Z-Image Examples¶
These examples generate images with the distilled Z-Image-Turbo model through TeleFuser or a fixed Diffusers reference path.
Model Source¶
| Model | HuggingFace | ModelScope | Purpose |
|---|---|---|---|
| Z-Image-Turbo | Tongyi-MAI/Z-Image-Turbo | Tongyi-MAI/Z-Image-Turbo | Distilled text-to-image checkpoint |
Feature Support¶
| Feature | Support | Notes |
|---|---|---|
| Text-to-image | Supported | TeleFuser and Diffusers reference paths |
| Multi-GPU inference | Supported | CFG and Ulysses parallelism through --gpu_num |
| LoRA | Unsupported | No LoRA loader is exposed |
| Quantization | Supported | The TeleFuser model path supports FP8 configuration |
| CPU offload | Unsupported | No CPU-offload option is exposed |
| Feature cache | N/A | Not used by this distilled model example |
| Server API | Supported | The TeleFuser script exposes the standard pipeline functions |
Requirements¶
- GPU: one CUDA GPU; additional GPUs may be selected with
--gpu_num - Software: the standard TeleFuser installation; Diffusers is required for the reference script
- Input assets: none
Install TeleFuser by following the development setup.
Model Directory¶
/path/to/Z-Image-Turbo/
|-- transformer/
| \-- diffusion_pytorch_model-*.safetensors
|-- vae/
|-- text_encoder/
| \-- model-*.safetensors
|-- tokenizer/
\-- scheduler/
The default root is ${TF_MODEL_ZOO_PATH}/Z-Image-Turbo.
Quick Start¶
python examples/z_image/z_image_turbo_t2i_h100.py \
--model_root /path/to/Z-Image-Turbo \
--prompt "A watercolor landscape at sunrise" \
--output work_dirs/z-image-turbo.png
The command writes the generated image to work_dirs/z-image-turbo.png.
Examples¶
TeleFuser Inference¶
z_image_turbo_t2i_h100.py¶
python examples/z_image/z_image_turbo_t2i_h100.py \
--model_root /path/to/Z-Image-Turbo \
--aspect_ratio 16:9 \
--gpu_num 1 \
--output work_dirs/z-image-turbo.png
Key options:
| Option | Default | Description |
|---|---|---|
--model_root | ${TF_MODEL_ZOO_PATH}/Z-Image-Turbo | Local model directory |
--aspect_ratio | 16:9 | Output aspect ratio |
--gpu_num | 1 | GPU count |
--output | Generated name | Output PNG path |
Diffusers Reference¶
z_image_turbo_official.py¶
This fixed-configuration script uses the same model directory and writes z_image_example.png.