lightx2v

/

Hy1.5-Quantized-Models

like 39

what version of pytorch is installed on your env? do u use sage attention? ur problem has nothing to do with the models

2.6，3.12，no Sage Attention, still black .hy15_480p_i2v_cfg_distilled_fp8_e4m3_lightx2v.safetensors

Got black output, but models from ComfyUI (https://huggingface.co/Comfy-Org/HunyuanVideo_1.5_repackaged/tree/main/split_files/diffusion_models) are working in the same workflow.

The models here are indeed not intended for ComfyUI use, they are for the lightx2v inference codebase: https://github.com/ModelTC/LightX2V

Like mentioned above, use the ones in comfy-org repo: https://huggingface.co/Comfy-Org/HunyuanVideo_1.5_repackaged/tree/main/split_files/diffusion_models

Same issue.

Same issue.

you're using wrong model
re-download https://huggingface.co/Comfy-Org/HunyuanVideo_1.5_repackaged/tree/main/split_files/diffusion_models