Thread by @mfranz_on

Diffusion models is the new trend and everyday new enthusiasts approach these tools to make art and create awesome applications.
There is a ton of beautiful pictures made with #generativeart but have you ever tried and failed?
Me, many times😁
A thread to improve your prompts⬇️

1. Sometimes less is more. Writing hundreds of characters is not the right way to produce beautiful images. Consider that there are "classes" like 4k or high-resolution which embeds a lot of characteristics.
These are already available in @StableDiffusion or #discoart @JinaAI_

2. More detailed, more precised. If you use a specific jargon to obtain a certain style or effect, it is quite sure that you will be impressed from the result!
Try something very specific using technicalities like
infrared photo or sticker illustration, also tattoo works!

3. Just art ? NO. One of the main usage of generative models is for art. But they may be relevant in other fields like photography, design, illustration or advanced editing, like #MagicBoard.

Have you other hints or suggestions to improve prompts ? Let me know !
I want to share much more knowledge on "prompt engineering" in futures thread!
In the meanwhile happy #diffusion 🤖