These loras were trained on images from Megami Magazine from late 2018 to 2026. Anima already knows the Megami Magazine tag but it's riddled with japanese text. These loras were trained with the text manually removed. Please read the About This Version for more information.
To get the results you see in the images above, these are the settings I used.
Visit My Site For Models I Can't Post On CivitAI: https://qtgurl.com/
The loras for Anima V1 were trained at 1792 resolution. So they're capable of generating at around 1792x1792 or higher.(I use 1600x1984) The P2 and P3 loras were trained at 1536 resolution, so they're capable of generating around 1536x1536. (I use 1408x1792 for most gallery images.) All of the loras are capable of generating smaller sizes just fine.
MeMaAni - V2+B
Trained on 51-68 (mostly) new and old images. Dataset was re-collected and edited for the best possible quality.
MeMaAni - V1
Trained on the same data used in Preview 3.
MeMaAni - Anima Preview 3
Trained on the same data used in Preview 2 with slight changes to improve results. Most of the improvements come from the Anima P3.
MeMaAni - Anima Preview 2
Trained on the same data used in 2d Gold Fish. I guess you could call it MeMa F, but I suppose that would make it A for this model. It uses a lot of images from the A dataset so it's close enough.
Comment On Anima
The final release of Anima 1.0 is fantastic. The model has so much capability. If you haven't made the switch, now is the time. It doesn't matter if all you post is one girl, the jump in quality, prompt comprehension, 16 channel VAE, newer anime knowledge and just raw stable performance makes it the worth while model to switch to as an anime genner.
Description
This version was trained on 51 images at 1792 from the brand new B dataset. Well, most of the images are from the MeMaXL D dataset but they were all re-gathered and edited to delivery the highest quality.
The goal of this version was to stabilize the eyes. There wasn’t much attention placed on the background this time around. More than half of the images were white background or pattern background. If you take this version and the first V2, set their weights to 50% and use them together on the same image, you’ll get the best of both worlds.
FAQ
Comments (4)
Hi, i didnt understand very well this part "If you take this version and the first V2 and mix them at 50%, you’ll get the best of both worlds." Do you mean as using both loras in the same image or mixing the loras altogether?
I mean using both lora at 50% strength on a single image.
Tiny bit confused, is this lora V2 mixed with B already or should I mix it myself?
It's B, not mixed. It's trained on a different set of images than the first V2. I kept the version number at 2 because it uses the same settings and approach as the first version 2. I only change version numbers when I feel the settings and approach warrant a distinction.