muapi Logomuapi Logo
muapi Logomuapi Logo

Qwen Image Edit 2511 LoRA: Lora Support

Edit up to 3 reference images at once with custom LoRA styles on Qwen Image Edit 2511 — stronger multi-image consistency and identity preservation. Try free — pay per generation.

Interactive model controls

Qwen Image Edit 2511 LoRA is an enhanced version with custom LoRA support for personalized styles. It delivers stronger edit consistency, robust multi-person identity/pose consistency, custom LoRA styles, enhanced industrial/product design, and improved geometric reasoning for structure-preserving edits.

📝

Overview

About this model

Qwen Image Edit 2511 LoRA is the enhanced build of Qwen Image Edit with custom LoRA support, adding stronger edit consistency, robust multi-person identity and pose consistency, and improved geometric reasoning for structure-preserving edits. It accepts up to 3 reference images at once via images_list, plus up to 3 LoRA weight files to apply a trained style on top of the edit.

1Multi-image composition: Combine up to 3 reference images into one consistently edited output.
2Identity-preserving edits: Keep multiple people's identity and pose consistent across an edited scene.
3Product design iteration: Apply a trained style LoRA while editing industrial or product renders.
4Structure-preserving retouches: Make geometrically precise edits (e.g. object placement, perspective) without distorting the scene.
💰

Pricing & Value

Cost analysis

muapiapp$0.04 per generation

Flat per-generation price regardless of how many reference images or LoRA weights are used.

Fal.ai$0.05 per generation

Muapiapp is roughly 20% cheaper for comparable Qwen Image Edit LoRA generations.

Replicate$0.05 per generation

Usage-based pricing comparable to Fal.ai; muapiapp remains the cheaper option.

** Competitor pricing is estimated based on similar model architectures and usage tiers.

⚙️

Technical Details

Configuration schema

Promptstring

Text prompt describing the image edits.

Default ValueA cinematic ocean wave at sunrise, highly detailed
Image URLsarray

The images to edit (maximum 3 reference images).

Default Valuehttps://d3adwkbyhxyrtq.cloudfront.net/webassets/videomodels/qwen-image-edit-2511-in.jpg
Aspect RatioEnum (9 options)

Aspect ratio of generated image.

Default Value1:1
LoRAsarray

List of LoRAs to apply (maximum 3).

Default Value[object Object]
Output FormatEnum (3 options)

Format of the output image.

Default Valuejpeg
📖

Implementation Guide

Developer documentation

How to Use Qwen Image Edit 2511 LoRA

  1. Provide Reference Images

    • Set images_list to up to 3 image URLs you want edited or used as reference.

  2. Write the Edit Prompt

    • Describe the desired change across the provided images.

  3. Apply LoRA Weights (Optional)

    • Add up to 3 entries to loras, each with a path (URL to a .safetensors file) and a scale (0–4, default 1).

  4. Set Output Options

    • Choose aspect_ratio and output_format (jpeg, png, or webp).

  5. Submit via the API

curl -X POST https://api.muapi.ai/api/v1/qwen-image-edit-2511-lora \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Blend the two people into one consistent scene with matching lighting.",
    "images_list": ["https://example.com/ref1.jpg", "https://example.com/ref2.jpg"],
    "loras": [{"path": "https://example.com/my-style.safetensors", "scale": 1}]
  }'
  1. Poll and Download
    • Poll GET /api/v1/predictions/{request_id}/result until status is completed, then retrieve the image URL from output.
❓

Common Questions

Frequently asked

What does Qwen Image Edit 2511 LoRA do?

It edits up to 3 reference images together according to your prompt, applying up to 3 custom LoRA weight files to steer style, while preserving multi-person identity, pose, and geometric structure.

How is this different from qwen-image-edit-lora?

This 2511 build accepts up to 3 reference images via `images_list` (vs. a single `image_url`) and offers stronger multi-image edit consistency, identity/pose preservation, and geometric reasoning.

How many LoRAs can I apply at once?

Up to 3, each with its own `scale` multiplier (0–4, default 1).

Where do the LoRA weights come from?

Any publicly reachable `.safetensors` URL, including output from a muapiapp LoRA trainer endpoint or a third-party host.

What output formats are supported?

`jpeg`, `png`, or `webp`, set via `output_format`.