Subscribe
Sign in
Stable Diffusion XL — Richer images & aesthetics from shorter prompts
Create descriptive images with shorter prompts and generate words within images. This is a significant advancement in image generation capabilities, offering enhanced image composition and face generation resulting in stunning visuals and realistic aesthetics.
13
Replies
Best
Chris Messina
Top Hunter
Hunter
📌
Another big drop from StableDiffusion — the compounding rate of improvement is insane.
Now you can create arbitrary text in the images you generate? What!! 🤯
Report
Share
@chrismessina I tried generating text but the results are still incorrect. Have you tried it?
Woah, can it generate text?
And love to see open source catching up with midjourney
Hope to have this on our API soon at evoke-app.com
I'm definitely very impressed with the quality of people in the images I'm generating. I love the UX and speed of the app. As with most AI so far though, I'm still a little frustrated at tweaking the prompt to get something simple corrected.
For instance, I tried creating an image of an influencer getting ready to record their video. I tried many times to get the AI to make sure the camera on the tripod is facing the influencer, but it always showed the camera facing us. The quality was really good, but feels very strange that I can't get it to make corrections like this.
I admittedly have a lot to learn about AI prompts, so maybe I'm just a noob on this!
I'm excited for it to be released as an open-source model! By the way, you can already test out SDXL on ClipDrop (created by Stability.ai). It definitely looks like a big improvement over 2.1!
We have embedded Stable Diffusion XL combined with ChatGPT tuned for prompt engineering in our app and it is a BLAST!!! … going live tomorrow: AI Genie for iOS
