ComfyUI Node
Qwen3-VL Prompt Optimization Assistant
A ComfyUI node in Henjicc/Text with 9 inputs and 3 outputs.
Qwen3-VL Prompt Optimization Assistant
- media
- clip
- text
- visuals
- media
◄taskauto►
◄frames_per_second2►
◄segment_seconds10►
◄analysis_seconds0►
◄max_tokens512►
◄pixel_budget_mp16►
◄custom_instruction►
CategoryHenjicc/Text
Inputs (9)
| Name | Type | Default | Description |
|---|---|---|---|
| media | HENJICC_MEDIA | — | |
| clip | CLIP | 通过原生 CLIP 加载器选择完整 Qwen3-VL；可共享千问 2.1 编码器。H3 专用 50 层编码器不能直接生成文本。 / Select a full Qwen3-VL with Load CLIP; Qwen-Image 2.1 can share it. | |
| task | COMBO | auto | 7 options: auto, describe, qwen_auto, h3_swap, h3_reference, general, +1 |
| frames_per_second | COMBO | 2 | 2 options: 2, 4 |
| segment_seconds | INT | 105–10 | — |
| analysis_seconds | INT | 00–60 | — |
| max_tokens | INT | 51264–2048 | — |
| pixel_budget_mpopt | FLOAT | 164–24 | — |
| custom_instructionopt | STRING | — |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| text | STRING | — |
| visuals | IMAGE | — |
| media | HENJICC_MEDIA | — |