ComfyUI Node
FL PenguinVL Video Query
A ComfyUI node in FL PenguinVL with 10 inputs and 1 output.
FL PenguinVL Video Query
- model
- images
- text
◄promptDescribe what happens in this video.►
◄system_prompt►
◄max_frames16►
◄max_new_tokens512►
◄temperature0.70►
◄top_p0.90►
◄do_sampletrue►
◄max_long_side512►
CategoryFL PenguinVL
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| model | FL_PENGUINVL_MODEL | — | |
| images | IMAGE | — | |
| prompt | STRING | Describe what happens in this video. | — |
| system_promptopt | STRING | — | |
| max_framesopt | INT | 161–128 | Max frames to sample from the batch. More frames = more VRAM. Model supports up to 128. |
| max_new_tokensopt | INT | 5121–4096 | — |
| temperatureopt | FLOAT | 0.700–2 | — |
| top_popt | FLOAT | 0.900–1 | — |
| do_sampleopt | BOOLEAN | true | False = greedy decoding (deterministic) |
| max_long_sideopt | INT | 5120–2048 | Resize long side of each frame to this value. 0 = no resize. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| text | STRING | — |