Flux 2 LoRA Edit: Precision Image-to-Image Editing AI

blackforestlabs/flux-2/lora/edit

Transform and edit your images using the Black Forest Labs Flux 2 LoRA model to apply custom styles, characters, or details via LoRA adapters on RunComfy.

Idle

The rate is $0.021 per image.

Introduction to Flux 2 LoRA Edit

Flux 2 LoRA edit allows creators to transform, restyle, and edit existing images by combining the precision of Black Forest Labs' architecture with custom LoRA adapters. This specialized Flux 2 LoRA workflow is perfect for teams needing to maintain consistent character details or apply specific artistic styles to uploaded visuals. For developers, Flux 2 LoRA edit on RunComfy can be used both in the browser and via an HTTP API, so you don’t need to host or scale the model yourself.

Examples of Flux 2 LoRA Creations

Flux 2 LoRA on X: Insights and Releases

For developers: API & integration quickstart

Integrate the Flux 2 LoRA image editing capabilities into your applications with a few lines of code. This endpoint supports uploading source images and applying multiple LoRAs simultaneously.

Access / API key

To access the Flux 2 LoRA API, navigate to your dashboard settings on RunComfy to generate your unique API Key. This key authenticates your requests to our playground API.

Response format & Parameters

The API returns a JSON object containing the run_id and the final output URL of the generated image. Here are the key parameters for this Flux 2 LoRA mode:

Parameter	Type	Description
`prompt`	String	(Required) Text description of the desired edit or style.
`image_urls`	Array	URL of the source image(s) to be edited (Max 3).
`loras`	Array	List of adapters. Contains `path` (HuggingFace ID/URL) and `scale` (0.0 to 4.0).
`guidance_scale`	Float	Controls prompt adherence. Default is 2.5.
`num_inference_steps`	Integer	Steps for generation quality. Default is 28.
`image_size`	String	Presets like `square_hd`, `portrait_16_9`, etc.

Model overview

Flux 2 LoRA for Image Editing

While the standard Flux 2 LoRA Text-to-Image model excels at creating visuals from scratch, the Flux 2 LoRA image-to-image variant is engineered for transformation. It leverages the underlying Rectified Flow transformer of Flux 2 but conditions the generation on an input image. This allows users to keep the composition of a source photo while radically changing the art style or subject details using LoRA weights.

Capabilities

Flux 2 LoRA offers a versatile toolkit for modern AI image pipelines:

Image-to-Image Restyling: Apply a "Sketch" or "Oil Painting" LoRA to a photograph to change its medium while keeping the subject.
Character Consistency: Upload a sketch or base render and apply a Character LoRA to finalize the look.
Text-to-Image Generation: For generating images from scratch without input images, please use our Flux 2 LoRA Text-to-Image page.
Multi-LoRA Composition: Blend up to 3 distinct LoRAs (e.g., Style + Clothing + Lighting) onto an existing image.

Technical specs & requirements

Architecture: Flux 2 (Transformer-based) with Image-Conditioning.
Input Types: Text Prompts + Image URLs (JPEG/PNG/WEBP).
LoRA Compatibility: Supports standard Flux 2 .safetensors adapters.
Resolution: Optimized for high-definition outputs (up to 2048px on the longest side).
License: The base Flux 2 model follows the Black Forest Labs license. Individual LoRAs adhere to their creators' specific licenses (check Hugging Face for details).

Run Flux 2 LoRA Edit on RunComfy

Playground Experience

Our interactive playground allows you to test Flux 2 LoRA editing workflows instantly. Simply drag and drop your source image, paste the LoRA URL (or HuggingFace ID), and adjust the lora_scale slider. This visual interface is perfect for finding the right balance between the original image structure and the LoRA's influence.

Use as API

Once you have refined your workflow in the playground, you can seamlessly transition to production. Click the "API" tab to view the code snippet for the Flux 2 LoRA endpoint, or visit the Flux 2 LoRA API Page for documentation. RunComfy handles the caching of LoRA files, ensuring your image-to-image tasks run with minimal latency.

Official resources & ecosystem

Hugging Face LoRA Library

https://huggingface.co/models?other=lora

Base Model Card (Flux.2-dev)

https://huggingface.co/black-forest-labs/FLUX.2-dev

Official GitHub

https://github.com/black-forest-labs/flux2

Related Models

flux-2/flash/edit

Accelerate visual editing with dynamic precision and open-weight adaptability for brand-consistent designs.

seedream-4-0/edit-sequential

Create cohesive visual sequences with precise style and continuity control.

flux-2/flash/text-to-image

Generate accurate design visuals with refined control and repeatable detail.

qwen-edit-2509/lora/inscene

Advanced AI editing merges scenes and styles with precise structure control for designers.

imagen-4/fast/text-to-image

Generate images fast from text with Google Imagen 4 Fast.

qwen-image/qwen-image-edit-2511/lora

LoRA-based visual editing model offering structure-aware asset transformation for creative pros

Frequently Asked Questions

What are the primary capabilities of Flux 2 Lora image-to-image?

Flux 2 Lora image-to-image is designed to modify existing visuals rather than generating them from scratch. It leverages the Flux 2 architecture to restyle photographs, maintain character consistency across different poses, or apply artistic effects using specific adapters. Unlike the standard text-to-image mode, this Flux 2 Lora playground accepts source images to guide the composition.

Can I use Flux 2 Lora on RunComfy for commercial purposes?

Commercial usage of Flux 2 Lora depends on two factors: the license of the base model (Flux 2 Dev is generally Non-Commercial without a separate agreement) and the license of the specific LoRA adapters you choose to load. You must ensure your project complies with Black Forest Labs' terms and the individual creator's rights for any LoRA files used in your Flux 2 Lora workflow.

What are the input limits for images and LoRAs?

The Flux 2 Lora image-to-image endpoint allows you to upload up to 3 source images via the image_urls parameter. Similarly, you can stack up to 3 distinct LoRA adapters simultaneously using the loras list. This multi-input capability makes Flux 2 Lora highly flexible for complex editing tasks involving style mixing and character insertion.

How does RunComfy handle performance and latency for Flux 2 Lora?

RunComfy utilizes high-performance cloud GPUs and an intelligent caching system. Once a specific adapter is loaded for a Flux 2 Lora task, subsequent requests using that same LoRA are processed significantly faster. This ensures low latency for interactive applications, even when handling the heavy computational load of the Flux 2 Lora architecture.

How do I transition from the Playground to the API?

Moving to production is seamless. After testing your prompts and LoRA weights in the Flux 2 Lora playground, simply click the "API" tab. RunComfy provides the exact

RunComfy

RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.

Flux 2 LoRA Edit: Precision Image-to-Image Editing AI | RunComfy