They're probably the only retailer with a big enough range and significant enough global presence for it to work well for users and scale.
Show HN: I made an app that uses AI to change your interior
21–26 of 26 posts
Re: Show HN: I made an app that uses AI to change your interior
#22Earlier quoted context omitted.
I am still not sure whether it was a typo in his math. He said "100,000 renders * 1/2 cent per render"[0] when someone asked him how come he was paying that much. And when someone asked a similar question to how the math worked out, there was no reply. Still not sure how to interpret this, no matter how I look at it. 100k renders at 1/2 cent each comes out to only $500 total. Someone guessed that he could've meant $0…
Sorry I'm not US, with 1/2 cent I meant 1 to 2 cent. Actually it's more like 1-4 cents depending on image dimensions. A square image is 1 cent, a wide image can be 4 cents. But yes it's not cheap
Re: Show HN: I made an app that uses AI to change your interior
#23This is awesome, but I'm a bit concerned about the "latest renders" feature potentially doxxing people. Are you checking for identifiable information somehow?
Images would still contain EXIF data from your phone. So you could take a picture, then take a screenshot of that picture before uploading the screenshot.
Re: Show HN: I made an app that uses AI to change your interior
#24Have a chat with IKEA - if you could integrate this with their inventory and then link to the specific items shown in generated images (for a revenue share or some other arrangement) you could have a slam dunk on your hands. They're probably the only retailer with a big enough range and significant enough global presence for it to work well for users and scale.
Re: Show HN: I made an app that uses AI to change your interior
#25Re: Show HN: I made an app that uses AI to change your interior
#26It's a very clever idea for an awesome, interactive app but in practice the results are extremely terrible. The output has almost nothing to do with the input, the outputs are all really just slight variations on what was clearly a very small training set, and the results are hardly even "possible" or "sane" suggestions with extremely obvious AI artifacts like impossible shapes or walls that are also chairs that are…