IOPaint
An open-source local image processing tool built on image inpainting models. It supports watermark removal and batch processing, can run on CPU or GPU, and requires no network upload.
Timeline (last 90 days)
IOPaint is an open-source project that can remove watermarks from images in batches, and all processing is done locally, with images not uploaded to the internet.
IOPaint's models can run on a CPU or be accelerated with a GPU, and GPU mode only supports Nvidia cards.
IOPaint does not simply erase watermarks; instead, it lets the AI infer the background covered by the watermark, and the restoration quality depends on the model's ability to infer the image's semantics and textures.
When deploying IOPaint, the extraction directory must not contain Chinese characters in its path, otherwise it may cause environment errors.
When installing IOPaint, double-clicking the Install script automatically initializes the environment and installs the dependency packages, after which double-clicking Win Start launches the watermark-removal interface.
In IOPaint's single-image processing, after the user paints over the watermark region with the brush and clicks the eraser, the model repairs the selected region, and upon completion the mask and the processed image can be saved together.
The mask file downloaded by IOPaint is named MASK.JPG by default, and it needs to be copied to the software's main directory and renamed to MASK.PNG for subsequent batch processing to call it correctly.
During IOPaint batch processing, the same mask is applied to all images, so the watermark position and shape must be highly consistent across every image.
IOPaint batch processing requires placing all images to be processed into a designated input directory and double-clicking the "Batch Watermark Removal.bat" script, after which the resulting images are stored in the OUT directory.
IOPaint's restoration of logo-type watermarks located in solid-color, gradient or regularly textured regions is usually quite good; if the watermark covers a person's face or a complex structure, the repair traces may be more noticeable.
2 more timeline events
All Facts (12)
IOPaint is an open-source project that can remove watermarks from images in batches, and all processing is done locally, with images not uploaded to the internet.
50%UnverifiedIOPaint's models can run on a CPU or be accelerated with a GPU, and GPU mode only supports Nvidia cards.
50%UnverifiedIOPaint does not simply erase watermarks; instead, it lets the AI infer the background covered by the watermark, and the restoration quality depends on the model's ability to infer the image's semantics and textures.
50%UnverifiedWhen deploying IOPaint, the extraction directory must not contain Chinese characters in its path, otherwise it may cause environment errors.
50%UnverifiedWhen installing IOPaint, double-clicking the Install script automatically initializes the environment and installs the dependency packages, after which double-clicking Win Start launches the watermark-removal interface.
50%UnverifiedIn IOPaint's single-image processing, after the user paints over the watermark region with the brush and clicks the eraser, the model repairs the selected region, and upon completion the mask and the processed image can be saved together.
50%UnverifiedThe mask file downloaded by IOPaint is named MASK.JPG by default, and it needs to be copied to the software's main directory and renamed to MASK.PNG for subsequent batch processing to call it correctly.
50%UnverifiedDuring IOPaint batch processing, the same mask is applied to all images, so the watermark position and shape must be highly consistent across every image.
50%UnverifiedIOPaint batch processing requires placing all images to be processed into a designated input directory and double-clicking the "Batch Watermark Removal.bat" script, after which the resulting images are stored in the OUT directory.
50%UnverifiedIOPaint's restoration of logo-type watermarks located in solid-color, gradient or regularly textured regions is usually quite good; if the watermark covers a person's face or a complex structure, the repair traces may be more noticeable.
50%UnverifiedIOPaint's advantages are zero membership fees, zero network uploads and unlimited batch runs, making it suitable for creators who frequently clean up assets to deploy locally and reuse long-term.
50%UnverifiedGPU mode requires correctly installing the CUDA driver and the corresponding PyTorch version, which poses a certain configuration barrier for ordinary users; CPU mode, though slower, is compatible with almost all computers and is entirely acceptable for processing within a few dozen images.
50%