logo
RunComfy

AI model workflow

FLUX ComfyUI Workflows

Generate and edit images with ready-to-run FLUX ComfyUI workflows for text-to-image, image-to-image, inpainting, LoRA, ControlNet, and consistent characters. Run them on RunComfy Cloud, download workflow JSON, or deploy supported workflows as APIs.

47 workflows

Run nowView all workflows

Featured workflow

Flux 2 Dev | Photoreal Text-to-Image Generator

The workflows

47 FLUX ComfyUI Workflows

Each FLUX ComfyUI example below is a ready-to-run workflow for FLUX.1, FLUX.1 Dev, FLUX.1 Kontext, FLUX.1 Fill, FLUX.1 Redux, FLUX.2 Dev, FLUX.2 Klein 4B, FLUX.2 Klein 9B, FLUX Krea Dev, LoRA, ControlNet, PuLID, IPAdapter, inpainting, outpainting, upscaling, and face restoration. Run it on RunComfy Cloud, download the workflow JSON, or deploy supported workflows as APIs.

32 of 47 workflows shown

About FLUX

What is a FLUX ComfyUI workflow?

A FLUX ComfyUI workflow is a reusable node graph that connects a Black Forest Labs FLUX model with its text encoders, VAE, prompts or reference images, and any LoRA, ControlNet, mask, or post-processing nodes. Instead of building the graph from scratch, you can open a working example, change the inputs, and run or download it.

Choose a FLUX.2 ComfyUI workflow for current image generation and editing, including multi-reference edits and faster Klein variants. Use a FLUX.1 Dev ComfyUI workflow for high-quality text-to-image and LoRA work, FLUX.1 Schnell for faster generation, FLUX.1 Kontext for instruction-based image-to-image editing and consistent characters, or FLUX.1 Fill for inpainting and outpainting. Model size and quantization determine how much VRAM the graph needs.

On RunComfy, each workflow page shows a real output and the files the graph requires. Click Run now to open it in a cloud ComfyUI environment with the setup prepared, or download the FLUX ComfyUI workflow JSON and import it into your own ComfyUI. Local use requires the matching checkpoint, text encoder, VAE, and custom nodes.

References and model resources: Black Forest Labs FLUX.2 overview, Black Forest Labs FLUX GitHub, ComfyUI GitHub, Explore FLUX models on Hugging Face, Explore FLUX models on Civitai.

Capabilities

What you can create with these FLUX workflows

Use a FLUX ComfyUI image-to-image workflow for prompt-guided edits, or choose dedicated graphs for text-to-image generation, inpainting and outpainting, and structure control.

FLUX | A New Art Image Generation

Text-to-image generation

Generate images from text prompts through a general-purpose FLUX ComfyUI workflow.

Run now

FLUX image-to-image editing

Edit an existing image with text instructions and references while retaining its structure and identity.

Run now

Structure-guided generation

Guide image generation with pose, depth, Canny, soft-edge, or grayscale inputs.

Run now

Inpainting and outpainting

Fill masked regions or extend image boundaries using prompt and mask inputs.

Run now

FLUX models on RunComfy

Choose the right FLUX model. Run the complete workflow.

RunComfy turns FLUX ComfyUI workflows into ready-to-run setups, not just examples to copy. Each workflow opens in native ComfyUI with the required models, checkpoints, LoRAs, custom nodes, and input assets already prepared, so you can start generating or editing without rebuilding the graph locally. For heavier FLUX pipelines, choose a cloud GPU that matches the workflow's memory requirements.

Run a FLUX workflow now

  • Native ComfyUI graphs for FLUX generation and editing

    Open text-to-image, Kontext editing, Fill inpainting and outpainting, and FLUX.2 workflows as complete node graphs. Change prompts, references, masks, samplers, guidance, and resolution in native ComfyUI.

  • FLUX checkpoints, LoRAs, and control models in the cloud

    Bring compatible FLUX checkpoints, custom LoRAs, ControlNet models, and supporting weights from Civitai, Hugging Face, or Google Drive and keep them available across sessions.

  • Custom nodes for specialized FLUX workflows

    Use ComfyUI-Manager to add and manage nodes for FLUX ControlNet, PuLID, IPAdapter, face restoration, and upscaling workflows without rebuilding your local environment.

  • Cloud GPU memory for small and large FLUX setups

    Choose a cloud GPU that fits FLUX.2 Klein, FLUX.1 Dev, or larger FLUX.2, ControlNet, and LoRA pipelines instead of reducing the workflow to fit a local GPU.

Applications

What people use these FLUX workflows for

Start from the image result you want, then open a matching FLUX workflow to inspect and customize the complete graph.

AI Face Swap

Replace or transfer a face into a target photo while preserving lighting, expression, and style.

Run now

Object Placement

Place a product or object reference into a target scene while matching its scale, perspective, and surroundings.

Run now
Flux Consistent Characters | Input Image

Consistent Character Images

Reuse a character reference to create new poses, outfits, and scenes without losing identity.

Run now

Character Turnaround Sheet

Create front, side, back, and three-quarter character reference views for design work.

Run now

Photo Restoration

Restore faces and upscale low-resolution images for cleaner portraits and archival photos.

Run now
FLUX LoRA (RealismLoRA) | Photorealistic Images

FLUX ComfyUI LoRA Workflow

Apply FLUX LoRAs for realism, trained styles, characters, or brand-specific image looks.

Run now

Frequently asked questions

FLUX workflow FAQ

Answers for choosing a FLUX ComfyUI workflow, matching it to the right model version, and running or downloading the graph.

Which FLUX model should I use?

Choose the model by what you want to create. Use a FLUX.2 ComfyUI workflow for current generation and multi-reference image editing, with FLUX.2 Klein when speed matters. Choose a FLUX.1 Dev ComfyUI workflow for high-quality text-to-image and LoRA work, FLUX.1 Schnell for faster generation, FLUX.1 Kontext for instruction-based editing and consistent characters, or FLUX.1 Fill for inpainting and outpainting. Use a workflow built for the same FLUX version instead of swapping an unrelated checkpoint into the graph.

How much VRAM do I need to run FLUX in ComfyUI?

It depends on the FLUX version, model format, image size, and workflow. Quantized FP8 or GGUF workflows can make basic FLUX generation possible on some 8GB to 16GB GPUs, while full-size models and workflows using LoRAs, ControlNet, or high-resolution processing may need substantially more memory. If your local GPU does not have enough VRAM, you can run the workflow on a RunComfy cloud GPU instead of changing the model or reducing the workflow to fit your computer.

Should I use FLUX Kontext or FLUX Fill for image editing?

Use FLUX Kontext when you want to describe a change in plain language, such as changing a background, restyling an image, replacing an object, or keeping the same character across different scenes. Use FLUX Fill when you want to edit a specific masked area, remove an object, repair part of an image, or extend the image beyond its original borders. For workflows that combine several reference images, choose a compatible FLUX.2 workflow.

How do I keep a character consistent with FLUX?

Start with a FLUX Kontext or identity-focused workflow and reuse the same reference image across generations. For stronger identity matching, choose a workflow built around PuLID, ACE, or a character LoRA. Keeping the FLUX model, reference image, aspect ratio, and core character description consistent will make it easier to change the pose, outfit, background, or scene without losing the character's identity.

How do I use LoRA or ControlNet in a FLUX ComfyUI workflow?

Choose a FLUX ComfyUI workflow with LoRA to add a trained character, product, or visual style. Choose a compatible FLUX ControlNet workflow when you need pose, depth, Canny edges, or other structural guidance. The base model and add-on must target the same FLUX version. To train your own LoRA, choose FLUX.1, FLUX.2, FLUX.2 Klein 4B, or FLUX.2 Klein 9B in RunComfy Trainer.

Where can I download a FLUX ComfyUI workflow JSON?

Click Download JSON on a workflow card or page to save the FLUX ComfyUI workflow to your computer. Import the JSON with ComfyUI's Workflows > Open command and continue editing the graph. Local ComfyUI still needs the checkpoint, text encoders, VAE, and custom nodes required by that workflow. You can also click Run now to open the prepared workflow on RunComfy Cloud instead.

What do I need for a local FLUX ComfyUI install?

For local use, install ComfyUI, then add the checkpoint, text encoder, VAE, and any custom nodes required by the workflow JSON. The exact files differ between FLUX.1 Dev, Schnell, Kontext, Fill, and FLUX.2 workflows. If you do not want to install or manage those dependencies, click Run now to open the prepared workflow on RunComfy Cloud.

Related resources

More ways to use FLUX

Train a LoRA with a supported FLUX base model, or use FLUX models directly when you do not need a full ComfyUI workflow.

Train a FLUX.1 LoRA

Train a FLUX.1 Dev LoRA for a character, product, style, or other reusable visual concept.

Open FLUX.1 Trainer

Train a FLUX.2 LoRA

Create a FLUX.2 LoRA for a character, product, brand style, or visual look in RunComfy Trainer.

Open FLUX.2 Trainer

Train a FLUX.2 Klein 4B LoRA

Choose FLUX.2 Klein 4B for a smaller, faster LoRA training and inference workflow.

Open Klein 4B Trainer

Train a FLUX.2 Klein 9B LoRA

Choose FLUX.2 Klein 9B when you want the higher-capacity Klein base model for LoRA training.

Open Klein 9B Trainer

Use FLUX models directly

Open the FLUX Models page when you want direct model generation instead of starting from a workflow.

Browse FLUX Models

Ready to run FLUX ComfyUI workflows?

Run now