Update README.md

README.md

@@ -1,3 +1,158 @@
 ---
-license: unknown
+license: other
+license_name: krea-2-community-license
+license_link: https://huggingface.co/krea/Krea-2-Turbo/blob/main/LICENSE.pdf
+base_model:
+- krea/Krea-2-Turbo
+base_model_relation: quantized
+pipeline_tag: text-to-image
+library_name: gguf
+tags:
+- gguf
+- comfyui
+- krea-2
+- krea2
+- text-to-image
+- quantized
+- lora-merge
+- rebelrealai
+language:
+- en
 ---
+
+
+# Krea-R-Turbo · GGUF
+
+GGUF quants of **Krea 2 Turbo** with two RealRebelAI style LoRAs baked directly into
+the transformer, so you get the blended look in a single file with nothing to stack at
+runtime. Built to run on consumer GPUs (8 GB VRAM and up) through the ComfyUI-GGUF
+loader.
+
+This is a **derivative** of [`krea/Krea-2-Turbo`](https://huggingface.co/krea/Krea-2-Turbo):
+LoRA-merged, then quantized. It is not the original checkpoint.
+
+> [!IMPORTANT]
+> **Stock ComfyUI-GGUF will not load these.** Krea 2 uses a custom architecture
+> (`krea2`) that isn't in city96's upstream loader. You must use the krea2-aware fork:
+> **[RealRebelAI/ComfyUI-GGUF_KREA-2](https://github.com/RealRebelAI/ComfyUI-GGUF_KREA-2)**.
+> Loading with any other GGUF node will fail with `unknown model architecture: 'krea2'`.
+
+---
+
+## What's baked in
+
+The base transformer is Krea 2 Turbo. Merged on top, in weight space (not runtime patches):
+
+| LoRA | Strength | What it does |
+|------|----------|--------------|
+| **RebelReal** | 0.31 | Photoreal grounding — skin, light, material realism |
+| **RebelMidjourney** | 0.13 | A light MJ-style aesthetic lift on composition and color |
+
+Both were trained on **Krea 2 Raw** (per Krea's "train on Raw, run on Turbo" guidance)
+and merged into the Turbo checkpoint before quantization. Strengths are fixed in the
+weights — no LoRA loaders needed in your graph.
+
+---
+
+## Quants
+
+Pick by how much VRAM you can spare for the diffusion model (leave room for the text
+encoder + VAE). Sizes are approximate — see the **Files** tab for exact numbers.
+
+| File | Type | ~Size | Notes |
+|------|------|-------|-------|
+| `Krea-R-Turbo-Q8_0.gguf`   | Q8_0   | ~14 GB  | Near-lossless. For 16 GB+ cards or offload. |
+| `Krea-R-Turbo-Q6_K.gguf`   | Q6_K   | ~11 GB  | Excellent quality, hard to tell from Q8. |
+| `Krea-R-Turbo-Q5_K_M.gguf` | Q5_K_M | ~9.4 GB | Great quality/size balance. |
+| `Krea-R-Turbo-Q4_K_M.gguf` | Q4_K_M | ~8.0 GB | **Recommended all-rounder.** 12 GB cards, or 8 GB with offload. |
+| `Krea-R-Turbo-Q3_K_M.gguf` | Q3_K_M | ~6.4 GB | **Best pick for 8 GB VRAM.** Small quality trade. |
+| `Krea-R-Turbo-Q2_K.gguf`   | Q2_K   | ~5.5 GB | Tightest fit. Visible degradation but usable. |
+
+On an 8 GB card (e.g. RTX 3070), Q3_K_M or Q2_K fit most comfortably alongside the VAE
+and text encoder. Larger quants still run via ComfyUI's `--lowvram` / block-swap
+offload — just slower.
+
+---
+
+## Usage in ComfyUI
+
+**1. Install the loader node** (one time):
+```
+cd ComfyUI/custom_nodes
+git clone https://github.com/RealRebelAI/ComfyUI-GGUF_KREA-2
+```
+Restart ComfyUI.
+
+**2. Place the files:**
+
+| File | Folder |
+|------|--------|
+| `Krea-R-Turbo-QX.gguf` (this repo) | `ComfyUI/models/unet/` (or `diffusion_models/`) |
+| `qwen3vl_4b` text encoder | `ComfyUI/models/text_encoders/` |
+| `qwen_image_vae.safetensors` | `ComfyUI/models/vae/` |
+
+The text encoder (Qwen3-VL 4B) and VAE (Qwen-image VAE) are the standard Krea 2 companions
+and can be pulled from the ComfyUI repack:
+[`Comfy-Org/Krea-2`](https://huggingface.co/Comfy-Org/Krea-2).
+
+**3. Wire it up:** load the GGUF with the **Unet Loader (GGUF)** node from the fork above,
+feed it the Qwen3-VL text encoder and Qwen-image VAE, and sample as normal Krea 2 Turbo.
+
+### Recommended sampling
+
+Krea 2 Turbo is a distilled, few-step model:
+
+- **Steps:** ~8
+- **Guidance:** CFG-free — set CFG to `1.0` (no negative prompt), or guidance `0`
+- **Sampler / scheduler:** `er_sde` + `simple` works well (or `euler` + `simple`)
+- **Resolution:** up to 2048², 1024–1280 sweet spot on 8 GB
+
+No positive/negative CFG split is needed; this checkpoint was distilled without it.
+
+---
+
+## Requirements
+
+- ComfyUI (recent build with Krea 2 support)
+- [RealRebelAI/ComfyUI-GGUF_KREA-2](https://github.com/RealRebelAI/ComfyUI-GGUF_KREA-2) custom node
+- Qwen3-VL 4B text encoder + Qwen-image VAE (from [`Comfy-Org/Krea-2`](https://huggingface.co/Comfy-Org/Krea-2))
+
+This repo contains the **diffusion model (transformer) only** — it will not generate
+anything on its own without the VAE and text encoder above.
+
+---
+
+## License & attribution
+
+These weights are a derivative of **Krea 2 Turbo** © Krea.ai, Inc., and remain governed
+by the **Krea 2 Community License Agreement** and the **Krea Acceptable Use Policy** —
+exactly as the original model. See the license:
+[LICENSE.pdf](https://huggingface.co/krea/Krea-2-Turbo/blob/main/LICENSE.pdf).
+In the event of any conflict, the Krea Acceptable Use Policy and Krea 2 Community License
+control.
+
+**Deployer obligation:** if you serve or redistribute this model, the license requires
+you to implement content-filtering or equivalent review to prevent generation or
+distribution of unlawful or policy-violating content. Report harmful or illegal outputs
+to `safety@krea.ai`.
+
+Upstream:
+- Base model: [`krea/Krea-2-Turbo`](https://huggingface.co/krea/Krea-2-Turbo) (distilled from [`krea/Krea-2-Raw`](https://huggingface.co/krea/Krea-2-Raw))
+- ComfyUI repack (VAE / text encoder): [`Comfy-Org/Krea-2`](https://huggingface.co/Comfy-Org/Krea-2)
+- GGUF tooling based on city96's [ComfyUI-GGUF](https://github.com/city96/ComfyUI-GGUF)
+
+The **RebelReal** and **RebelMidjourney** LoRAs are original work by RealRebelAI, trained
+on Krea 2 Raw. The merge and quantization are also RealRebelAI's work; the underlying
+model rights belong to Krea.ai, Inc.
+
+---
+
+## Credits
+
+Merged, quantized, and released by **RealRebelAI**.
+
+- GitHub: [github.com/RealRebelAI](https://github.com/RealRebelAI)
+- Hugging Face: [huggingface.co/realrebelai](https://huggingface.co/realrebelai)
+
+If these are useful, a like on the repo helps others find them. Issues with loading go on
+the [ComfyUI-GGUF_KREA-2](https://github.com/RealRebelAI/ComfyUI-GGUF_KREA-2) tracker.