---
license: apache-2.0
tags:
  - image-editing
  - text-to-image
  - multimodal
  - diffusion
  - fp8
  - quantized
base_model: jdopensource/JoyAI-Image-Edit
---

# JoyAI-Image-Edit — FP8 Quantized

FP8 (float8_e4m3fn) quantized version of the [JoyAI-Image-Edit](https://huggingface.co/jdopensource/JoyAI-Image-Edit) model transformer.

## What's Different

| Component | Original | FP8 |
|---|---|---|
| DiT Transformer | 32.5 GB (bf16) | **16.3 GB (fp8_e4m3fn)** |
| VAE | 485 MB (bf16) | 485 MB (bf16, unchanged) |
| Text Encoder (Qwen3VL) | ~17 GB (bf16) | ~17 GB (bf16, unchanged) |

The transformer's 407 linear-layer weight matrices (2D+, ≥1024 elements) are stored in `float8_e4m3fn`. All biases, normalization weights, and small tensors remain in `bfloat16`.

## Why FP8

The full bf16 transformer is 32.5 GB — too large for consumer GPUs like the RTX 4090 (24 GB VRAM). At FP8, the transformer fits in ~16 GB, leaving headroom for activations and the VAE during inference. The text encoder (Qwen3VL) can be offloaded to CPU after conditioning.

RTX 4090 (Ada Lovelace, SM89) has native FP8 hardware support.

## Inference Tool

A Gradio UI, CLI, and REST API for running inference are available at **[SanDiegoDude/JoyAI-Image](https://github.com/SanDiegoDude/JoyAI-Image)** — a handy way to run the model until proper ComfyUI integration lands. Features include auto-download from HuggingFace, multiple memory modes, and bitsandbytes quantization for the text encoder.

### Quick start

```bash
git clone https://github.com/SanDiegoDude/JoyAI-Image.git
cd JoyAI-Image
python -m venv .venv && source .venv/bin/activate
pip install -e .

# Models auto-download on first run (FP8 transformer is the default)
# Default: FP8 transformer + 8-bit text encoder, offload mode
python app.py

# Minimum VRAM (~13 GB active, fits RTX 4090)
python app.py --nf4-dit --4bit-vlm

# CLI inference
python inference.py \
  --prompt "Turn the plate blue" \
  --image test_images/test_1.jpg \
  --output result.png \
  --steps 18 --guidance-scale 4.0 --seed 42

# Headless REST API (with optional ComfyUI connector node)
python app.py --headless-api 7500
```

## Also available

- **[Full bf16 safetensors](https://huggingface.co/SanDiegoDude/JoyAI-Image-Edit-Safetensors)** — original precision, also the source weights for runtime NF4 quantization (~8 GB)

## Conversion

Quantized using a straightforward per-tensor cast to `torch.float8_e4m3fn` with value clamping to the representable range (±448). This matches the approach used by ComfyUI and other diffusion model tooling.

## Model Details

JoyAI-Image is a unified multimodal foundation model comprising:
- **8B MLLM** (Qwen3VL) for text/image understanding
- **16B MMDiT** for image generation (this is the quantized component)
- **Wan2.1 VAE** for encoding/decoding

See the [original repository](https://github.com/jd-opensource/JoyAI-Image) for full documentation.

## Credits

Original model by [JD Open Source](https://github.com/jd-opensource/JoyAI-Image), released under Apache 2.0.
FP8 conversion by SanDiegoDude.