table Diffusion 2.0 is here and it bring big improvements and amazing new features.
* New Text-to-Image Diffusion Models using a new OpenCLIP text encoder will bring better results that are closer to the prompt and are trained on an aesthetic subset of the LAION-5B. It can also have a base Resolution of 768 x 768 px.
* The new Super-resolution Upscaler Diffusion will bring more Quality to upscales, but more GPU efficient and upscale to 2048 x 2048 px.
*The Models Depth-to-Image Diffusion Model will create depth maps that are important for coherent output, so you can change the prompt, but the composition of the image and the position of the elements in the image will stay the same.
*The Inpainting Diffusion Model will bring improvements to to inpaiting, but also to outpainting, with the Method i showed you here:
#### Stabel Diffusion 2.0 LINKS ####
Announcement:
Github:
Online Demo:
Additional Download links:
#### Follow me and join my AI Group ####
Daily AI Insta Posts:
AI Facebook Group:
My AI Discord:
Support my Channel:
Subscribe to my Newsletter for FREE: My Newsletter:
How to get started with Midjourney:
Midjourney Settings explained:
Best Midjourney Resources:
Make better Midjourney Prompts:
My Affinity Photo Creative Packs:
My Patreon Page:
All my Social Media Accounts:
source
