Pengfei Xuan

Ubuntu 18.04 has reached its end of life, making it challenging to install compatible CUDA and dependency versions. As you suggested, I removed the hard-coded OS version and also enabled...

Vulkan backend seems very promising: - https://blog.nomic.ai/posts/gpt4all-gpu-inference-with-vulkan - https://github.com/DTolm/VkFFT - https://github.com/shg8/3DGS.cpp/tree/main

Kompute looks really good. Right now, it's an incubating project in LF AI & Data landscape: https://landscape.lfai.foundation/card-mode?category=programming

PlayCavas's implementation seems working great: - https://github.com/playcanvas/engine/blob/release-1.69/src/scene/gsplat/gsplat.js#L259

[This](https://github.com/francescofugazzi/3dgsconverter/blob/37dc41e969db2c041d3a815ab0692552a80d236d/gsconverter/utils/base_converter.py#L97) method can give some help. But it’s a little computation heavy.

Good idea. Open3D also has a few similar built-in functions: https://www.open3d.org/docs/latest/tutorial/Advanced/pointcloud_outlier_removal.html

Perhaps removing floaters could be considered as a post-processing step after the 3DGS model gets fully trained. Additionally, I found enhancing the training code, improving camera pose, and removing blurry...

Maybe you can try [this](https://github.com/nerfstudio-project/nerfstudio/issues/2876#issuecomment-1988953975) and [this](https://github.com/nerfstudio-project/nerfstudio/issues/2876#issuecomment-2027407812) workarounds.

@parkerlreed Before launching docker, you may need to expose host GPUs to docker engine first. e.g. https://rocm.docs.amd.com/projects/install-on-linux/en/latest/how-to/docker.html#accessing-gpus-in-containers. ``` docker run --device /dev/kfd --device /dev/dri ``` To support ROCm 6.0.2, we...

It seems like you are using iGPU, which is not supported by ROCm very well. Maybe you can try `export HSA_OVERRIDE_GFX_VERSION=11.0.0` or with a different version and see if it...