🌊🚒 FlowState Flux Engine

Streamline your entire generation workflow with the FlowState Flux Engine.

An all-in-one node that seamlessly combines:

• Model, VAE, and CLIP loading (with Model Persistence to prevent subsequent reloads.)
• Prompt conditioning & guidance.
• Sage Attention model patching (for users with that capability)
• The integrated FlowState Latent Source for flexible resolution selection & easy i2i style transfer.

Reduce your node clutter and go from concept to final image faster than ever within a single, powerful engine.

Inputs:

Name	Type	Description
model_filetype	STRING	Model File Type: The type of model file to load.
model_name	STRING	Full Diffusion Model List: List of all available Diffusion Models.
weight_dtype	STRING	Weight Datatype (DType): The data type to be used for your models weights.
sage_attention	STRING	Sage Attention Mode: The type of Sage Attention to use.
lora_model	STRING	LoRA Model: Select a LoRA model to use.
lora_strength	STRING	LoRA Strength: If using a LoRA, specify strength.
clip_1_name	STRING	CLIP / Text Encoder List: List of available Text Encoders and CLIP models.
clip_2_name	STRING	CLIP / Text Encoder List: List of available Text Encoders and CLIP models.
vae_name	STRING	VAE List: List of available Variational Autoencoders (VAE).
resolution	STRING	Resolution Selector: Select custom to use the entered width & height, or select a resolution.
orientation	STRING	Orientaion Selector: Resolutions given in horizontal orientation. Select vertical to swap resolution aspect ratio.
latent_type	STRING	Latent Type: Your choice of an empty latent (all zeros) or an image as a latent.
custom_width	INT	Width: Defines the width of the image.
custom_height	INT	Height: Defines the height of the image.
custom_batch_size	INT	Custom Batch Size: The number of images you want to generate.
image	STRING	Uploaded Image: Path to the image file to be used when latent_type is 'Uploaded Image'.
seed	INT	Seed: Seed used to generate inital random noise.
sampling_algorithm	STRING	Sampling Algorithm: List of available Sampling Algorithms.
scheduling_algorithm	STRING	Scheduling Algorithm: List of available Scheduling Algorithms.
guidance	FLOAT	Guidance: Defines the guidance scale.
steps	INT	Steps: Defines the number of steps to take in the sampling process.
denoise	FLOAT	Sampler Denoise Amount: The amount of denoising applied.
prompt	STRING	Positive Prompt: Positive text prompt describing your desired output.
input_img (optional)	IMAGE	Input Image: An optional image passed from another node.

Outputs:

Name	Type	Description
model	MODEL	The selected Diffusion Model.
clip	CLIP	The selected CLIP.
vae	VAE	The selected VAE.
image	IMAGE	The image batch.
latent	LATENT	The latent batch.

Limitations:

• Only tested on core Flux models: Dev, Schnell, Krea, Fill, Chroma, Kontext. Cannot guarantee compatiblity with versions from other developers such as De-Distilled, etc.

Video tutorials can be found on YouTube at 🌊 FlowState Creator Suite Playlist (Coming Soon)