Recently made this guide for the Perchance Hub Learn articles to document the experiments that I did for making prompts with the text-to-image-plugin.
It discusses the options/terms in the plugin as well as how to effectively use them.
There are also some tricks from making prompts with the syntax from AUTOMATIC1111 which works in the model in Perchance (tag emphasis).
Each topic also has a generator for you to play around with. It looks like this:
Here is the generator…
Feel free to give feedback and if you have other suggestions or tricks that you found with the plugin, please let me know and I’ll add them to the guide!
P.S. Also need feedback on the Table of Contents as I might add them to the articles on the Perchance Hub Learn articles!
Amazing! Great job on this.
RE this point:
Unfortunately, we cannot de-emphasize tags.
I just tried this:
bald man, hiking, forest, small village in background, (army tank:0.1), morning sunrise
And I went through about 30 images, and the only ones with prominent/obvious tanks in them were the
(army tank:1)images, so I’m pretty sure it does work?(btw, I really love how you’re building up a bunch of “infrastructure” that you use across your gens, like callout, mermaid graph, etc.)
Might actually be an error on my end 😅, tested with a clean prompt and negative prompt and the de-emphasizing was working, albeit subtle on some tags. Here are some results of my recent testing (will add to the guide).
Army Tank Prompt - Army Tank Prompt, 0.9 added a small army tank, 1 added the tank fully.
Flowers Prompt - Changed ‘army tank’ to ‘flowers’ - 0.7 below seems the same but 1 to 0.7 subtly reduced the flowers.
Goth Aesthetic Prompt - Goth Aesthetic Prompt w/ Negative Prompts, some significant de-emphasis seen here.
Hippie Aesthetic Prompt - Hippie Aesthetic Prompt w/ Neg Prompt, less significant de-emphasis seen here.
About the ‘infrastructure’, I just like having a consistent layout/look xD, but thanks!
