fantastic work by @EstebanSuarez go check out the example and fork it https://x.com/estebansuarez/st...

@LinusEkenstam it's very impressive, but the tech is not yet there I think It easily mess small details and you absolutely need to describe clothing item in details (@EstebanSuarez has this too, a prompt for each clothing), otherwise it start to mix them together The best way for me was to

@xakpc @EstebanSuarez You are talking about pipeline problems, nobody is suggesting a zero shot pipeline here, you could fix things like size estimate, proper alignment, and less fabric distortion. The main thing to focus on here is speed. 2.5 year ago I told ebay how to do virtual try-on. it

@LinusEkenstam It can help but the return rates will go up because the size will be displayed incorrect- that’s why real models with their actual size are displayed so you see exactly how a “Large” Jacket looks on a 185cm guy. We need more accurate models.

@dreambeam6 I don’t agree, if you know anything about return rates, you’d know that primary reason for return is not being able to visualize how you’d look in a multi-piece outfit. Size is less of an issue, but also solvable with AI. ~27% is the average return rate, if you move that. 0.5%

@LinusEkenstam Cool idea, but it often alter how the real product looks like. Web shops need disclaimers to avoid angry customers.

@RyszardRzepa This is a tiny problem, that will go away just like hands. Plus with Seeddream, Nanobanana, et al. you see that future unfolding infront of your eyes.

@LinusEkenstam This will backfire on brands, massively. Shoppers like to imagine how things look on them and assume they look just as "pretty" as the model showcasing it. Showcasing their face and body with the product will be a reality check and sales will go nowhere.

@_hk2025_ I don’t agree, and also it’s not about boosting sales, it’s about reducing returns. 27% global return rates, if you can move that needle just -0.5 % you have a massive opportunity infront of you

@LinusEkenstam would be cool to take body dimensions for a more precise fitting

@daniil_bh we can do that too.

@LinusEkenstam Ahah, Esteban's trying to sell us on augmented try-on... meanwhile I've been using Drift to automate my own outfit suggestions for years. Just saying

@LinusEkenstam Brands would benefit massively. In a world where data is gold, this is just another way to learn what people actually look like, and know exactly what to sell you for the next decade.

@LinusEkenstam Very cool idea . So simple, but a really powerful hook

@LinusEkenstam so cool! will cover in newsletter on monday http://shortcut.email

@LinusEkenstam Thanks for sharing!

@LinusEkenstam @Sephora does this with makeup that you can “try on” through the app, which is awesome.

@LinusEkenstam ikr this is hella sick

@LinusEkenstam I build this one

@LinusEkenstam Except its absolutely terrible at doing the most important aspect of how clothing fits - the relationship between body shape, materials and how things sit on a body - bet @dieworkwear would have lots to say about this...

@LinusEkenstam firsr result: return rate spikes up: "the product is not like in the image"

@LinusEkenstam Good work bro

@LinusEkenstam And finally standardize sizes.. the standards are there, make sure an M is always the same so I know what I fit..

@LinusEkenstam I did kind of the opposite: you upload your photo into a Chrome extension, and it add the clothing onto you. https://x.com/zat0ichi/status/...