A small peek behind the scene provided by the dev would be awesome. I guess, there is something to be learned :-)…

Yes… I tried some self hosted models (=> ComfyUI), but I miss a secret sauce. The results are far below perchance.org T2I results regarding… well… the “overall feeling”. Most of them are plain and simple bad at quality, composition and consistency. Z-Image-Turbo was the only model which performed good (to a certain extent), but the training material seems to have been heavily filtered for “all good looking and shiny things and especially people”. It’s quite incapable of generating some more down-to-earth or slightly ugly stuff. perchance.org seems less biased here, which results in a wider variety of “person types” you could achieve. If the devs would disclose, what they exactly do, to modify or enhance a model - that would be quite interesting.

Don’t know which model is behind T2I, but some models just do not support negative prompting at all… The solution is, to simply not use them, but to focus, what to show - instead of focusing on what not to show. I tried a girl face with a simple prompt, and could generate any number of girl faces without one earring in sight:

Just try it as a positive description… it works (at least most of the time ;-) )

I checked some offerings and options - including running my own local generators. But there is nothing remotely as easy and free/unbiased as perchance.org… In spite of the fact, that things here might change or break without prior notice. After all it’s a toy we are given for free… which might get removed at any time. So the only viable alternative is to do it yourself - in that way you are in control. But that’s hard work or expensive or both…

In that way: thanks to the Dev(s)!