- Introducing SCEPTER v1, supporting customized image edit tasks! Simply provide 10 image pairs, SCEPTER will tune an edit tuner for your own Image-to-Image tasks, like Clay Style, De-Text, Segmentation, etc. - Compatible with tuner models (LoRA) from community.
0.0.5
What's Changed * Support [StyleBooth](https://ali-vilab.github.io/stylebooth-page/) demo on SCEPTER Studio for Text-Based Style Editing * Support tuner model import from **modelscope/local** * Support tuner model **export to local**/**share to modelscope** * Optimize the Data Management module, which currently supports batch preprocessing of images and automatic annotation through language models such as **BlipImageBase**, **QWVL**, **QWVLQuantize**
What's Changed * Enhanced interaction design for the training module * New task management capabilities that allow for asynchronous task submission and termination * New model management module that supports saving of custom models * Support `outpainting`, `inpainting` and `virtual try on` applications by integrating [LAR-Gen](https://arxiv.org/abs/2403.19534), a novel method that enables inpainting with text and subject guidance simultaneously