Midjourney Inpainting and Vary Region: Edit Parts of Your Images
📋 Table of Contents
What Is Vary Region?
Vary Region (Midjourney's version of inpainting) lets you select a specific part of a generated image and regenerate just that area — while keeping everything else intact.
This is a major workflow upgrade. Instead of regenerating an entire image because one element isn't right, you can surgically fix just the problem area:
- Replace a cluttered background with a clean one
- Change a character's clothing
- Fix a hand (AI's notorious weakness)
- Add or remove objects
- Correct text that rendered incorrectly
- Change the sky in a landscape
How to Access Vary Region
- Generate an image in Midjourney (web or Discord)
- Upscale the image you want to edit (click U1, U2, U3, or U4)
- Below the upscaled image, click Vary (Region)
- A canvas editor opens in your browser
- Draw over the area you want to change
- Type your new prompt for that region
- Click the generate arrow
Using the Selection Tools
In the Vary Region editor, you have two selection tools:
Rectangle selection: Good for clean, geometric areas — backgrounds, skies, rectangular objects.
Freehand brush: Better for organic shapes — a character's outfit, specific objects, irregular areas.
Selection tips:
- Select slightly more than the problem area — the boundary blending looks better with some overlap
- Don't select tiny areas (under 10% of the image) — results are less reliable
- Don't select the entire image — at that point, just use regular Vary or regenerate
Practical Use Cases
Fixing Backgrounds
One of the most common uses: you love the subject but the background is wrong.
Original prompt generated: Person in portrait, great face, but messy background
Vary Region workflow:
- Select the background area
- Type: "clean white studio background, soft shadow, professional photography"
- Generate
You keep the perfect portrait, replace only the background.
More background swaps:
- "modern city skyline at golden hour, blurred bokeh"
- "minimalist office interior, soft lighting"
- "autumn park, fallen leaves, natural light"
- "white seamless studio backdrop"
Changing Clothing and Outfits
Select a character's clothing area and describe the new outfit:
- "tailored navy blue suit, white shirt, no tie"
- "casual denim jacket over white t-shirt"
- "elegant red evening gown, strapless"
Keep the selection tight around the clothing and away from the face/hands for best results.
Fixing AI-Generated Hands
Hands are where AI image generators most frequently fail. Vary Region gives you a fix.
- Zoom in on the upscaled image to confirm the hand issue
- Select just the hand area (freehand brush)
- Prompt: "natural human hand, correct anatomy, realistic fingers, five fingers"
- Generate several times and pick the best result
This often takes 2–3 iterations but consistently improves on the original.
Replacing Objects
Remove an unwanted object or swap it for something else:
- Select the object
- Prompt: "empty table surface, natural wood, clean" (to remove)
- Or: "ceramic coffee mug, white, steam rising" (to replace)
Fixing Text in Images
Midjourney struggles with readable text. Use Vary Region to attempt fixes:
- Select the text area
- Prompt: "clear readable text: 'YOUR TEXT HERE', clean typography, black on white"
Results vary — Midjourney still isn't reliable for precise text — but Vary Region gives you the best chance at improvement without full regeneration.
Adjusting Lighting on Subjects
Select a face or subject:
- "soft warm studio lighting, catchlights in eyes, no harsh shadows"
- "dramatic side lighting, dark background, cinematic"
Combining Vary Region with Other Features
After Vary Region → Upscale again: If you made Vary Region edits on an upscale, you may need to upscale again for full resolution.
Zoom Out + Vary Region: Use Zoom Out to expand the canvas, then use Vary Region to clean up the new edges that Zoom Out generated.
Multiple passes: You can run Vary Region multiple times on the same image for different areas. Fix the background, then fix the hands, then adjust the clothing — iterating to the final result.
Tips for Best Results
Describe what you WANT, not what to remove: Instead of "no power lines in sky", say "clear blue sky with white clouds". Positive descriptions work better than negative ones.
Match the style of the rest of the image: If your image is shot in a specific photography style, mention it in the Vary Region prompt. "Replace background with office interior, same soft natural light, same shallow depth of field" keeps the image feeling unified.
Generate multiple variations: The arrow button in the editor generates multiple results. Don't settle for the first one — run it 2–3 times and pick the best.
Keep a copy of the pre-edit upscale: Before running Vary Region, right-click and save the original upscale. If the edits make things worse, you still have the original.
Use smaller selections for precision: A tighter selection gives Midjourney less room to deviate from the surrounding context, leading to more seamless results.
Vary Region is one of those features that doesn't sound exciting until you need it — then it becomes essential. The ability to fix specific problems without losing a good overall composition saves enormous time and allows you to get images to exactly the quality you need.
📚 Continue Learning
How to Create Consistent Characters in Midjourney
Learn how to generate the same character across multiple Midjourney images using Character Reference, style references, and detailed prompting — essential for storytelling and branding.
Midjourney for Business: Create Professional Marketing Visuals with AI
Learn how to use Midjourney to create professional marketing images, social media graphics, product mockups, and brand visuals — without hiring a designer for every project.
Midjourney Beginner's Guide: Create Your First AI Image in 15 Minutes
A complete step-by-step guide to getting started with Midjourney — setting up Discord, understanding the interface, writing your first prompts, and generating high-quality AI images from scratch.
Midjourney Prompt Writing: How to Describe Images Like a Pro
The complete guide to writing Midjourney prompts that generate exactly what you want — covering structure, style descriptors, composition terms, lighting language, and the keywords that transform average results into stunning images.