistakes account for most rejections, and none of them are about photographic quality.

1. 1The background is not white enough. Off-white, cream and light grey all get flagged where pure white is required — this is the single most common Meesho rejection.
2. 2There is a watermark, logo, brand name or price tag on the primary image. Sellers add these to stop image theft; the platforms read them as promotional overlay and reject.
3. 3Wrong aspect ratio — usually a square uploaded to Myntra.
4. 4No model on an apparel listing where one is mandatory.
5. 5Too few images. Myntra in particular expects a set, not a single photo.
6. 6The garment is wrinkled. Myntra rejects visibly creased or badly fitted samples more readily than the others.

A workflow that feeds all four

1. 1Produce one master image per design at 2000×2667 px (3:4), on a model or ghost mannequin, on white or very light neutral, with the full garment in frame and margin around it.
2. 2Export the Myntra version at 1500×2000 directly from the master.
3. 3Centre-crop the master to 1:1 and export at 1500×1500 for Meesho, Flipkart and Amazon. Check that the hem and neckline both survive the crop.
4. 4Add the supporting images each platform wants — detail shots of the embroidery and neckline, a back view, a flat-lay.
5. 5Keep the master. When a platform changes its specification, you re-export rather than reshoot.

That last point is the one that pays for itself. Sellers who keep a high-resolution master handle a specification change in an afternoon. Sellers who only kept the exported JPEGs courier their stock to a studio again.

Where AI catalogue images fit

This four-way split is the specific problem AI product photography is good at, because the expensive part of satisfying it conventionally is producing both a model shot and a clean product shot of the same garment. Both come out of the same upload — an on-model portrait for Myntra and a clean studio shot on plain backdrop for the others — and you crop to each platform's ratio afterwards.

It is not a complete answer. AI images cannot verify fit, and if a platform ever requires photography of the physical item as sold, generated images will not satisfy it. Read each marketplace's current policy on AI-assisted imagery before you build a catalogue on it — the rules in this area are moving, and a policy change is a worse surprise than a rejected image.