Try any saree on yourself with AI: a copy-paste prompt (Cloakroom-style)

You may have seen the reel. As reported, Cloakroom uses an in-store kiosk: staff scan a saree, the customer gets a QR code, and the screen shows the customer already wearing that saree.

This page explains what Cloakroom is (based only on what has been reported publicly), then gives you a prompt you can use right now with an AI image tool to get a similar result for yourself — no kiosk and no developer.

What is Cloakroom?

Cloakroom is an AI saree try-on product built by Sohan M Rai, who goes by @zikiguy on Instagram. His studio is listed as @thetuesdaylabs.

Based on LinkedIn coverage, here is what has been reported about how it works:

• Staff photograph the folded saree.
• The AI identifies the pallu, borders, and body fabric from that single photo and uses them to recreate the drape.
• A customer scans a QR code, and a kiosk displays the customer wearing the saree.
• One LinkedIn post notes the project was built using code written through ChatGPT.

Sohan has not published his exact prompts or code anywhere we could find. We were not able to open his Instagram post to read its caption. So this page is not affiliated with Cloakroom and the prompt below is our own starter prompt, not Cloakroom's.

Make your own try-on in 3 steps — no coding

You need two photos and an AI image tool that accepts multiple image uploads plus a text prompt. Tools such as Gemini or ChatGPT can take image uploads with a prompt; check that your version accepts two images. Results will vary by tool and by photo quality.

Step 1 — Photo of the saree

Lay the saree flat on a clean surface or hold it folded. Make sure the border, body, and pallu end are all visible. Use natural light or bright indoor light. Avoid shadows across the fabric.

Step 2 — Photo of the person

A full-length photo of the person standing straight, facing the camera, on a plain background. Only use photos of people who have agreed to it. The cleaner the background, the better the result.

Step 3 — Upload and paste the prompt

Open your AI image tool. Upload the saree photo first (Image 1), then the person photo (Image 2). Then paste this prompt exactly:

Things to try if the result is not right

• If a pattern changes or disappears — regenerate, or add "do not change any pattern" a second time at the end of the prompt.
• If the pallu looks wrong — upload a second close-up photo of just the pallu and add "Image 3 is a close-up of the pallu" to the prompt.
• To change the drape style — replace "Nivi" with the drape name you want (for example: Bengali, Gujarati, Maharashtrian Nauvari).

For developers: building an app or kiosk

If you want to build something like Cloakroom rather than just use a prompt manually, here are three starting points. None of these are Sohan's code — they are separate projects and APIs.

Google Virtual Try-On API

Google's Virtual Try-On API (model ID: virtual-try-on-001) takes a person image and a product image and returns the person wearing the product. It is a general clothing API, not built specifically for sarees. The official documentation includes a Python code example.

Saree try-on app built with Claude Code + Gemini 2.5 Flash

A developer on LinkedIn shared how he built a saree try-on app for his aunt using Claude Code and the Gemini 2.5 Flash image generation API. He noted that a plain overlay approach looked bad until he switched to the Gemini image generation API. This is a different, independently-built project — not Cloakroom.

Open-source virtual try-on repo by satish-rathod

There is also an open-source repository called virtual-tryon by GitHub user satish-rathod — a separate project, not Sohan's. It takes a single floor-shot saree photo and produces an image of a fixed model wearing it, aiming to preserve the border, pallu and motifs. We only read its README; we did not run or test the code ourselves.