In this categoryImage Tools · 47
- How to Generate Your First Image with Midjourney on DiscordStart
- How to Write Midjourney Prompts That Actually Work
- How to Upscale and Export High-Resolution Images in Midjourney
- How to Create Consistent Characters in Midjourney with --cref
- How to Use Image Prompts and the /blend Command in Midjourney
- How to Lock a Visual Style Across Images with Midjourney --sref
- How to Set the Right Aspect Ratio for Each Platform in Midjourney
- How to Edit Parts of an Image with Vary (Region) in Midjourney
- How to Extend an Image with Pan and Zoom Out in Midjourney
- How to Batch Many Variations with Permutation Prompts in Midjourney
- How to Organize and Rerun Your Images from the Midjourney Website
- How to Install ComfyUI on Windows with an NVIDIA GPU
- How to Run Flux.1 Dev in ComfyUI
- How to Install ComfyUI Manager to Add Nodes and Models Easily
- How to Generate Fast Images with Flux.1 Schnell in ComfyUI
- How to Use a LoRA in ComfyUI
- How to Write Better Prompts for Flux in ComfyUI
- How to Upscale AI Images in ComfyUI with an ESRGAN Model
- How to Inpaint and Fix Part of an Image in ComfyUI
- How to Batch Generate Many Image Variations in ComfyUI
- How to Control Image Composition with ControlNet in ComfyUI
- How to Fix Out-of-Memory and Slow Generations in ComfyUI
- How to Upscale a Photo in Magnific Without Hallucinating New Detail
- How to Restore an Old Scanned Photo with Topaz Photo AI
- How to Remove the Background from a Product Photo in Photoroom
- How to Generate a Lifestyle Background for a Product in Photoroom
- How to Upscale and Restore an Old Photo with AI
- How to Batch Upscale a Folder of Images with Topaz Photo AI
- How to Add Detail to AI Art with Magnific's High Creativity Mode
- How to Relight a Flat Product Photo Using Magnific Relight
- How to Batch-Edit Product Photos for Shopify with Photoroom
- How to Choose Between Topaz Photo AI and Gigapixel for Heavy Enlargements
- How to Add a Realistic Shadow to a Cut-Out Product in Photoroom
- How to Batch Edit Photos with AI Presets
- How to Fix a Blurry Face in a Portrait Across Topaz and Magnific
- How to Make a Logo in Canva from a Text Prompt
- How to Create a Logo with Readable Text in Ideogram
- How to Design a Clickable YouTube Thumbnail in Canva
- How to Remove an Image Background in Canva
- How to Resize One Graphic for Every Platform in Canva
- How to Build an Instagram Carousel in Canva
- How to Set Up a Brand Kit in Canva for Consistent Graphics
- How to Make a Social Quote Graphic in Ideogram
- How to Design a Blog Featured Image in Canva
- How to Create Podcast Cover Art with Ideogram and Canva
- How to Upscale an AI Image for Print Quality
- How to Write Image Prompts That Get Usable Results
How to Inpaint and Fix Part of an Image in ComfyUI
Use the mask editor and an inpaint workflow to repaint a specific region, like fixing a hand or removing an object, without touching the rest.
Inpainting lets you change only a masked part of an image while keeping everything else. It is how you fix a broken hand, swap a background object, or remove an unwanted element. ComfyUI has a built-in mask editor plus dedicated inpaint nodes that make this reliable.
What you need
- An image to edit (generated or imported)
- A checkpoint; an inpainting-specialized checkpoint works best but a normal one is fine
- A few minutes to paint the mask
Step 1: Load the image
Add a Load Image node and select your file, or drag the image straight onto the node. If you want to edit something you just generated, save it first, then load it back in.
Step 2: Paint the mask
Right-click the Load Image node and choose Open in MaskEditor. Paint over the region you want to change with the brush, adjust the brush size as needed, then click Save to node. The painted area becomes the mask.
Step 3: Encode for inpainting
Add a VAE Encode (for Inpainting) node. Feed it the image, the mask, and your VAE. This produces a latent that tells the sampler which pixels are locked and which are free to change.
Step 4: Prompt for only the masked area
Your positive prompt should describe what you want in the masked region, in the context of the whole scene. To remove an object, describe what should be there instead, for example the wall or sky behind it. Connect the encoded latent into a KSampler.
Step 5: Decode, check, repeat
Run a VAE Decode and Save Image. If the seam shows or the fix is not quite right, change the seed or grow the mask slightly and queue again. Inpainting is iterative, so expect a few attempts.
a clean red brick wall, even lighting, matching the surrounding textureResult: a targeted edit that blends with the original image. The unmasked pixels stay byte-for-byte the same, so only your chosen region changes.
Watch related tutorials
26:00
17:00
13:00
13:20
11:05
14:18New guides in your inbox
Fresh step-by-step how-to guides as we publish them. One email a week, no more.