

Diffusionmat
Overview :
DiffusionMat is a novel image matting framework that uses a diffusion model to transform rough to fine alpha matting. Unlike traditional methods, our approach treats image matting as a gradual learning process, starting from adding noise to the trimmed map and iteratively denoising through a pre-trained diffusion model, gradually guiding the prediction towards a clean alpha matting. A key innovation in our framework is a correction module, which adjusts the output in each denoising step to ensure that the final result aligns with the structure of the input image. We also introduce Alpha Reliability Propagation, a novel technique aimed at maximizing the utility of available guidance by selectively enhancing the alpha information in the trimmed map regions with confidence, thus simplifying the correction task. To train the correction module, we have designed a specific loss function to target the accuracy of alpha matting edges and the consistency of opaque and transparent areas. We have evaluated our model on several image matting benchmarks, and the results show that DiffusionMat always outperforms existing methods.
Target Users :
Used for image matting
Features
Transforms the alpha matting process from rough to fine using the diffusion model
Gradually guides the prediction towards clean alpha matting
The correction module adjusts the output to ensure that the final result is consistent with the structure of the input image
Alpha Reliability Propagation technique to maximize the utility of available guidance, thereby simplifying the correction task
A dedicated loss function targeting the accuracy of the alpha matting edges and the consistency of the opaque and transparent areas
Featured AI Tools

Remove Background Webgpu
remove-background-webgpu is a browser-based mini-program that utilizes WebGPU technology to achieve fast image background removal. It allows users to quickly obtain images without backgrounds without downloading any additional software.
AI Image Editing
227.1K

Stable Fast 3D
Stable Fast 3D (SF3D) is a large reconstruction model based on TripoSR that can create textured UV-mapped 3D mesh assets from a single object image. The model is highly trained and can produce a 3D model in less than a second, offering a low polygon count along with UV mapping and texture processing, making it easier to use the model in downstream applications such as game engines or rendering tasks. Additionally, the model predicts material parameters (roughness, metallic) for each object, enhancing reflective behaviors during rendering. SF3D is ideal for fields that require rapid 3D modeling, such as game development and visual effects production.
AI Image Generation
130.3K