ComfyUI Node
TS H3 Fast VAE Decode
NVIDIA StreamRender-H3 noncausal 24-layer decoder, FP32 or native INT8 convrot weights, PyTorch SDPA path. Connect a standard H3 video VAE for automatic memory fallback.
TS H3 Fast VAE Decode
- latent
- fallback_vae
- images
- report
◄vae_name▾►
CategoryTS/Video
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| latent | LATENT | — | |
| vae_name | COMBO | 1 options: (missing fast_h3_vae_noncausal_24l) | |
| fallback_vaeopt | VAE | Standard H3 video VAE; used if fast decoding needs more VRAM than is available. Audio is ignored. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| images | IMAGE | — |
| report | STRING | — |