[tobr]

I don’t know that I understand the value of this. The examples show structural elements like doors and windows being changed. To be of any use it would need to know what in the input image is impossible or impractical to change, and keep those things exactly identical in the output. Otherwise it’s like looking at a somewhat similar room, and then you might as well just go look for actual photographic inspiration, or use any image generator to create an image from scratch.

[paulcn]

Yes, these random layout glitches can occur with certain styles. I’m working on improving this. It's currently just an MVP, and I’m gauging interest before investing more time into it.

[theendisney]

It is certainly interesting and you could take it in coutless directions. For example, say one provides a good number of photos of each piece of furniture in a room. A good number of pictures of the room dimensions then get a slide show with every possible configuration.

This would be worth something.

People do this by moving things around and with paper cutouts. Much easier to pay 20-30 and see all variations.

You then take pictures from something you want to buy and pay the fee again to see 50 or 200 implementations.

[paulcn]

Man, this idea of visualizing all possible furniture configurations in a room is genius! need to build this

[theendisney]

It was just an example. I think I can make them up non stop. [Say] Product placement in the current implementation would probably allow for more profit, a free to use tool and interesting updates. One would have to check periodically. You would also get the trendy styles in your tool for free.