40 lines
2.5 KiB
Markdown
40 lines
2.5 KiB
Markdown
## TT SD3.5 Text Encoder *TT_Sd35TextEncoderNode*
|
|
|
|
*This description was generated with AI assistance. If you spot any errors, please create an issue on GitHub.*
|
|
|
|
Encodes text prompts for SD3.5 using all three text encoders (CLIP-L, CLIP-G, T5-XXL) and returns a configured Guider object ready for sampling.
|
|
|
|
### Inputs
|
|
|
|
| Parameter | Type | Required | Description |
|
|
|-------------------|--------|----------|-----------------------------------------------------------|
|
|
| `model` | MODEL | Yes | SD3.5 diffusion model |
|
|
| `clip` | CLIP | Yes | SD3.5 text encoders (CLIP-L, CLIP-G, T5-XXL) |
|
|
| `cfg` | FLOAT | No | Classifier-free guidance scale. Default: `5.0` |
|
|
| `clip_l_positive` | STRING | No | Positive prompt for CLIP-L encoder |
|
|
| `clip_g_positive` | STRING | No | Positive prompt for CLIP-G encoder |
|
|
| `t5xxl_positive` | STRING | No | Positive prompt for T5-XXL encoder |
|
|
| `clip_l_negative` | STRING | No | Negative prompt for CLIP-L encoder |
|
|
| `clip_g_negative` | STRING | No | Negative prompt for CLIP-G encoder |
|
|
| `t5xxl_negative` | STRING | No | Negative prompt for T5-XXL encoder |
|
|
| `lora_triggers` | STRING | No | LoRA trigger words appended to positive prompts |
|
|
| `width` | INT | No | Output width for resolution conditioning. Default: `512` |
|
|
| `height` | INT | No | Output height for resolution conditioning. Default: `512` |
|
|
| `target_width` | INT | No | Target width for resolution conditioning. Default: `512` |
|
|
| `target_height` | INT | No | Target height for resolution conditioning. Default: `512` |
|
|
|
|
### Outputs
|
|
|
|
| Parameter | Type | Description |
|
|
|-----------|--------|---------------------------------------------------------|
|
|
| `GUIDER` | GUIDER | Configured guider with encoded conditioning for sampler |
|
|
|
|
### Usage
|
|
|
|
Connect `MODEL` and `CLIP` from a SD3.5 models loader node. Connect `GUIDER` to a sampler node. Use different prompts per encoder for fine-grained
|
|
control — T5-XXL handles long descriptive prompts best, CLIP-L and CLIP-G work better with short style tags.
|
|
|
|
---
|
|
|
|
[View on GitHub](https://github.com/tenser-tensor/ComfyUI-TenserTensor)
|