ComfyUI Node
Donut Krea2 Image Conditioning
A ComfyUI node in Donut/conditioning with 12 inputs and 1 output.
Donut Krea2 Image Conditioning
- clip
- image1
- image2
- image3
- vae
- conditioning
◄prompt—►
◄text_strength1.00►
◄rebalancetrue►
◄strength11.00►
◄strength21.00►
◄strength31.00►
◄multiplier1.00►
CategoryDonut/conditioning
Inputs (12)
| Name | Type | Default | Description |
|---|---|---|---|
| clip | CLIP | Krea2 CLIP model (Qwen3-VL 4B) | |
| prompt | STRING | Text instruction describing the edit or generation | |
| text_strength | FLOAT | 1.000–10 | Text influence: 0=no text baseline, 1=full, >1=overemphasis |
| rebalance | BOOLEAN | true | Reboost blended conditioning to match the requested text_strength magnitude. When lowering image strength, text gets proportionally louder so overall CFG stays constant. |
| image1opt | IMAGE | First reference image | |
| strength1opt | FLOAT | 1.000–10 | Image1 influence: 0=text-only, 1=full, >1=overemphasis |
| image2opt | IMAGE | Second reference image | |
| strength2opt | FLOAT | 1.000–10 | — |
| image3opt | IMAGE | Third reference image | |
| strength3opt | FLOAT | 1.000–10 | — |
| vaeopt | VAE | Optional VAE. Stores reference_latents on the conditioning for forward-compatibility. | |
| multiplieropt | FLOAT | 1.00-1000–1000 | Global gain on the conditioning tensor. 1.0 = unchanged. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| conditioning | CONDITIONING | — |