-
Notifications
You must be signed in to change notification settings - Fork 88
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
Question about 2D GS #42
Comments
I have same question. Have you see mesh file? how about mesh quality? |
I didn't export the mesh. How is your mesh result? |
I used 30 images, After instantsplat I got pose and point cloud. Then put them into 2dgs. The mesh is so bad! |
@Boris-Cui Hi, thank you for your interest! It’s possible that during the early stages of training, the scale of the Gaussian points is relatively small. However, when photometric constraints are insufficient, the Gaussians may elongate along the viewing direction as training progresses (as shown below), thus leading to "artifacts that resemble light reflections around the surrounding areas". Here are three solutions that could work.
|
@ddenglina Hi, could you show us the mesh visualization results so we can analyze them? |
@ddenglina Hi, you can find the code here: https://github.com/NVlabs/InstantSplat/tree/instantsplat-2dgs. You can run the script |
Hello, thank you for your work. I would like to try integrating 2D GS into the method you used for 3D GS. However, after the integration, the following issue arises: surface reconstruction can be completed, but there are artifacts resembling light reflections around the surroundings, as shown in the image below. Moreover, this phenomenon intensifies with the increase in the number of iterations. At 500 iterations, it is almost non-existent; at 1000 iterations, a few artifacts start to appear. The following image shows the situation in 3000 iterations.
Could you analyze what might be causing this?
The text was updated successfully, but these errors were encountered: