Z-Image-Turbo_clear

• Guide (External site): English | Japanese

• Hugging Face

• VAE: Z-Image_clear_vae

Compare to Z-Image-Turbo (Original)

Z-Image-Turbo_clear | Z-Image-Turbo

The Z-Image-Turbo_clear model is a fine-tuned version of Z-Image-Turbo that improves details compared to the original.

For better results, please use it with the dedicated VAE, Z-Image_clear_vae.

Model Formats and Image Quality

Generally, the image quality ranks in the following order:

FP32 > FP16 ≒ BF16 > INT8_ConvRot_HQ > MXFP8 > FP8_svd_scaled > FP8

• FP32 is mainly recommended for training purposes.

• FP16 / BF16 can generate high-quality illustrations.

• INT8_ConvRot_HQ offers an excellent balance of speed and quality while using roughly half the VRAM of FP16 / BF16.

• MXFP8, FP8_svd_scaled, and FP8 are currently not recommended, as they are inferior to INT8_ConvRot_HQ in both speed and image quality.

For a detailed comparison of each format, please refer to this page.

Recomended Settings

• Sampler:

  • euler

  • res_multistep

• Scheduler: beta

• Steps: 9

• ModelSamplingAuraFlow: shift 7.5 (3-9)

Upscaler

your choice, examples:

• 4x-SwinIR-L_GAN

• 4x_NMKD-Superscale-SP_178000_G

Tips (External Sites)

• ComfyUI VRAM management: English | Japanese

• FP / BF Format Image Quality and Generation Time: English | Japanese

• Steps and Scheduler: English | Japanese

• Color Correct custom nodes in ComfyUI: English | Japanese

License

• Apache-2.0