EasyUse/Loadersgenerated

EasyLoader (Flux)(fluxLoader)

The EasyLoader (Flux) node, class fluxLoader, loads a Flux checkpoint for inference, assembling the model, VAE, LoRA adjustments, prompt, and latent dimensions...

flux Loader

model_override
clip_override
vae_override
optional_lora_stack
optional_controlnet_stack
pipe
model
vae
ckpt_name
COMBO
vae_name
COMBO
lora_name
COMBO
lora_model_strength
1.00
lora_clip_strength
1.00
resolution
1024 x 1024
empty_latent_width
INT
empty_latent_height
INT
positive
STRING
batch_size
1
Easy Use

Description

The EasyLoader (Flux) node, class fluxLoader, loads a Flux checkpoint for inference, assembling the model, VAE, LoRA adjustments, prompt, and latent dimensions into a pipeline. It also supports optional model/CLIP/VAE overrides, a LoRA stack, and a ControlNet stack, making it a flexible single-node loader for Flux workflows.

Inputs

Required

NameTypeDefaultNotes
ckpt_nameCOMBOCheckpoint to load.
vae_nameCOMBOVAE to load.
lora_nameCOMBOLoRA to apply to the loaded model.
lora_model_strengthFLOAT1Strength of the LoRA on the model weights. Range: -10 to 10, step 0.01. Negative values invert or reduce the effect.
lora_clip_strengthFLOAT1Strength of the LoRA on the CLIP weights. Range: -10 to 10, step 0.01. Negative values invert or reduce the effect.
resolutionCOMBO1024 x 1024Preset resolution for the empty latent. Selecting a preset updates empty_latent_width and empty_latent_height.
empty_latent_widthINTWidth of the empty latent in pixels.
empty_latent_heightINTHeight of the empty latent in pixels.
positiveSTRING""Positive prompt text.
batch_sizeINT1Number of latents to generate in the batch. Range: 1 to 64.

Optional

NameTypeNotes
model_overrideMODELUse an existing model instead of loading from ckpt_name.
clip_overrideCLIPUse an existing CLIP instead of loading from ckpt_name.
vae_overrideVAEUse an existing VAE instead of loading from vae_name.
optional_lora_stackLORA_STACKApply multiple LoRAs as a stack.
optional_controlnet_stackCONTROL_NET_STACKApply multiple ControlNets as a stack.

Outputs

TypeDescription
PIPE_LINEThe assembled pipeline containing the model, VAE, conditioning, and latent settings.
MODELThe loaded or overridden model.
VAEThe loaded or overridden VAE.

Usage Notes

  • When model_override, clip_override, or vae_override are provided, they take precedence over the checkpoint and VAE selections.
  • Use the resolution presets for common Flux dimensions, or set empty_latent_width and empty_latent_height directly for custom sizes.
  • The LoRA strength inputs allow negative values, which can be used to weaken or invert a LoRA's influence.
  • optional_lora_stack and optional_controlnet_stack let you compose multiple adapters in a single load.
  • The PIPE_LINE output can be passed directly to supported sampler or conditioning nodes for further workflow processing.

Comments

Sign in with GitHub to join the discussion.

Loading comments…