1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
const axios = require('axios');
const fs = require('fs');
const path = require('path');
async function toB64(imgPath) {
const data = fs.readFileSync(path.resolve(imgPath));
return Buffer.from(data).toString('base64');
}
const api_key = "YOUR API-KEY";
const url = "https://api.segmind.com/v1/qwen-image-edit-plus-eigen-banana";
const data = {
"prompt": "Replace the current plain wall background with a sophisticated, softly lit indoor event space, featuring warm golden ambient lighting, elegant architectural details such as decorative panels or subtle artwork, and a slightly blurred depth of field to keep the focus on the subjects while ensuring the new background's rich, muted tones complement their attire.",
"image_1": "toB64('https://segmind-resources.s3.amazonaws.com/input/efcc5a54-c995-454d-9474-41a8515380c3-example4-input.png')",
"image_2": "toB64('')",
"image_3": "toB64('')",
"lora": "eigen_banana",
"aspect_ratio": "match_input_image",
"seed": 87568756,
"image_format": "webp",
"quality": 95,
"base64": false
};
(async function() {
try {
const response = await axios.post(url, data, { headers: { 'x-api-key': api_key } });
console.log(response.data);
} catch (error) {
console.error('Error:', error.response.data);
}
})();Attributes
Text prompt describing the desired image edit or generation
Primary input image (URL or base64)
Secondary input image (URL or base64)
Tertiary input image (URL or base64)
Pre-configured LoRA model to apply
Allowed values:
Additional LoRA model URL. Public direct url or huggingface url pointing to the lora file
Additional LoRA model URL. Public direct url or huggingface url pointing to the lora file
Output image aspect ratio
Allowed values:
Random seed for reproducibility. Use -1 for random
min : -1,
max : 2147483647
Output image format
Allowed values:
Output image quality (1-100)
min : 1,
max : 100
Return image as base64 encoded string
To keep track of your credit usage, you can inspect the response headers of each API call. The x-remaining-credits property will indicate the number of remaining credits in your account. Ensure you monitor this value to avoid any disruptions in your API usage.
Eigen-Banana-Qwen-Image-Edit: AI Image Editing Model
Edited by Segmind Team on November 23, 2025.
What is Eigen-Banana-Qwen-Image-Edit?
Eigen-Banana-Qwen-Image-Edit is a fine-tuned LoRA extension of the Qwen-Image-Edit model, designed for text-guided image editing. It performs detailed transformations of existing images based on natural language instructions, including altering objects, transferring styles, changing scene elements, and adjusting facial expressions. It is trained on Apple’s extensive Pico-Banana-400K dataset, which includes around 400,000 text-image-edit pairs spanning 35 different editing operations. Eigen-Banana-Qwen-Image-Edit provides fast and high-quality results while supporting prompts in both English and Chinese, making it a versatile tool for creators and developers worldwide, particularly for multilingual image editing solutions.
Key Features of Eigen-Banana-Qwen-Image-Edit
- Multi-Image Support: It enables users to edit single images or up to three images simultaneously for complex compositions and layered effects.
- Diverse Editing Operations: It can handle 35+ editing tasks, including object insertion, removal, style changes, and expression modification.
- Bilingual Processing: It includes native support for English and Chinese text prompts.
- LoRA Flexibility: Users can configure custom LoRA models for specific aesthetic styles and visual effects.
- Precision Control: It offers fixed seed support for reproducible edits across multiple generations.
- Format Options: It can render visual outputs in JPEG, PNG, or WebP with adjustable quality settings.
- Flexible Aspect Ratios: It comes with support for 11 standard ratios or automatic input matching.
Best Use Cases
-
E-commerce Product Photography: It expedites the customization of product backgrounds, adjusts lighting, or adds contextual elements without reshoots.
-
Content Creation: It is effective in transforming social media images, creating variation sets for A/B testing, and maintaining brand consistency across visual assets.
-
Creative Studios: It supports rapid prototyping for concept art, style exploration, and client presentations with iterative refinements.
-
Localization Teams: Users can utilize it to generate culturally adapted visuals with multilingual prompt support (English and Chinese) for international campaigns.
-
Game Development: It is a perfect tool to create character expression variations, environment modifications, and asset iterations efficiently.
Prompt Tips and Output Quality
- Write precise and detailed prompts describing the source image and desired transformation. So, instead of "change the background," use "replace the white studio background with a sunset beach scene with palm trees," for a more impactful output.
Layer complexity gradually while using multiple images; start with image_1 as your primary source, then use image_2 and image_3 for blending or compositional elements.
- Lock seed value (default: 987654321) to refine prompts to isolate the effect of text changes; use seed=-1 for random variation exploration.
Aspect ratio will render desired formats, such as using "match_input_image" will preserve original dimensions. Select specific ratios (16:9, 4:5) for social media platforms that need fixed formats.
- Maximum quality will be possible when you set quality to 100 and choose PNG format for images with sharp edges or text. Use WebP (default) for balanced size and quality in web applications.
FAQs
Is Eigen-Banana-Qwen-Image-Edit open-source?
The model is based on the open Qwen-Image-Edit architecture with proprietary LoRA fine-tuning. Additionally, the base Qwen model follows Tongyi Qianwen's licensing terms.
How does this differ from other image editing models?
Unlike generative inpainting models, Eigen-Banana specializes in instruction-following edits trained on over 400,000 diverse editing operations. Its LoRA architecture provides faster inference than full model fine-tuning while maintaining edit precision.
What aspect ratio should I use for Instagram posts?
Use 4:5 for feed posts (portrait), 1:1 for square grid consistency, or 9:16 for Stories and Reels. Set "match_input_image" to preserve original photo dimensions.
Can I combine multiple LoRA models?
Yes, you can seamlessly use lora_2_url and lora_3_url parameters to layer up to three LoRA styles. This option supports complex aesthetic combinations, such as "vintage film + anime style" transformations.
Why are my edits inconsistent across generations?
To achieve consistent output across multiple generations, set a fixed seed value instead of using the default random seed, because reproducibility requires identical prompts, images, and seed values across API calls.
Does prompt language affect edit quality?
Since prompts in English and Chinese work equally well due to bilingual training, opt for the language that enables you to provide precise editing instructions.
Popular Models
