Flux

Type: comfy.text_to_image.Flux

Namespace: comfy.text_to_image

Description

Generates images from text prompts using the Flux model. image, text-to-image, generative AI, flux

ComfyUI-native Flux implementation in the comfy.text_to_image namespace.

Properties

Property Type Description Default
unet_model hf.text_to_image The UNet/diffusion model to use. {'type': 'hf.text_to_image', 'repo_id': '', 'path': None, 'variant': None, 'allow_patterns': None, 'ignore_patterns': None}
clip_model hf.clip The primary Flux CLIP checkpoint (clip-l). {'type': 'hf.clip', 'repo_id': 'Comfy-Org/stable-diffusion-3.5-fp8', 'path': 'text_encoders/clip_l.safetensors', 'variant': None, 'allow_patterns': None, 'ignore_patterns': None}
clip_model_secondary hf.clip The secondary Flux CLIP checkpoint (t5xxl). {'type': 'hf.clip', 'repo_id': 'Comfy-Org/stable-diffusion-3.5-fp8', 'path': 'text_encoders/t5xxl_fp8_e4m3fn_scaled.safetensors', 'variant': None, 'allow_patterns': None, 'ignore_patterns': None}
vae_model hf.vae The Flux VAE checkpoint. {'type': 'hf.vae', 'repo_id': 'ffxvs/vae-flux', 'path': 'ae.safetensors', 'variant': None, 'allow_patterns': None, 'ignore_patterns': None}
prompt str The prompt to use. ``
negative_prompt str The negative prompt to use. ``
width int   1024
height int   1024
steps int   20
guidance_scale float   5.0
seed int   0
denoise float   1.0
scheduler Enum['normal', 'karras', 'exponential', 'sgm_uniform', 'simple', 'ddim_uniform', 'beta', 'linear_quadratic']   simple
sampler Enum['ddim', 'ddpm', 'dpm_2', 'dpm_2_ancestral', 'dpm_adaptive', 'dpm_fast', 'dpmpp_2m', 'dpmpp_2m_sde', 'dpmpp_2m_sde_gpu', 'dpmpp_2s_ancestral', 'dpmpp_3m_sde', 'dpmpp_3m_sde_gpu', 'dpmpp_sde', 'dpmpp_sde_gpu', 'euler', 'euler_ancestral', 'heun', 'heunpp2', 'lcm', 'lms', 'uni_pc', 'uni_pc_bh2']   euler

Outputs

Output Type Description
output image  

Metadata

Browse other nodes in the comfy.text_to_image namespace.