Just yesterday, Stability AI released the version 2.0 of Stable Diffusion and it went online. Players exclaimed that the V1 version has not been fully understood, and the V2 version is coming. In fact, after Stable Diffusion 1.0 was open-sourced, many functions of Pickup Smart AI were upgraded and optimized based on this model.
As far as image restoration is concerned, Pickup Intelligent AI has added a new encoding model training to Stable Diffusion 1.0, which can increase the resolution of generated images from 2 times to 4 times, which is comparable to the resolution of Stable Diffusion 2.0 , or even higher.
Long before the release of Stable Diffusion 2.0, Pickup Smart's technology has been able to increase the 640x640 pixel image by more than 4 times, and users can go to the official website to test it.
From the point of view of text generation images, Pickup Intelligent AI has greatly improved the quality of image generation based on the V1 version model, and can generate a default resolution of 512x512 pixels. At present, the "AI painting" function on the official website of Pickup Smart is to make the C-end user experience better and generate faster, only the default resolution is 512x512. If you access the API, you can set 768x768, 1080x1080, or even higher resolution images.
In addition, Pickup Smart has also expanded the image-image feature in the V1 version, providing a new possibility for creative applications. The imagination doesn't stop there.
It is worth mentioning that since the release of the first generation of Stable Diffusion, the domestic R&D team of Pickup Smart has been working hard to optimize the model, hoping that as many people as possible can use it without code. The Stable Diffusion model can not only make the image clear, but also make the video high-definition. The follow-up of Pickup Smart will soon launch the "video to high-definition" function on the official website, which can quadruple the resolution, so stay tuned!