Image Editing lets you modify existing images by describing the changes you want in natural language. Upload one or more reference images and tell the AI what to change.
How It Works
- Upload the image(s) you want to edit as reference images
- Write a prompt describing the edit (e.g., “Remove the background and replace with a sunset”)
- Choose a provider and model
- Generate - the AI creates a new image with your requested changes
Providers
| Provider | Models | Strengths |
|---|---|---|
| Gemini | Nano Banana 2 / Lite / Pro / classic | Multi-image edits, instruction following |
| Qwen | 3.0 / 3.0 Pro / 2.0 / Pro / Edit Max | Style changes, negative prompts |
| Flux | Flex and full Flux 2 line | Reference-guided edits |
| Seedream | 5.0 Pro / Lite / 4.5 | Realistic edits, multi-ref |
| Wan | 2.7 Image / Pro | Multi-ref transforms |
| OpenAI | ChatGPT Image 2.0 | Up to 16 refs |
| Kling | 3.0 Omni | Omni-image refs |
| xAI (Grok) | Imagine 2.0 / Imagine / Quality | General purpose |
| Microsoft | MAI Image 2.5 / 2.5 Pro | Single-image edits |
| Ideogram | 4.0 | Remix |
| Meta | Muse Image 1.0 | Up to 10 refs; fixed 2K output; contact sheet |
See Image Providers for the full list.
Input Slots
- Slot count is model-dependent (e.g. Nano Banana Pro up to 14, OpenAI up to 16, many models 1–10)
- The first reference is usually the primary subject
- Additional images provide identity, product, or style context
Quick Edit Inpainting
You can also launch Quick Edit straight from the lightbox or asset context menu when you want to patch a specific image instead of rebuilding the full prompt from scratch.
Inside Quick Edit, paint or object-select the area you want to change, write the replacement instruction, then generate a masked inpaint result on that exact asset version. Hylope now supports direct inpainting models in Quick Edit, including Nano Banana and Flux 2 Klein 4B, with Qwen-style quick edit flows replaced by native mask pipelines.
Reformat Preview
If you enable reformat, Hylope previews the source image inside the output frame before you generate. Drag the frame edges to snap across supported aspect ratios, or double-click the frame to return to Auto when that option is available.
Common Edits
- Background replacement - “Replace the background with a tropical beach”
- Object removal - “Remove the person on the left”
- Style transfer - “Make this look like a watercolor painting”
- Color grading - “Apply warm golden hour lighting”
- Object addition - “Add a red sports car in the parking lot”
- Composition changes - “Zoom out to show more of the scene”
Tips
- Be specific in your edit instructions - “make the sky more blue” works better than “improve this”
- Gemini handles complex multi-step edits best
- Upload multiple reference images for style guidance
- Use negative prompts (Qwen) to prevent unwanted changes
- Use reformat preview to check crop pressure before you spend credits on a new aspect ratio
- The more context you provide in the prompt, the better the result