Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Multiple connected panoramas generated? #2

Open
antoinemacia opened this issue Nov 23, 2023 · 6 comments
Open

Multiple connected panoramas generated? #2

antoinemacia opened this issue Nov 23, 2023 · 6 comments

Comments

@antoinemacia
Copy link

Hello,

Wondering how multiple generated panoramas can be connected to one another (essentially maintaining the style as well as whatever artefacts/furnitures have been generated)

Use case: multiple panoramas for a home

@ArcherFMY
Copy link
Owner

Hi,

It is a very interesting application and we have discussed within the team.

We think one straightforward method is that we can generate a main pano-image (like a living room) and convert it to a 3D structure using depth. Then find a new location to render a sparse pano-image and fill the holes via inpainting.

Another data-driven idea is that we can treat it as a pano-video generation problem. The AnimateDiff (https://animatediff.github.io/) can be a try. The motion module may help to maintain the continuity of different panoramas. But we have not tried yet.

@antoinemacia
Copy link
Author

Yeah nice!

The first option is pretty sound, this project could be of interest (I believe does more or less what you're describing)

Im curious how styled could be maintained across rooms however...

@ArcherFMY
Copy link
Owner

Yeah nice!

The first option is pretty sound, this project could be of interest (I believe does more or less what you're describing)

Im curious how styled could be maintained across rooms however...

Maybe we can train a LoRA for a certain decoration style :-)

@antoinemacia
Copy link
Author

Yeah nice! Also realise I have shared the wrong paper previously in regards to depth conversion, this is what I meant to share https://cyhsu14.github.io/OmniNeRF/

@ArcherFMY
Copy link
Owner

Yeah nice! Also realise I have shared the wrong paper previously in regards to depth conversion, this is what I meant to share https://cyhsu14.github.io/OmniNeRF/

Interesting! Thanks for sharing!

@zhanghongyong123456
Copy link

Yeah nice! Also realise I have shared the wrong paper previously in regards to depth conversion, this is what I meant to share https://cyhsu14.github.io/OmniNeRF/

Do you have any implementation cases? I found that a video is the effect you want, but I don't know what the implementation process is.

1.19.mp4

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants