ComfyUI Node
HunyuanVideo 1.5 Model Loader
A ComfyUI node in HunyuanVideoWrapper1.5 with 20 inputs and 1 output.
HunyuanVideo 1.5 Model Loader
- hunyuanvideo_model_config
◄model_path►
◄attn_modeflash►
◄byt5_max_length256►
◄vision_encoder_typesiglip►
◄vision_encoder_precisionfp16►
◄text_encoder_typeqwen-2.5vl-7b►
◄text_encoder_tokenizer_typeqwen-2.5vl-7b►
◄text_encoder_max_length1000►
◄text_encoder_precisionfp16►
◄text_encoder_hidden_state_skip_layer2►
◄text_encoder_apply_final_normfalse►
◄text_encoder_reproducefalse►
◄resolution480p►
◄taskt2v►
◄enable_offloadingfalse►
◄enable_group_offloadingfalse►
◄transformer_dtypebfloat16►
◄device_optcuda►
◄hf_token►
◄flow_shift—►
CategoryHunyuanVideoWrapper1.5
Inputs (20)
| Name | Type | Default | Description |
|---|---|---|---|
| model_pathopt | STRING | 1. If you have already manually downloaded the model to your directory, specify the model folder path hyvideo-1.5 here. 2. If not specified, it will be automatically downloaded to the default folder for continued use. | |
| attn_modeopt | COMBO | flash | 3 options: flash, ptm_sparse_attn, flash3 |
| byt5_max_lengthopt | INT | 256 | — |
| vision_encoder_typeopt | STRING | siglip | — |
| vision_encoder_precisionopt | STRING | fp16 | — |
| text_encoder_typeopt | STRING | qwen-2.5vl-7b | — |
| text_encoder_tokenizer_typeopt | STRING | qwen-2.5vl-7b | — |
| text_encoder_max_lengthopt | INT | 1000 | — |
| text_encoder_precisionopt | STRING | fp16 | — |
| text_encoder_hidden_state_skip_layeropt | INT | 2 | — |
| text_encoder_apply_final_normopt | BOOLEAN | false | — |
| text_encoder_reproduceopt | BOOLEAN | false | — |
| resolutionopt | COMBO | 480p | 2 options: 480p, 720p |
| taskopt | COMBO | t2v | 2 options: t2v, i2v |
| enable_offloadingopt | BOOLEAN | false | — |
| enable_group_offloadingopt | BOOLEAN | false | — |
| transformer_dtypeopt | COMBO | bfloat16 | 9 options: float32, float64, float16, bfloat16, uint8, int8, +3 |
| device_optopt | COMBO | cuda | 2 options: cpu, cuda |
| hf_tokenopt | STRING | — | |
| flow_shiftopt | FLOAT | When the resolution is 480p, the recommended shift value is 5, and when the resolution is 720p, the recommended shift value is 7. If you do not set this value, it will be automatically configured according to the recommendations of this rule. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| hunyuanvideo_model_config | HUNYUANVIDEO_MODEL_CONFIG | — |