🔡 CLIP Text Encode Simple
Positive-only encoding with translate, styles, and dynamic prompts
- clip
- positive
- prompt
This is a CLIP text encoder with three quality-of-life features baked in, stripped down to just the positive prompt. You write your prompt, and before it gets encoded the node can translate it to English, apply a curated style preset, and expand any dynamic-prompt syntax. Out comes the CONDITIONING for your sampler - plus, usefully, the final resolved prompt string so you can see exactly what got encoded. It's the "simple" (positive-only) member of the pack's encoder family; the fuller SDVN encoder does positive and negative together.
Why the extras matter: most diffusion models were trained overwhelmingly on English captions, so if you think in another language, translating first genuinely improves adherence. And the style library (219 presets, from the pack's styles.csv / my_styles.csv) lets you slap on a look with one dropdown instead of memorizing a wall of quality tags.
How it works
Your positive text runs through a small pipeline before encoding: translate (if set) → apply the chosen style card → expand dynamic prompts → encode with the clip you provide. The style cards wrap your prompt in pre-written positive (and sometimes negative-flavored) text. The dynamic-prompt step is what turns {a|b|c} or wildcards into a concrete pick, and the seed is what decides which pick - same seed, same expansion.
The inputs and outputs that matter
positive(multiline STRING) - your prompt.clip(CLIP) - from your checkpoint loader; this is what does the actual encoding.style- one of 219 presets, orNone.translate(language picker, defaultNone) - translate to English before encoding.seed- drives dynamic-prompt rerolls.
Outputs: positive (CONDITIONING) into your KSampler, and prompt (STRING) - the fully resolved text after translate/style/dynamic, ideal for logging or piping into an Any show so you can see what the model actually got.
How to install it
Comes with the SDVN pack. ComfyUI Manager, search SDVN_Comfy_node; or:
cd ComfyUI/custom_nodes
git clone https://github.com/StableDiffusionVN/SDVN_Comfy_node
pip install -r custom_nodes/SDVN_Comfy_node/requirements.txt
from the ComfyUI root, then restart.
Common issues & troubleshooting
Dynamic prompts don't expand. The pack's dynamic-prompt support needs the separate ComfyUI-DynamicPrompts node installed, with wildcard files living in /ComfyUI/wildcards. Without it, the {a|b|c} and __wildcard__ syntax passes through as literal text.
Your custom styles aren't there. The built-in style cards work out of the box, but custom ones come from my_styles.csv - copy the provided my_styles.csv.example to my_styles.csv and add your rows, per the README.
It's not translating. The translate picker defaults to None, which passes text through untouched. Set it to your source language to actually translate to English.
Same dynamic result every run. That's the seed - a fixed seed reproduces the same expansion. Randomize it for variety.
Where's my negative? This node is positive-only by design. Encode your negative with a separate encoder and wire both into the sampler.
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| positive | STRING | Prompt tích cực mô tả nội dung bạn muốn sinh ra. | |
| style | COMBO | None | Chọn style mẫu có sẵn để thêm vào prompt. |
| translate | COMBO | Ngôn ngữ dịch prompt. | |
| seed | INT | 00–18446744073709550000 | Seed ngẫu nhiên cho prompt. |
| clip | CLIP | Mô hình CLIP dùng để mã hóa prompt. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| positive | CONDITIONING | Điều kiện chứa văn bản đã mã hóa để hướng dẫn mô hình sinh ảnh. |
| prompt | STRING | — |