ComfyUI Node
CLIP Text Encode (Advanced)
Advanced text encoding with support for multi-layer tap rebalancing (Krea2) and global multiplier controls.
CLIP Text Encode (Advanced)
- clip
- conditioning
◄text—►
◄rebalance_presetnone►
◄per_layer_weights1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 2.5, 5.0, 1.1, 4.0, 1.0►
◄multiplier1.00►
◄renormalizetrue►
◄krea2_only_multiplierfalse►
Category🌒 Smart Model Loader/ Conditioning
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| clip | CLIP | The input CLIP model used to tokenize and encode the prompt. | |
| text | STRING | The text prompt to be converted into conditioning embeddings. | |
| rebalance_preset | COMBO | none | Select layer-gain rebalancing profile for Krea2 multi-layer tap models. 'none' disables rebalancing; 'balanced' boosts late taps (up to 5x); 'detail' dampens early and aggressively boosts late taps (up to 6x); 'subtle' gently boosts late taps (up to 2x); 'uniform' applies a flat 1.0; 'custom' uses 'per_layer_weights'. |
| per_layer_weights | STRING | 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 2.5, 5.0, 1.1, 4.0, 1.0 | Comma-separated float gains (e.g. 12 values for Krea2) applied to individual layer outputs. Active only when preset = 'custom'. |
| multiplier | FLOAT | 1.00-1000–1000 | Global strength multiplier applied to the conditioning embedding tensor. Values > 1.0 amplify prompt influence; < 1.0 reduce it. |
| renormalize | BOOLEAN | true | When enabled, rescales the altered conditioning tensor's RMS to match the original, keeping the overall prompt strength/volume constant after layer weighting. |
| krea2_only_multiplier | BOOLEAN | false | When enabled, the global multiplier is only applied to Krea2 models. For standard models, the multiplier is ignored (treated as 1.0). |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| conditioning | CONDITIONING | A conditioning containing the embedded text used to guide the diffusion model. |