Z-Image gguf file quanted with sd doesn't work

[entauri]

Used

E:\AI\SDcpp\sd-cli -M convert -m E:\AI\ComfyUI_windows_portable\ComfyUI\models\diffusion_models\z_image_de_turbo_v1.safetensors -o E:\AI\ComfyUI_windows_portable\ComfyUI\models\diffusion_models\z_image_de_turbo_v1-q8_0.gguf -v --type q8_0

got

[stduhpf]

That's not an issue from sd.cpp. The GGUF node for ComfyUI makes assumptions about the types of each tensor in the model (which are supposed to be quantized with a special fork of llama.cpp), it is not the case in sd.cpp.