StabilityAI Launches StableFast3D Fast 3D Model Generation in Just 0.5 Seconds

TapTechNews August 2nd news, StabilityAI today launched an AI model named StableFast3D today, which focuses on 'image-to-3D model'. It is claimed that it can quickly generate a complete and natural 3D model in just 0.5 seconds. Currently, the model has been listed on the HuggingFace platform, and interested friends can (click here to visit).

TapTechNews learned that StableFast3D can quickly generate high-quality 3D models with just a single image. The related model comes with materials and corresponding parameters, and also provides accurate UV unwrapping (UVUnwrapping) for users to make later modifications. If users need, they can also choose the quadrilateral or triangular mesh reconstruction function, and the related work can be completed only by spending an additional 100 to 200 milliseconds, allowing users to control the mesh topology of the model more precisely.

The official also mentioned that StableFast3D can minimize the traces of the main body affected by the illumination in the original image as much as possible when generating the model from the image, so that the generated model can show the correct color under different ambient illuminations.

StabilityAI compared the StableFast3D model with the previous SV3D model and emphasized the speed of the StableFast3D image-generation model. It is claimed that the inference time of a single model is shortened from 10 minutes to 0.5 seconds while ensuring accuracy, which has certain practical benefits.