Nodes/HunyuanVideo-1.5 nodes/HunyuanVideo Transformer
ComfyUI Node

HunyuanVideo Transformer

A ComfyUI node in HunyuanVideoWrapper1.5 with 21 inputs and 1 output.

By yuanyuan-spec·Created 9 months ago·Updated 8 months ago· 27
HunyuanVideo Transformer
  • hyvid_cfg
  • transformer
  • vae_concat
  • hyvid_embeds
  • vision_states
  • extra_kwargs
  • latents_dict
  • transformer_latent
n_tokens1
steps50
target_dtypebfloat16
embedded_guidance_scale
guidance_rescale0.00
autocast_enabledtrue
eta0.00
enable_cachefalse
cache_start_step11
cache_end_step45
no_cache_block_id53
cache_step_interval4
cache_typedeepcache
enable_sageattnfalse
CategoryHunyuanVideoWrapper1.5

Inputs (21)

NameTypeDefaultDescription
hyvid_cfgHYVID15CFG
n_tokensINT11–1000
stepsINT501–1000
transformerHYVID15TRANSFORMER
vae_concatHYVID15VAECONCAT
hyvid_embedsHYVIDEMBEDS
vision_statesHYVID15VISIONSTATES
extra_kwargsHYVID15EXTRAKWARGS
latents_dictHYVID15LATENTSDICT
target_dtypeCOMBObfloat169 options: float32, float64, float16, bfloat16, uint8, int8, +3
embedded_guidance_scaleoptFLOATAdditional control guidance scale, if supported
guidance_rescaleoptFLOAT0.00
autocast_enabledoptBOOLEANtrue
etaoptFLOAT0.000–1
enable_cacheoptBOOLEANfalseEnable cache.
cache_start_stepoptINT11Cache start step.
cache_end_stepoptINT45Cache end step.
no_cache_block_idoptINT53No cache block id.
cache_step_intervaloptINT4Cache step interval.
cache_typeoptCOMBOdeepcacheCache type.
enable_sageattnoptBOOLEANfalseEnable sageattn.

Outputs (1)

NameTypeDescription
transformer_latentHYVID15TRANSFORMERLATENT