v1

I found myself pressed for time, so the release took longer than anticipated. However, the merge turned out fairly well, with a cute and detailed style that differentiates itself from its counterparts. The outcome of this merge was satisfying. The style is relatively consistent and a good nsfw model acted as the base, ensuring that the merge is solid at NSFW. I'd very much enjoy it if you post some of your runs :D

Thanks and enjoy!

Small note: The model likely lost compatability capabilities with XL loras due to v1.

v0.5 - EXPERIMENTAL MERGE

Cute anime style bordering the boundaries of XL and Pony, as models from both were used in its making. The model recipe is quite large. Using Ponymagine as the base, it is merged with LeekAI (an XL model), and a touch of Hassaku XL (Hentai). After that, ElyPonyXl and BAXL are merged with the previous checkpoint.

Issues to tackle (and v1 progress) As of the moment, v1 has been released.

• Takes on too many of XL's features, giving it a soft anime style but also low reliability, prompt adherence, and suffers with anatomy - Solved by v1 ✅️

• Does not inherent BAXL's Blue Archive knowledge ;( (might merge it with a Blue Archive Lora) - no lora merge yet ❌️

• Doesn't know too many characters - Solved somewhat by v1 (knows characters recognized by Pony)

• Struggles to use Pony Loras (Maybe also XL Loras) - Solved by v1 ✅️

• A bit too horny (Likes outputting boobas even when not explicitly prompted) - Solved by v1 ✅️ half-solved. Likes showing boobas's "peaks". Present but not too prevalent.

• The model should struggle with NSFW as it seems to be leaning XL rather than Pony. - Solved by v1 ✅️

  These issues shouldn't be too hard to solve if this model is merged with another, with that model being the base model. -

NOTE: The next version will strive to better the current model. As of now, a VAE is strongly encouraged but is not necessary. My favorite for use with this model is

SharpSpectrumVAEXL.

The note has become redundant with v1. On that note (xD), v1 seems to have some style volatility. It outputs cute images with varying styles. Might need further testing than anticipated. It might be downgraded to a v0.7 and then released. Either that or this "v1" will be scrapped in favor of a more desirable merge. But since it is a good model regardless, it could be released in some capacity soon.

Journey to v1

While some testing is needed, v1 has mostly been prepared. Expect a release soon. It uses another model as the base, making it less XL-ish while still retaining much of the style. I hope you guys enjoy the model once it is ready for release. Few people read this, so dm me and I might give you 100 buzz ;)