Fail to generate 360 scene

[fangchuan]

Hi @eckertzhang , thanks for your wonderful work!

I have some problems when I follow your tutorial to generate a 360 scene, using my custom text prompt.

CUDA_VISIBLE_DEVICES=0 python text2nerf_main.py --config 'configs/text2nerf_scenes.txt' --expname 'text000_360' --prompt 'The bedroom has eight walls. The room has two windows and a bed ' --datadir 'data_example/text000' --pose_traj 'circle' --regen_pose

After 600 epochs, it seems still cannot get the complete neural radiance field of the 3d room, here is the final rendered rgb video and depth video

https://github.com/eckertzhang/Text2NeRF/assets/8519718/266020c7-5089-4243-a6d3-66947a74fd8c

https://github.com/eckertzhang/Text2NeRF/assets/8519718/8f40d864-e6a2-4b83-90b6-ce9d1d9daaf1

Could you help me figure out what's wrong with my cmd or configurations in code? Many thanks!

[lxy198]

Has the issue been resolved? I'm encountering the same problem

[lxy198]

Do you think we should increase the number of epochs ？ Many thanks!

[Maroceannn]

I got the same error. The training process was correct, but there were some issues in the "renderer". Maybe you can set the parameter "is_train" as True

[lxy198]

I got the same error. The training process was correct, but there were some issues in the "renderer". Maybe you can set the parameter "is_train" as True

我遇到了同样的错误。训练过程是正确的，但是“渲染器”中出现了一些问题。 也许你可以将参数“is_train”设置为True

Has the issue been resolved?

[jeremy123z]

Has the issue been resolved? I'm encountering the same problem

i even cannt work the code successfully,there are some errors in my code project and idl how to download the pre-trained ,could u please help me to solve.i wanna have a touch link with u! thxs soooo much