This guide will walk you through completing a ComfyUI native HiDream-I1 text-to-image workflow example
Model Name | Description | Inference Steps | Repository Link |
---|---|---|---|
HiDream-I1-Full | Full version | 50 | 🤗 HiDream-I1-Full |
HiDream-I1-Dev | Distilled dev | 28 | 🤗 HiDream-I1-Dev |
HiDream-I1-Fast | Distilled fast | 16 | 🤗 HiDream-I1-Fast |
cfg
parameter to 1.0
during sampling. We have noted the corresponding parameter settings in the relevant workflows.
ComfyUI/models/diffusion_models/
folder.
Load Diffusion Model
node is using the hidream_i1_full_fp8.safetensors
fileQuadrupleCLIPLoader
are loaded correctly
Load VAE
node is using the ae.safetensors
fileshift
parameter in ModelSamplingSD3
to 3.0
Ksampler
node, you need to make the following settings
steps
to 50
cfg
to 5.0
sampler
to lcm
scheduler
to normal
Run
button, or use the shortcut Ctrl(cmd) + Enter
to execute the image generationComfyUI/models/diffusion_models/
folder.
Load Diffusion Model
node is using the hidream_i1_dev_fp8.safetensors
fileQuadrupleCLIPLoader
are loaded correctly
Load VAE
node is using the ae.safetensors
fileshift
parameter in ModelSamplingSD3
to 6.0
Ksampler
node, you need to make the following settings
steps
to 28
cfg
to 1.0
sampler
to lcm
scheduler
to normal
Run
button, or use the shortcut Ctrl(cmd) + Enter
to execute the image generationComfyUI/models/diffusion_models/
folder.
Load Diffusion Model
node is using the hidream_i1_fast_fp8.safetensors
fileQuadrupleCLIPLoader
are loaded correctly
Load VAE
node is using the ae.safetensors
fileshift
parameter in ModelSamplingSD3
to 3.0
Ksampler
node, you need to make the following settings
steps
to 16
cfg
to 1.0
sampler
to lcm
scheduler
to normal
Run
button, or use the shortcut Ctrl(cmd) + Enter
to execute the image generation