bad result for configs/diffusion/inference/768px.py on one GPU

[zhengdian1]

I got extremely bad results by directly using text-to-video generation in 768px, one GPU. What are the possible reasons?

[SimonWXW]

could you give me your command when you run? Is there any modifications in the inference code or config file?

[zhengdian1]

I have tried the commands below and the results are all terrible. I do not modify any code. The results of FLUX is correct. torchrun --nproc_per_node 1 --standalone scripts/diffusion/inference.py configs/diffusion/inference/256px.py --prompt "raining, sea"

torchrun --nproc_per_node 8 --standalone scripts/diffusion/inference.py configs/diffusion/inference/768px.py --prompt "raining, sea"

torchrun --nproc_per_node 1 --standalone scripts/diffusion/inference.py configs/diffusion/inference/t2i2v_256px.py --save-dir samples --prompt "raining, sea"

torchrun --nproc_per_node 1 --standalone scripts/diffusion/inference.py configs/diffusion/inference/t2i2v_768px.py --save-dir samples --prompt "raining, sea"

[SimonWXW]

Theoratically, config t2i2v_768px.py generates the best result. Could you upload your generated video or the screenshots? meanwhile you can try some other prompts to see if it is also bad

[zhengdian1]

Could you please re-clone the code and run it again according to the instructions in your README? I find it a bit puzzling.

[SimonWXW]

Could you please re-clone the code and run it again according to the instructions in your README? I find it a bit puzzling.

https://github.com/user-attachments/assets/e5b42be2-4e39-45b6-96cd-1784d31d0597 this is the video I reclone the code and generate using the default settings with t2i2v_256, is it the same quality you get?

[zhengdian1]

No, could you check the gmail? I send you my wechat.

[github-actions[bot]]

This issue is stale because it has been open for 7 days with no activity.

[github-actions[bot]]

This issue was closed because it has been inactive for 7 days since being marked as stale.