Pony with Z-image refiner workflow

Pony with Z-image refiner is a ComfyUI workflow for using a Z-image refiner with Pony; it uses six model files and includes PreviewImage and SaveImage nodes for image outputs.

Its graph includes EmptyLatentImage, VAEEncode, ImageScaleToMaxDimension, KSampler, LatentUpscaleBy, ModelSamplingAuraFlow, and VAEDecode.

The listed model files are ae.safetensors, qwen_3_4b.safetensors, z_image_turbo_bf16.safetensors, pony-20-real-dream-float8_e4m3fn.safetensors, real-dream-z-2.1.safetensors, and sd_xl_base_1.0.safetensors.

The workflow lists PreviewImage and SaveImage as its image-output nodes.

It is published by Sinatra on Civitai and uses ZImageTurbo as its base model.

ae.safetensors, qwen_3_4b.safetensors, and z_image_turbo_bf16.safetensors are associated with Comfy-Org/z_image_turbo.

sd_xl_base_1.0.safetensors is listed with stabilityai/stable-diffusion-xl-base-1.0; no source is listed for pony-20-real-dream-float8_e4m3fn.safetensors or real-dream-z-2.1.safetensors.

Named ComfyUI nodes include CheckpointLoaderSimple, CLIPLoader, CLIPTextEncode, ConditioningZeroOut, EmptyLatentImage, ImageScaleToMaxDimension, KSampler, LatentUpscaleBy, ModelSamplingAuraFlow, PreviewImage, SaveImage, UNETLoader, VAEDecode, VAEEncode, and VAELoader.

The workflow description says CTRL + B activates the Lora Node.

Before use, verify that pony-20-real-dream-float8_e4m3fn.safetensors, real-dream-z-2.1.safetensors, and sd_xl_base_1.0.safetensors are available, and that VAEEncode, UNETLoader, PreviewImage, and VAELoader are available in your ComfyUI setup.

Suggestion · not verified

No license is listed, so verify usage terms before use.

Suggestion