Skip to content

Transform your 3D texturing workflow with the power of generative AI, directly within Blender!

update readme and manual installation guide for qwen integration

sakalond22d33b920dd7
diff --git a/README.md b/README.md
index f6074eb..fd0575e 100644
--- a/README.md
+++ b/README.md
@@ -6,7 +6,7 @@
**Transform your 3D texturing workflow with the power of generative AI, directly within Blender!**
-StableGen is an open-source Blender plugin designed to seamlessly integrate advanced diffusion models (SDXL, FLUX.1-dev) into your creative process. Generate complex, coherent, and controllable textures for your 3D models and entire scenes using a flexible ComfyUI backend.
+StableGen is an open-source Blender plugin designed to seamlessly integrate advanced diffusion models (SDXL, FLUX.1-dev, Qwen Image Edit 2509) into your creative process. Generate complex, coherent, and controllable textures for your 3D models and entire scenes using a flexible ComfyUI backend.
---
@@ -53,7 +53,7 @@ StableGen empowers 3D artists by bringing cutting-edge AI texturing capabilities
* Employ IPAdapter without an reference image for enhanced consistency in multi-view generation modes.
* Control IPAdapter strength, weight type, and active steps.
* ⚙️ **Flexible ComfyUI Backend:**
- * Connects to your existing ComfyUI installation, allowing you to use your preferred SDXL checkpoints and custom LoRAs. Experimental support for FLUX.1-dev.
+ * Connects to your existing ComfyUI installation, allowing you to use your preferred SDXL checkpoints, custom LoRAs, and the new Qwen Image Edit workflow alongside experimental FLUX.1-dev support.
* Offloads heavy computation to the ComfyUI server, keeping Blender mostly responsive.
* ✨ **Advanced Inpainting & Refinement:**
* **Refine Mode (Img2Img):** Re-style, enhance, or add detail to existing textures (StableGen generated or otherwise) using an image-to-image process. Choose to preserve original textures for localized refinement.
@@ -136,6 +136,7 @@ StableGen acts as an intuitive interface within Blender that communicates with a
* **Blender:** Version 4.2 or newer.
* **Operating System:** Windows 10/11 or Linux.
* **GPU:** **NVIDIA GPU with CUDA is recommended** for ComfyUI. For further details, check ComfyUI's github page: [https://github.com/comfyanonymous/ComfyUI](https://github.com/comfyanonymous/ComfyUI).
+ * At least 8 GB of VRAM is required to run SDXL at a usable speed; plan for 16 GB or more when running FLUX.1-dev or the Qwen-Image-Edit pipeline.
* **ComfyUI:** A working installation of ComfyUI. StableGen uses this as its backend.
* **Python:** Version 3.x (usually comes with Blender, but Python 3 is needed for the `installer.py` script).
* **Git:** Required by the `installer.py` script.
@@ -185,11 +186,11 @@ The `installer.py` script (found in this repository) automates the download and
```
Replace `<YourComfyUIDirectory>` with the actual path. If omitted, the script will prompt for it.
3. **Follow On-Screen Instructions:**
- * The script will display a menu of installation packages (Minimal, Essential, Recommended, Complete SDXL). Choose one based on your needs.
+ * The script will display a menu of installation packages (Minimal, Essential, Recommended, Complete SDXL, plus Qwen-specific bundles). Choose the option that matches the feature set you want to install.
* It will download and place files into the correct subdirectories of `<YourComfyUIDirectory>`.
4. **Restart ComfyUI:** If ComfyUI was running, restart it to load new custom nodes.
-*(For manual dependency installation (and for installation of FLUX.1-dev and its dependecies), see `docs/MANUAL_INSTALLATION.md`.)*
+*(For manual dependency installation—including FLUX.1-dev and Qwen Image Edit setups—see `docs/MANUAL_INSTALLATION.md`.)*
### Step 3: Install StableGen Blender Plugin
@@ -225,7 +226,8 @@ Here’s how to get your first texture generated with StableGen:
* In the StableGen panel, click "**Add Cameras**". Choose `Object` as center type. Adjust interactively if needed, then confirm.
5. **Set Basic Parameters:**
* **Prompt:** Type a description (e.g., "ancient stone wall with moss").
- * **Checkpoint:** Select a checkpoint (e.g., `sdxl_base_1.0`).
+ * **Architecture:** Pick the diffusion family (`SDXL`, `Flux 1`, or `Qwen Image Edit`) that matches the workflow you set up.
+ * **Checkpoint:** Select a checkpoint or GGUF file suited to the chosen architecture (e.g., `sdxl_base_1.0` or `Qwen-Image-Edit-2509-Q3_K_M.gguf`).
* **Preset:** Choose a preset and apply it. `Default` or `Characters` are good starting points.
6. **Hit Generate!** Click the main "**Generate**" button.
7. **Observe:** Watch the progress in the panel and the ComfyUI console. Your object should update with the new texture! Output files will be in your specified "Output Directory".
@@ -260,7 +262,7 @@ These are your primary controls for defining the generation:
* **Prompt:** The main text description of the texture you want to generate.
* **Checkpoint:** Select the base SDXL checkpoint.
-* **Architecture:** Choose between `SDXL` and `Flux 1` (experimental) model architectures.
+* **Architecture:** Choose between `SDXL`, `Flux 1` (experimental), and `Qwen Image Edit` (experimental) model architectures.
* **Generation Mode:** Defines the core strategy for texturing:
* `Generate Separately`: Each viewpoint generates independently.
* `Generate Sequentially`: Viewpoints generate one by one, using inpainting from previous views for consistency.
@@ -333,7 +335,7 @@ Encountering issues? Here are some common fixes. Always check the **Blender Syst
* Check firewall settings.
* **Models Not Found (Error in ComfyUI Console):**
* Run the `installer.py` script.
- * Manually ensure models are in the correct subfolders of `<YourComfyUIDirectory>/models/` (e.g., `checkpoints/`, `controlnet/`, `loras/`, `ipadapter/`, `clip_vision/`).
+ * Manually ensure models are in the correct subfolders of `<YourComfyUIDirectory>/models/` (e.g., `checkpoints/`, `controlnet/`, `loras/`, `ipadapter/`, `clip_vision/`, `clip/`, `vae/`, `unet/`).
* Restart ComfyUI after adding new models or custom nodes.
* **GPU Out Of Memory (OOM):**
* Enable `Auto Rescale Resolution` in `Advanced Parameters` > `Output & Material Settings` if disabled.
@@ -406,7 +408,7 @@ Here are some features we plan to implement in the future (in no particular orde
* **Advanced IPAdapter support:** Support for custom IPAdapter models, support for advanced IPAdapter parameters.
* **Upscaling:** Support for upscaling generated textures.
* **Custom VAE, CLIP model selection:** Ability to select custom VAE and CLIP models in addition to custom ControlNet and LoRA models.
-* **Qwen-Image-Edit integration:** Qwen-Image-Edit show potential for generating higher quality textures with better prompt adherence. We plan to add support for it as an alternative to SDXL/Flux.1-dev.
+* **Qwen workflow refinements:** Deeper integration for Qwen Image Edit, including UV inpainting and refine modes.
* **Automatic camera placement improvements:** More advanced camera placement algorithms (e.g., based on model geometry).
* **Mesh generation:** Integration of mesh generation capabilities.
@@ -421,4 +423,4 @@ Ondřej Sakala
* X/Twitter: `@sakalond`
---
-*Last Updated: October 31, 2025*
+*Last Updated: November 3, 2025*
diff --git a/docs/MANUAL_INSTALLATION.md b/docs/MANUAL_INSTALLATION.md
index bc28644..6752797 100644
--- a/docs/MANUAL_INSTALLATION.md
+++ b/docs/MANUAL_INSTALLATION.md
@@ -194,3 +194,52 @@ FLUX.1 is a newer diffusion architecture and its usage in StableGen is optional.
* **Download URL:** [https://huggingface.co/InstantX/FLUX.1-dev-IP-Adapter/resolve/main/ip-adapter.bin?download=true](https://huggingface.co/InstantX/FLUX.1-dev-IP-Adapter/resolve/main/ip-adapter.bin?download=true)
* **License:** FLUX-1-dev Non-Commercial License
* **Size:** ~6.5 GB
+
+**6. (Optional) Qwen Image Edit Setup**
+
+StableGen can interface with the experimental Qwen Image Edit 2509 workflow for rapid texture refinements. These components are large and require the ComfyUI GGUF loader.
+
+* **a) Install the ComfyUI GGUF Loader (Custom Node)**
+ 1. Navigate to your ComfyUI custom nodes directory:
+ ```bash
+ cd <YourComfyUIDirectory>/custom_nodes/
+ ```
+ 2. Clone the GGUF loader repository:
+ ```bash
+ git clone https://github.com/city96/ComfyUI-GGUF.git
+ ```
+ 3. Restart ComfyUI to load the new node.
+
+* **b) Download the Qwen Image Edit UNet (GGUF)**
+ * **Directory:** `<YourComfyUIDirectory>/models/unet/`
+ * **Filename:** `Qwen-Image-Edit-2509-Q3_K_M.gguf`
+ * **Download URL:** [https://huggingface.co/QuantStack/Qwen-Image-Edit-2509-GGUF/resolve/main/Qwen-Image-Edit-2509-Q3_K_M.gguf?download=true](https://huggingface.co/QuantStack/Qwen-Image-Edit-2509-GGUF/resolve/main/Qwen-Image-Edit-2509-Q3_K_M.gguf?download=true)
+ * **License:** Apache 2.0
+ * **Size:** ~9.5 GB
+
+* **c) Download the Qwen Image VAE**
+ * **Directory:** `<YourComfyUIDirectory>/models/vae/`
+ * **Filename:** `qwen_image_vae.safetensors`
+ * **Download URL:** [https://huggingface.co/Comfy-Org/Qwen-Image_ComfyUI/resolve/main/split_files/vae/qwen_image_vae.safetensors?download=true](https://huggingface.co/Comfy-Org/Qwen-Image_ComfyUI/resolve/main/split_files/vae/qwen_image_vae.safetensors?download=true)
+ * **License:** Apache 2.0
+ * **Size:** ~254 MB
+
+* **d) Download the Qwen Text Encoder (FP8)**
+ * **Directory:** `<YourComfyUIDirectory>/models/clip/`
+ * **Filename:** `qwen_2.5_vl_7b_fp8_scaled.safetensors`
+ * **Download URL:** [https://huggingface.co/Comfy-Org/Qwen-Image_ComfyUI/resolve/main/split_files/text_encoders/qwen_2.5_vl_7b_fp8_scaled.safetensors?download=true](https://huggingface.co/Comfy-Org/Qwen-Image_ComfyUI/resolve/main/split_files/text_encoders/qwen_2.5_vl_7b_fp8_scaled.safetensors?download=true)
+ * **License:** Apache 2.0
+ * **Size:** ~9.2 GB
+
+* **e) Install the Core Qwen Lightning LoRA (Required for StableGen Presets)**
+ * **Directory:** `<YourComfyUIDirectory>/models/loras/`
+ * **Filename:** `Qwen-Image-Edit-2509-Lightning-4steps-V1.0-bf16.safetensors`
+ * **Download URL:** [https://huggingface.co/lightx2v/Qwen-Image-Lightning/resolve/main/Qwen-Image-Edit-2509/Qwen-Image-Edit-2509-Lightning-4steps-V1.0-bf16.safetensors?download=true](https://huggingface.co/lightx2v/Qwen-Image-Lightning/resolve/main/Qwen-Image-Edit-2509/Qwen-Image-Edit-2509-Lightning-4steps-V1.0-bf16.safetensors?download=true)
+ * **License:** Apache 2.0
+ * **Size:** ~850 MB
+
+* **f) (Optional) Additional Qwen Lightning LoRAs**
+ * **Directory:** `<YourComfyUIDirectory>/models/loras/`
+ * **Qwen Image Edit Lightning 8-Step (bf16):** [https://huggingface.co/lightx2v/Qwen-Image-Lightning/resolve/main/Qwen-Image-Edit-2509/Qwen-Image-Edit-2509-Lightning-8steps-V1.0-bf16.safetensors?download=true](https://huggingface.co/lightx2v/Qwen-Image-Lightning/resolve/main/Qwen-Image-Edit-2509/Qwen-Image-Edit-2509-Lightning-8steps-V1.0-bf16.safetensors?download=true)
+ * **Qwen Image Lightning 4-Step (bf16):** [https://huggingface.co/lightx2v/Qwen-Image-Lightning/resolve/main/Qwen-Image-Lightning-4steps-V1.0-bf16.safetensors?download=true](https://huggingface.co/lightx2v/Qwen-Image-Lightning/resolve/main/Qwen-Image-Lightning-4steps-V1.0-bf16.safetensors?download=true)
+ * Each LoRA is licensed under Apache 2.0 and is approximately 850 MB.