jdopensource
/

JoyAI-Image-Edit-ComfyUI

@@ -9,54 +9,74 @@ tags:
 base_model: jdopensource/JoyAI-Image-Edit-Diffusers
 ---
-# JoyAI-Image-Edit (ComfyUI)
-Repackaged single-file weights of [`jdopensource/JoyAI-Image-Edit-Diffusers`](https://huggingface.co/jdopensource/JoyAI-Image-Edit-Diffusers) for ComfyUI's standard model-folder layout. Keys, filenames and directory layout follow the [Comfy-Org](https://huggingface.co/Comfy-Org) packaging convention used by `Qwen-Image-Edit_ComfyUI` and `Wan_2.1_ComfyUI_repackaged` — drop the files into `ComfyUI/models/` and they load natively.
 ## Files
 | File | Size | Goes into | Component |
 |------|------|-----------|-----------|
-| `split_files/diffusion_models/joyai_image_edit_bf16.safetensors` | ~32 GB | `ComfyUI/models/diffusion_models/` | `JoyImageEditTransformer3DModel` (bf16) |
-| `split_files/text_encoders/qwen_3vl_8b_joyimage_bf16.safetensors` | ~17 GB | `ComfyUI/models/text_encoders/` | Qwen3-VL-8B text encoder, JoyImage-finetuned (bf16) |
-| `split_files/vae/wan_2.1_vae.safetensors` | ~250 MB | `ComfyUI/models/vae/` | Wan 2.1 VAE (shared with Wan / Qwen-Image; bf16) |
-The VAE is the standard Wan 2.1 VAE — bit-identical to [`Comfy-Org/Wan_2.1_ComfyUI_repackaged/split_files/vae/wan_2.1_vae.safetensors`](https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/vae/wan_2.1_vae.safetensors). If you already have it from Wan or Qwen-Image, you can skip the VAE here.
-## Quick start
-```bash
-# Download into ComfyUI/models/ (skips split_files/ wrapper):
-hf download jdopensource/JoyAI-Image-Edit-ComfyUI \
-  --include "split_files/*" \
-  --local-dir /path/to/ComfyUI/models \
-  --local-dir-use-symlinks False
-# Or fetch each file individually into the matching model folder.
 ```
-## Requirements
-Native JoyImageEdit support is in flight upstream against [Comfy-Org/ComfyUI](https://github.com/comfyanonymous/ComfyUI). Until the PR lands, install ComfyUI from the JoyAI fork's branch:
 ```bash
-git clone https://github.com/jd-opensource/JoyAI-Image.git
-# follow the ComfyUI branch instructions in JoyAI-Image/README.md
 ```
-When the upstream PR merges, plain `comfyanonymous/ComfyUI` will detect and load these files with no extra setup.
-## Key namespaces
-For developers / quantizers — the top-level state-dict prefixes in each file:
-- **diffusion_models**: raw transformer keys (`double_blocks.*`, `condition_embedder.*`, `img_in.*`, `proj_out.*`, `norm_out.*`). No `model.diffusion_model.` wrapper. Auto-detected by ComfyUI's `model_detection.py`.
-- **text_encoders**: HF Qwen3-VL keys (`model.language_model.*`, `model.visual.*`, `lm_head.*`). ComfyUI strips the `model.` wrappers internally.
-- **vae**: standard Comfy-Org Wan 2.1 VAE keys (`encoder.{conv1,downsamples,middle}.*`, `decoder.{conv1,upsamples,middle,head}.*`, plus top-level `conv1.*`/`conv2.*` for the post-quant projections).
 ## Links
-- Source code & nodes: [github.com/jd-opensource/JoyAI-Image](https://github.com/jd-opensource/JoyAI-Image)
-- Diffusers weights: [jdopensource/JoyAI-Image-Edit-Diffusers](https://huggingface.co/jdopensource/JoyAI-Image-Edit-Diffusers)
-- GGUF quantizations (community): [huangfeice/JoyAI-Image-Edit-Diffusers-GGUF](https://huggingface.co/huangfeice/JoyAI-Image-Edit-Diffusers-GGUF)
-- License: Apache 2.0

 base_model: jdopensource/JoyAI-Image-Edit-Diffusers
 ---
+# JoyAI-Image-Edit (ComfyUI weights)
+Single-file `.safetensors` checkpoints of [JoyAI-Image-Edit](https://github.com/jd-opensource/JoyAI-Image), repackaged for **native ComfyUI** support (no custom node required).
+JoyAI-Image-Edit is the single-image instruction-guided editing model of the [JoyAI-Image](https://github.com/jd-opensource/JoyAI-Image) family. It takes one reference image plus a text instruction and generates the edited result.
 ## Files
 | File | Size | Goes into | Component |
 |------|------|-----------|-----------|
+| `diffusion_models/joy_image_edit_bf16.safetensors` | ~31 GB | `ComfyUI/models/diffusion_models/` | `JoyImageEditTransformer3DModel` (bf16) |
+| `text_encoders/qwen3vl_joyimage_bf16.safetensors` | ~17 GB | `ComfyUI/models/text_encoders/` | Qwen3-VL-8B text encoder (bf16) |
+| `vae/joy_image_edit_vae.safetensors` | ~243 MB | `ComfyUI/models/vae/` | `AutoencoderKLWan` |
+The repo's directory layout already matches `ComfyUI/models/`, so a single `hf download` into your models root drops every file where it needs to go.
+## Installation
+The model runs natively in ComfyUI. Native support is proposed upstream in [Comfy-Org/ComfyUI#14428](https://github.com/Comfy-Org/ComfyUI/pull/14428); until it is merged, install the fork branch:
+```bash
+git clone -b joyimage-edit-pr https://github.com/feice-huang/ComfyUI.git
+cd ComfyUI
+pip install -r requirements.txt
 ```
+Once the PR is merged upstream, the stock ComfyUI release will run these weights with no fork needed.
+Then download the weights straight into `ComfyUI/models/`:
 ```bash
+hf download jdopensource/JoyAI-Image-Edit-ComfyUI \
+  --local-dir /path/to/ComfyUI/models
 ```
+Restart ComfyUI.
+## Usage
+Build the graph from these native nodes:
+1. **Load Diffusion Model** (`UNETLoader`) → `diffusion_models/joy_image_edit_bf16.safetensors`
+2. **Load CLIP** (`CLIPLoader`) → `text_encoders/qwen3vl_joyimage_bf16.safetensors`, type `joyimage`
+3. **Load VAE** (`VAELoader`) → `vae/joy_image_edit_vae.safetensors`
+4. **Load Image** (`LoadImage`) for the reference
+5. **TextEncodeJoyImageEdit** — feed `clip`, `vae`, the instruction, and the reference `image`. Wire one instance for the positive prompt and one (empty prompt, same image) for the negative. The node bucket-resizes the reference to the 1024-base buckets, VAE-encodes it, and appends the reference latent to the conditioning; its `image` output feeds `VAEDecode` / empty-latent sizing.
+6. **KSampler** → **VAEDecode** → **SaveImage**
+Example workflow: [workflow_joyimage_edit.json](https://github.com/user-attachments/files/28871922/workflow_joyimage_edit.json)
+## Recommended parameters
+| Parameter | Value |
+|-----------|-------|
+| Steps | 40 |
+| CFG | 4.0 |
+| Sampler | `euler` |
+| Scheduler | `simple` |
+| dtype | bf16 |
+| Resolution | auto (1024-base buckets) |
+## GGUF quantizations
+Lower-bit GGUF quants of the transformer and text encoder are available at [huangfeice/JoyAI-Image-Edit-Diffusers-GGUF](https://huggingface.co/huangfeice/JoyAI-Image-Edit-Diffusers-GGUF) (community contribution). The VAE here is the only VAE you need — GGUF doesn't quantize the VAE.
 ## Links
+- Source code and documentation: [github.com/jd-opensource/JoyAI-Image](https://github.com/jd-opensource/JoyAI-Image)
+- Original Diffusers-format weights: [jdopensource/JoyAI-Image-Edit-Diffusers](https://huggingface.co/jdopensource/JoyAI-Image-Edit-Diffusers)
+- Multi-image edit model (ComfyUI): [jdopensource/JoyAI-Image-Edit-Plus-ComfyUI](https://huggingface.co/jdopensource/JoyAI-Image-Edit-Plus-ComfyUI)