Can this project run on the Windows ?

[EndrrSB]

Can this project run on the Windows ?

From the code, I can see that this project was developed in a Linux environment. Has it been tested for run on the Windows?

[yunhua-deng]

I tested in my Windows with an NVidia RTX4000 GPU, and it worked.

[GrebenkovaO]

Hi @EndrrSB, thanks for your interest in the project. We haven’t tested on Windows ourselves, but there’s nothing in the code that should prevent it from running there. If you run into any issues, please let us know!

A special thanks to @yunhua-deng for verifying it works on Windows with an NVIDIA RTX 4000 GPU.

[TamandTat]

我在 Windows 上使用 NVidia RTX4000 GPU 进行了测试，并且成功了。

I tested it on NVidia RTX4090 GPU, and the result was not good, not even as good as 3DGS. My input was sparse point cloud information after colmap. How is your running result?

[dimakot55]

Could you share some of your results? Then we might be able to help you. If possible please also share your results after the dense keypoints initialization as well.

[TamandTat]

您能分享一些您的结果吗？这样我们或许能帮到您。如果可以的话，也请分享一下密集关键点初始化后的结果。

I'm really sorry, I just tried EDGS at that time and didn't save the result. But my whole process is: First, use COLMAP to process the image data to obtain the sparse point cloud data in COLMAP format. Second, execute the command: "python train.py train.gs_epochs=30000 train.no_densify=True gs.dataset.source_path="xxx\xxx\xxx" gs.dataset.model_path="xxx\xxx\xxx" init_wC.matches_per_ref=20000 init_wC.nns_per_ref=3 init_wC.num_refs=180 wandb.mode=disabled". Finally, get the result. I think there is nothing wrong with my running process.

[dimakot55]

It is quite hard to tell what is the reason for that. The problem may lie in different things: COLMAP, in type of images you use (e.g. if you have medical images or aerial shots RoMa correspondences can be way too off), in scaling of your scene.