ComfyUI Node
FlashVSR
A ComfyUI node in Practical-Tools/SResolution with 11 inputs and 1 output.
FlashVSR
- frames
- image
◄modelFlashVSR-v1.1►
◄modetiny►
◄scale2►
◄color_fixtrue►
◄tiled_vaetrue►
◄tiled_dittrue►
◄unload_ditfalse►
◄tile_size256►
◄tile_overlap24►
◄seed0►
CategoryPractical-Tools/SResolution
Inputs (11)
| Name | Type | Default | Description |
|---|---|---|---|
| frames | IMAGE | Sequential video frames as IMAGE tensor batch | |
| model | COMBO | FlashVSR-v1.1 | Model version. |
| mode | COMBO | tiny | Using "tiny-long" mode can significantly reduce VRAM used with long video input. |
| scale | INT | 21–9223372036854776000 | — |
| color_fix | BOOLEAN | true | Use color fix algorithm to preserve original colors. |
| tiled_vae | BOOLEAN | true | Disable tiling: faster decode but higher VRAM usage. |
| tiled_dit | BOOLEAN | true | Significantly reduces VRAM usage at the cost of speed. |
| unload_dit | BOOLEAN | false | Unload DiT before decoding to reduce VRAM peak at the cost of speed. |
| tile_size | INT | 25632–1024 | DiT tile size; 0 disables tiling. |
| tile_overlap | INT | 248–512 | — |
| seed | INT | 00–1125899906842624 | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | — |