Image Retouching
Upload an image to start the retouching process.
How AI Retouch (Inpainting) Works
Retouching Result
Original
Retouched Result
Perfect photos, no distractions
An unexpected tourist? An object ruining the composition? Our AI removes unwanted elements and flawlessly reconstructs the background, as if they were never there.
From power lines to people, get back the photo you envisioned with just a few clicks.
Slide to compare before and after
Recover the original sharpness of your memories
The passage of time can leave its mark on your most cherished photographs. With our AI, you can digitally erase the ravages of time and rediscover the original clarity and emotion.
Slide to compare before and after
Enhance your natural beauty instantly
Forget about imperfections. Our AI smooths skin, removes blemishes or acne, and subtly reduces wrinkles, preserving your natural look for flawless portraits.
Get professional selfies and portraits effortlessly. Try our tool and be amazed by the results!
Slide to see the retouching magic
Bring back the splendor to your photos from the past
Do scratches, stains, or tears mar your precious memories? Our artificial intelligence removes this damage and reconstructs lost areas with incredible detail.
Transform deteriorated old photos into sharp, vibrant images. It's like traveling back in time!
Slide to see the change
Frequently Asked Questions
Quick answers about our Inpainting AI.
Image inpainting is a technique that uses advanced Artificial Intelligence to reconstruct missing or unwanted parts of an image. Our system uses the LaMa (Large Mask Inpainting) model, renowned for its ability to fill large and complex areas coherently and realistically.
LaMa is very versatile. You can use it to:
- Remove unwanted objects (people, posts, trash).
- Remove watermarks or text.
- Repair old or damaged photos (scratches, stains).
- Fill large areas that need to be reconstructed.
✨ LaMa is especially good with large masks and periodic structures (like patterns or textures).
LaMa uses Deep Convolutional Neural Networks (Deep Learning) and an innovative technique called Fast Fourier Convolutions. It has been trained on millions of images to learn to 'understand' the context of an image and predict how to fill in missing areas so they look natural and consistent with the rest of the photo.
💡 Fast Fourier Convolutions allow LaMa to have a global receptive field, meaning it considers the entire image to generate high-quality fills, even in large areas.
Yes! One of LaMa's great advantages is its resolution robustness. Although it was trained on moderately sized images (e.g., 256x256), it generalizes surprisingly well to much higher resolutions (up to ~2K or more!). This means you can expect good results even if your images are large.
🚀 This capability sets it apart from many other inpainting models that may struggle with large images.
You can upload your original image in common formats like PNG, JPG/JPEG, WebP.
To indicate the area to be filled, you will need to provide a mask. This is a black and white image where the white area (or sometimes black, depending on the implementation) represents the region that LaMa should fill. Typically, our tool will allow you to draw this mask directly on your image.
🎨 Tip: Precise masks help get the best results, but LaMa is quite forgiving!
Absolutely. Your privacy is our top priority. The images you upload are used solely for the inpainting process and are handled with strict security measures.
🔒 We do not permanently store your images or share them with third parties without your consent. See our Privacy Policy for more details.