Description: This model is the result of extensive analysis and refinement based on multiple popular anime-style models, which often struggle with issues like detail loss, poor framing, and repetitive elements. To address these challenges, we’ve directly integrated a custom LoRA into the model, ensuring that generated characters are more coherent, detailed, and visually consistent.
While this model is still a work in progress and not yet perfect, it already delivers more stable and reliable results than most alternatives currently available.
Key Features:
Optimized Framing: The built-in LoRA automatically adjusts character composition, preventing unwanted cropping and ensuring all critical details are preserved.
Reduced Repetition: The model minimizes repetitive patterns, delivering more diverse and natural results.
Enhanced Anime Quality: Perfect for users seeking a highly detailed, expressive, and professional anime style, with rich textures and balanced proportions.
Included Rule 34 (R34)
Description
I have significantly stabilized the model. It now rarely confuses humans with furries or produces unwanted hybrid features. The focus has shifted toward much better fabric rendering, realistic clothing textures, and subtle, high-quality body details (skin, anatomy, proportions). The next priority is improving background generation and scene coherence.
FAQ
Comments (8)
Is it possible to make the backgrounds a little less...visually noisy, I guess? Like in the pic linked below, there's a lot of background detail that is just really excessive and distracting. If it's possible to tone it down, or anyone knows a good negative prompt that will do that, I'd appreciate it! :)
https://civitai.com/posts/26537514
I see, you make a beautiful image, the background is getting this way because the model tries to follow the color pattern of the character, I'm working to separate it.
Thank you so much for your feedback, I hope you're enjoying the checkpoint.
I respect your work here, and I love to see something that isn't outwardly obsessed with the same waifu and sacrifices the whole model for the sake of that one character. However, this model is still very biased and freely ignores whatever it doesn't want to listen to. To be called Mega cock, it is constantly churning out female anatomy, especially hips, and the best you'll get if you want a cock is usually a femboy. If you want a man who looks like he's about 20 years old and isn't a femboy, good luck. I do hope you will continue to develop the model, though, because it does have potential and I'm seriously glad to see something that isn't just focused on bigger boobs or nothing but generic moe wives.
Thank you very much for your comment, I'm glad you liked it, well the name is just a joke, I really want to make this model coherent, one that understands dynamics of light, shadow, position, proportions ...
But when I started I saw that the models make very simple mistakes, like for example the first thing I fixed was the ability of the model to make a simple couple, (it's not perfect) Most models didn't differentiate the characters, created hybrids and took a lot of repetition, after I took out conflicts of humans and others, and if you use my lora (semi realistic) you'll see that now I'm solving lighting and background.
I want to create a base model and I'm seeing in the other models are being overloading mainly with each character that exists, or as you commented exaggerating proportions, certain things have to be kept in loras to not harm the checkpoint.
So, I've been trying around this model. Has A LOT of potential.
But there are a few things I would like to point out, after 20-30 mins of playing around.
First and most important for me. Fingers. It ignores if it's a fist, open hand or anything. Tends to add more fingers, even with strong negative prompts. I know ai always struggles with fingers and hands, but I feel this one struggles a bit more.
Secondly, color palette. Always bright. I like it sometimes, but when I put 'night, low light, hard shadows' and gives me bright colors, then ruins the result.
Third, tends to add things I didn't ask. For example, I make my own ai anime project, and it's heavily inspired by DB. I used a prompt of one of the characters to see the results. Overal, it's amazing. But adds elements that I didn't ask for, like some visual effects, or flying rocks, or messes the background (night sky in this prompt).
I believe the creator of this model can do wonders. But needs polishing. Good luck and I hope to see new update soon!
Thank you very much for your comment, I'm glad you pushed the model to its limit. I'm mainly aware of the color palette that bothers me, I'm working to improve these details and even in my new lora (semi realistic) some of these errors were corrected but I need to be more precise and specific when updating the base model so as not to generate conflicts or unwanted limits.
About adding unwanted things is a conflict that happened when I tried to resolve other conflicts.
I haven't tried to fix the hands yet, I'll have to experiment if it's better to correct in the model or leave a separate lora.
Thank you very much for your suggestions. I'll keep them in mind for the next updates.
@Lacike Take your time. Rushing will lean you nowhere. I am still working with the model to find anything that needs your attention. So far, it's great. Doing much more that I would expect. To be honest, I do love models that don't need loras or vae. My 2070 is not very happy when I put more stuff on it to work with 😂
But it's on you. I believe you can add everything and make a REALLY good, all-rounder model without anything extra.
Hope I helped, even with this much feedback. Hope to see an update soon! (No rush though)
Nice cock!