Skip to main content

Feature Request: Saved Images for Easier Design

Feature Request

Right now, I keep having to re-upload the same images of furniture and rooms to get the best results when transforming designs. Since the "Tweak/Transform" function can sometimes stray too far from the original image and doesn’t have an undo feature yet, it would be great if there was a way to save these images for easy access. I think this would streamline the design process and ensure consistently high-quality transformations.

Posted by vanillism ·
Status: Completed3 comments

Log in to comment and vote

Comments3

  • rlowell

    Team•

    Sep 2, 2025

    By ‘saving these images’, do you mean save the images of furniture and rooms?
    (If so, this is already in our roadmap and is something we’re working on this month).

    Also, we’re working on something really cool in the results page, where you can click a button to fix an ai result that is not architecturally accurate, plus we’re also making further upgrade to the ‘tweak’ consistency and output.

    • vanillism

      •

      Sep 2, 2025

      Yes, the images that users upload of their furniture/rooms. That’s awesome and I’m looking forward to seeing the changes! Really loving this tool.

    • Sue Fisher

      •

      Sep 2, 2025

      I saw the architectural accuracy fix mentioned in the last newsletter... that sounds great as there are parts of my house that the AI sometimes has problems with (angled walls, pony walls, alcoves) and it sometimes also widens rooms, but can you give us a bit more information on how it would work?

      For example:

      • Do you move/add/delete walls/windows/ceiling/etc. manually until they're correct? This could give the best accuracy but feels a bit hands-on compared to what Room Visualizer AI seems to be aiming at.

      • Does the AI have another go at reading the structure of the room from the original photo using a different algorithm? If so, is there a way to indicate the area that's wrong, or the type of problem? And if the second attempt at extracting the structure of the room is still wrong, can you keep going to get it more and more accurate?

      (I actually tried to have a look at the "advanced users" architectural change area that I haven't really played with to see if it gave any clues, which is when I discovered I couldn't log in any more; separate post logged for that.)

      Also, what are the downsides/tradeoffs in fixing architectural inaccuracy? There must be some, even if it's just extended processing time, otherwise the sensible thing to do would be to apply the architecture fix automatically all the time when generating AI designs. ;D

      Thanks