ModelsLab/ V1.3 - Pruned

About this version

Hi people!
I've a lot to say about this new version, and I really hope to remember all, eventually I'll modify in the future this change note.

First of all, this is a new total training, but merged with previous version of Aniverse, you'll understand why later.

I've do the training training to push at the limit of Stable Diffusion 1.5 (and my poor PC)
This model is trained at 1024x1024 pixel (the models before was at 768x768).
It was great, but to much "artefacts" at low resolution, so I preferred to do a merge with Aniverse 1.2.

I think the result is very good, but you'll decide, I really hope that you like it.
You don't need any more use the option HiRes.Fix (until a certain resolution, I saw that at 1280x720 it still work good, you can push at 1366x768 but sometimes it give you some problems).
Usually I use 512x904 and Adetailer, after when if I got a good image, I put it in 4x Ultrasharp tool

In model description I changed a little bit the Positive and Negative prompt, so check it out :)

Question: Now... why I put in "early preview" my model?

Answer:

• Because I spent over 650 hours of training ( I don't remember if 20 or 22 days), and I hope that you can support me using my Ko-Fi page, where you can "buy" the new model. (if you have already a subscription in civitai, you can download it from here, I don't want that you pay twice)

• For give to my supporters a "plus" to try the new models before the others.

I really hope that you can understand that I'm not looking to become "rich" selling my model, but I'm trying to figurate out how to pay the electricity bill :), nothing else.

So if you don't want be a my supporter, no problem, really :), wait only a couple of days more, to download the new version.

Enjoy!