gaussian-splatting icon indicating copy to clipboard operation
gaussian-splatting copied to clipboard

Real time renderer

Open jh27kim opened this issue 1 year ago • 11 comments
trafficstars

Hi, When I used the SIBR real time viewer, it gives me this warning. image

Also when I switch rendering type to Ellipsoids, the whole system seems to stop and I cannot move around the scene. Is this also related to the former warning ?

The command I used is ./<SIBR install dir>/bin/SIBR_gaussianViewer_app -m <path to trained model> --device 0

PyTorch: 1.12.1
CUDA: 11.7 
OpenGL: 4.5 (Mesa 21.2.6)
Device: NVIDIA A6000 (48G)

jh27kim avatar Dec 20 '23 04:12 jh27kim

  1. try --device 1 on your system if it have more then one GPU. Setting default GPU can help.
  2. Check what CUDA is dedicated to your GPU and reinstal it.

jaco001 avatar Dec 20 '23 06:12 jaco001

Hi, I tried --device 1 but it gives me the same warning. And I guess CUDA - GPU compatibility is fine since I have been using the same version on other conda envs and worked fine.

jh27kim avatar Dec 21 '23 06:12 jh27kim

Try search for another copy of viewier at (gs-catalog) \viewer\bin\SIBR_gaussianViewer_app.exe It is already compiled and can be different if you compiled your own.

jaco001 avatar Dec 21 '23 08:12 jaco001

Hi, I tried --device 1 but it gives me the same warning. And I guess CUDA - GPU compatibility is fine since I have been using the same version on other conda envs and worked fine. Hi, l met the same issue. May I ask what is your conda envs configuration that worked fine for Ellipsoids Rendering? By the way, are you using Ubuntu 22 or Ubuntu 20?

freemantom avatar Dec 22 '23 09:12 freemantom

hi @jh27kim, any news on this? I am using Ubuntu 22 and met the same CUDA/OpenGL interop issue.

Does it happen on Windows as well?

FedericoVasile1 avatar Feb 07 '24 13:02 FedericoVasile1

Do you have a system with more than GPU? Is the non-NVIDIA GPU the one that is connected to the monitor?

Snosixtyboo avatar Feb 09 '24 23:02 Snosixtyboo

I am running it on a laptop equipped with Nvidia GTX 1650 Ti. This is the output of the command:

user@3dd732b690b8:/workspace/repos/gaussian-splatting/SIBR_viewers$ ./install/bin/SIBR_gaussianViewer_app -m /workspace/host-area/e7b3eb6c-3
[SIBR] --  INFOS  --:	Initialization of GLFW
MESA: error: Failed to query drm device.
libGL error: glx: failed to create dri3 screen
libGL error: failed to load driver: iris
libGL error: failed to open /dev/dri/card0: No such file or directory
libGL error: failed to load driver: iris
[SIBR] --  INFOS  --:	OpenGL Version: 4.5 (Compatibility Profile) Mesa 23.0.4-0ubuntu1~22.04.1[major: 4, minor: 5]
[SIBR] ##  ERROR  ##:	FILE /workspace/repos/gaussian-splatting/SIBR_viewers/src/core/scene/ParseData.cpp
			LINE 560, FUNC getParsedData
			Cannot determine type of dataset at //workspace/data/bicycle
[SIBR] --  INFOS  --:	Did not find specified input folder, loading from model path
Number of input Images to read: 194
Number of Cameras set up: 194
[SIBR] --  INFOS  --:	Mesh contains: colors: 1, normals: 1, texcoords: 0
[SIBR] --  INFOS  --:	Mesh '/workspace/host-area/e7b3eb6c-3/input.ply successfully loaded. 1 meshes were loaded with a total of  (0) faces and  (54275) vertices detected. Init GL ...
[SIBR] --  INFOS  --:	Init GL mesh complete 
[SIBR] --  INFOS  --:	Loading 5717126 Gaussian splats
[SIBR] --  INFOS  --:	Initializing Raycaster
[SIBR] --  INFOS  --:	Interactive camera using (0.009,1100) near/far planes.
Switched to trackball mode.

I think the interesting lines are the ones related to MESA and libGL errors

FedericoVasile1 avatar Feb 10 '24 00:02 FedericoVasile1

Ok thanks. so does it actually run at all, with the non-interop mode? Because GTX 1650 is definitely older than what we aim to support.

Snosixtyboo avatar Feb 10 '24 00:02 Snosixtyboo

Hi, When I used the SIBR real time viewer, it gives me this warning. image

Also when I switch rendering type to Ellipsoids, the whole system seems to stop and I cannot move around the scene. Is this also related to the former warning ?

The command I used is ./<SIBR install dir>/bin/SIBR_gaussianViewer_app -m <path to trained model> --device 0

PyTorch: 1.12.1
CUDA: 11.7 
OpenGL: 4.5 (Mesa 21.2.6)
Device: NVIDIA A6000 (48G)

Hi, I also met this problem, have you solved it?

fangzhou2000 avatar Mar 12 '24 13:03 fangzhou2000

Hi, I also met this problem, have you solved it?

xuyaojian123 avatar Jun 11 '24 12:06 xuyaojian123

You can try the following command:

sudo apt-get install nvidia-prime
sudo prime-select nvidia

SFNCY avatar Aug 02 '24 07:08 SFNCY