SupremeFurry - The ultimate enhancer for furry content generated by PonyXL!

Let's face it, it's hard to make Pony behave sometimes. This model has immense potential, yet it can be challenging to get what you want and get it just right. Sometimes the colours just are too washed out, sometimes it's the contrast or the edges that are weird, or maybe you're just unsatisfied with the prompt in general...

Introducing SupremeFurry, a LoRA that has been measured to improve the majority of generations to fit the following requirements (as for v1_e6) :

• Better colours (+++)

• Better contrast (+++)

• Better poses / Less boring images (++)

• Better edges (++)

• Boost in cohesiveness (+)

Furthermore, this LoRA has been trained to be more versatile than ever, with an emphasis on working with other LoRA and LyCoris so that you can use your favorite character or style models alongside SupremeFurry.

How to properly weight SupremeFurry

From extended testing, SupremeFurry seems to function differently based on the weight of the LoRA.

Low weights (Under 0.4) can be useful when you want minor improvements and are already quite satisfied with the original generation(s).

Medium to high weights (0.4 to 0.8) are where the magic happens, depending on your prompt. The results may drastically change from time to time, but on average there's an increase in overall quality, colours, contrast, edges and so on.

Very high weights (Above 0.8) produce unstable results. Sometimes it results in a very different image, which might be better or worse. It is advised to not go above 0.8 in general unless you have special needs and/or know what you're doing.

What SupremeFurry is & aims to be

SupremeFurry isn't a magic "fixes everything" LoRA. You will sometimes get unsatisfactory results, as with any other LoRA and/or models. What we aim to do here is to increase on average the quality of the images generated. This feeling of "having a better image" is subjective to the user, but objectively, SupremeFurry is surprising in how it can turn a boring prompt around into something much better.

This LoRA is a fresh concept that I've been meaning to work on for some time now. It's still in its early stages and there WILL be upgrades to come. Training locally can take a while but I have the time and will to do so.