Subscribe
Sign in
FASHN Virtual Try-On — Transfer clothes from image to image [ Web app & API ]
Change the clothes in any model photo using another reference image with pixel-perfect accuracy. Available for commercial use through a user-friendly web app or developer API.
35
Replies
Aya Bochman
Maker
📌
Hello PH! 🐱
After working over a year on our virtual try-on model, we’re excited to officially launch it to the world!
FASHN is a self-funded, AI-first company focused on researching image-based generative models tailored for the fashion industry.
Our virtual try-on solution is built from the ground up, and carefully designed to meet the high standards brands expect when showcasing their collections.
Featuring:
⚡️ Lightning fast generation (~11 seconds per image)
🏞️ 384x576 output resolution
🫳 20 free generations upon signing up
🏦 Available for commercial use
How it works 👇
- Upload one image of yourself, a model, or an AI model (you can create that with our app)
- Upload one image of the garment (top, bottom, full-body)
- Choose the type, whether it's a flat lay photo or taken from another model
- Run and get the result! You can download the image and we're also creating a public link for you that you can use anywhere.
Our products:
🖥️ Web App
Our app offers a user-friendly interface that runs our virtual try-on solution behind the scenes, allowing you to easily create fashion photoshoots with a click of a button.
🌐 API
Build on top of FASHN by integrating our virtual try-on API into your own website or app.
Since we started, we've had one rule: Get the garment details spot-on before going big on resolution. Our latest release (384x576) sticks to that game plan.
We're definitely heading towards even larger, higher-res images, but we promise, we won't let the clothes lose their real look. We know how much this matters to fashion brands, and we're all-in to meet those high expectations.
Don’t forget to redeem today’s offer for the Pro plan using this coupon code: PH20OFF
⭐️ The Pro plan enables AI model generation + unlimited try-ons!
If you have any questions, we’re here for you!
— Aya & Dan
Share
Is your model based on an already existing model like stable diffusion? Or did you create this yourself? If you did create this yourself, do you mind sharing what kind of architecture you used?
Maker
@tim_pietrusky Hi Tim! Our model is not based or fine-tuned on an existing architecture.
Our latest version, that is currently in production, took a lot of inspiration from the architecture of Stable Diffusion 3 / FLUX, but it was adapted and trained from scratch for the virtual try-on task.
We collected an internal dataset of around 4M images and labeled them for this task, and we have invested more than $70k on GPU compute so far for experimenting and training our model.
Congratulations on the launch! The examples on the website look very promising. Is your model adjustable for different types of clothing like sportswear and formal wear?
Maker
@shachar_langer Thanks Shachar! sportswear and formal wear are part of our dataset, so it should work well. The challenge with sportswear can be the photography style, like dynamic poses, angles, etc. but I encourage you to try it on our app (it's free)!
Aya, congratulations on the launch.
1. Does your API offer customization for brands to tweak the AI model to match their specific styles?
2. Of course, I can’t not but ask. About the API itself, do you generate SDKs automatically and documentation?
Maker
@dmytro_krasun Thank you Dmytro!
To answer your questions:
1. Yes, we can fine-tune our AI model on a brand's collection to get even better try-on results and match their style. It is a custom, contract-based offering.
2. Regarding SDKs, we're definitely planning to do that for JS and Python! For now, you can check out our docs here: https://docs.fashn.ai/
Congrats for this launch Aya! Love the simplicity of the tool!
Looking forward to see more!
Dan and Aya have built an incredible product. Proud to be an early adopter and grateful to work with such talented founders! This app is the future of fashion🚀 🚀
Maker
@aharon_shapiro Many thanks 🙏 We need to do a use-case page/blog for your app for sure
Maker
@aharon_shapiro We're grateful to have you as an early adopter! Thank you for helping us improve 🤗
I've actually used this product and am wildly impressed with it. I can't wait to see its adoption across all sorts of eCommerce outlets, like Zalando, Amazon, and more thanks to their developer-friendly API. I also look forward to mass-market consumer adoption via their web user interface.
It is remarkable that a self-funded team of 2 was able to build so much!
Maker
@tejas_kumar1 Thanks Tejas! It means a lot coming from you. I am looking forward to hearing your podcast with @ayabo 🤗
Congratulations on the launch @ayabo and @danbochman 👏
I tried it a while ago but had a hard time generating quality images. The garment design was impressive, but the cloth folding was not as good. Also, the image surroundings, like legs or pets next to the person, were skewed.
I'm not sure if this has improved since, but I'll check it again.
Anyway, you have my vote, and I wish you much success.
P.S. If things improve, I might use your API to build a product I'm working on 🤫
Maker
@danbochman @samiralibabic
Hi Samir, thanks for the honest feedback!
If you haven't already tried our background preservation feature, which helps to keep the details in the background (If that's what you're referring to), I recommend to toggle it under Model Image Controls.
You can also read our user guide to get better results:
https://fashn.ai/blog/getting-th...
Either way we will keep doing our best to improve 🙏 thanks again!
Great job!
I wonder if fashion retailers or individual shoppers are your primary target audience?
Maker
@sergei_benkovich Thanks Sergei! Our solution is ideal for fashion brands who have limited resources to produce high-quality photos for their collections, but also developers who can build cool applications on top of our technology, for example, a Shopify plug-in for try-on
Maker
@florinmakes That's so great to hear Florin! If you got any ideas of anything we can do to improve, let us know 💛
Really looking forward to higher resolutions, and more versatile apparel types.
What would be awesome is that at some point you offer weighs so we can deploy it on our infrastructure and us to pay royalties and updates. Good luck guys!
Maker
@stefan_dragisic Thanks! Higher resolutions are definitely coming, just a matter of time (and $$$), our solution so far scales nicely and reliably.
Regarding offline weights, it is not just one model, e.g. we have a human parser which is also trained in-house in the pipeline. Maybe a deeper collaboration in the future that is not just based on money could work.