ShowMeCode

i am running the code on a remote terminal, so i need off-screen rendering

Hi, now i am using osmesa to render a 82 frames video with 960x720 size, but it takes more than 500s； in order to reduce the time, i am going...

@rfdeng Hi, i also find that osmesa backend is too slow, but my egl backend always get fllowing error. I am running on a remote terminal, do you know the...

Otherwise, My training loss is difficult to reduce, always near 0.88

@idunnostuff i also have the same issue on my linux machine, Did you solve this issue ?

I find a solution to solve this problem, i am running on a remote terminal, that need virtual display to show the rendering window. like this link: https://www.jianshu.com/p/7df287155ce0

hi, I also have this problem, does anybody know how to run it on th CPU

@Huang-Jin have you get the answer?

Hi, > Hi,I use these commands to train the dreamgaussian: > > ### training gaussian stage > python main.py --config configs/text.yaml prompt="a photo of an icecream" save_path=icecream > > ###...

it is useful to use rendered_alpha to calculate mask loss in some object reconstruction cases. rendered_depth can also provide some guidance e.g. depth loss between rendered_depth and monocular depth.