File size: 4,259 Bytes
ca9cb6a fec1489 ca9cb6a fec1489 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 | ---
license: apache-2.0
tags:
- lora
- krea2
- character
- comfyui
- text-to-image
base_model: KREA2
---
# Krea2 Character LoRA Collection
A collection of **34 character LoRAs** for the **KREA2** base model, each paired with a `persona .md` file describing the character's stable facial features, variable appearance, and expressions.
Every LoRA ships as `<Name>_KREA2.safetensors` alongside `<Name>_KREA2.md`.
## Fully automated pipeline
This entire repository — dataset, training, and upload — was produced end-to-end by an autonomous **[Hermes Agent](https://hermes-agent.nousresearch.com)** (served by the **Kimi K3** model) from a natural-language request. No manual review gates anywhere in the loop:
1. **Scraping** — automated image collection per character from web photo pools.
2. **Cropping & curation** — face/head detection **Ultralytics Yolo** and crop, with a hard content filter (No AI generated images, correct subject only, sharpness threshold).
3. **Aesthetic gating** — every candidate image scored by a discriminator built on the **ERNIE aesthetic image scorer**; low scorers rejected automatically with a hard pass/fail threshold.
4. **Captioning** — datasets captioned with **Gemma-4 26B (abliterated) vision MTP** for unrestricted, descriptive captions, fast (1.2s/i).
5. **Persona synthesis** — each `<Name>_KREA2.md` generated from caption trait-analysis (facial features, range of expressions).
6. **Infrastructure** — training pods created, configured, monitored, and torn down automatically via the **RunPod API (MCP)**.
7. **Training** — LoRAs trained on pods with **Ostris AI-Toolkit** with user predefined settings.
8. **Publishing** — weights and persona files uploaded to this repo by the agent.
## Usage
### Recommended: CRT-Nodes persona loader
Use the **Seeded Persona LoRA Loader (CRT)** node from [CRT-Nodes](https://github.com/PGCRT/CRT-Nodes). It loads both the LoRA **and** its paired `.md` file in one node, and returns:
- `MODEL` — the LoRA-patched model
- `STRING` — the persona text, ready to feed an LLM prompt enhancer
Wiring the string output into your prompt-enhancement LLM gives it grounded hints about the currently loaded character — stable features, typical styling, expressions — so prompts stay on-identity without you memorizing trigger details.
The same CRT-Nodes pack includes **ERNIE Image Aesthetic Score (CRT)** — the exact scorer used to gate these datasets, packaged as a ComfyUI node with automatic model download. Ready to use out of the box if you want to score your own outputs or curate your own datasets with the same standard.
### Standard usage
Load `<Name>_KREA2.safetensors` as a regular LoRA on top of KREA2 in ComfyUI (or any KREA2-compatible pipeline) — recommended strength is around **1.0**. The `.md` files are plain text — read them, or feed them to your prompt LLM manually.
## Contents
| | | |
|---|---|---|
| Adriana Lima | Alexandria Ocasio-Cortez | Ana De Armas |
| Anne Hathaway | Anya Taylor-Joy | Ariana Grande |
| Autumn Falls | Bella Poarch | Billie Eilish |
| Camille Cerf | Doja Cat | Dorozea |
| Dua Lipa | Elydia | Emilia Clarke |
| Eva Mendes | Hailey Rose | Hannah Kleit |
| Ho-yeon | Ice Spice | Inde Navarrette |
| Jessica Alba | Kaia Gerber | Kylie Jenner |
| Lea Elui | Nicki Minaj | Sab Zada |
| Shakira | Shay | Sydney Sweeney |
| Tinashe | Tokyo | Zendaya |
| Zoe Kravitz | | |
## Known limitations
- Trained for likeness; style flexibility depends on your prompt and sampler settings.
## Disclaimer
These LoRAs reproduce the likeness of real people and are shared for research and creative purposes. Please use them responsibly and respect the rights of the individuals depicted — don't create misleading, harmful, or non-consensual content. By downloading, you take responsibility for what you generate.
## Credits
- Pipeline automation: [Hermes Agent](https://hermes-agent.nousresearch.com) (Nous Research) on Kimi K3
- Training: [Ostris AI-Toolkit](https://github.com/ostris/ai-toolkit)
- Aesthetic scoring: ERNIE aesthetic scorer (CRT-Nodes ComfyUI packaging)
- Captioning: Gemma-4 26B (abliterated) vision MTP
- Infrastructure: RunPod
- User Discord https://discord.gg/uV6ZNgg33j
|