UwazumiMix_Anima

( Not my Model )


 <a href="https://civitai.red/models/2677866/uwazumimixanima?modelVersionId=3262727" rel="noopener noreferrer nofollow" target="_blank">
 https://civitai.red/models/2677866/uwazumimixanima?modelVersionId=3262727
 </a>




 About this version


 I guess it leans a little more towards realism now ...




 
 Steps: 30
 


 
 CFG: 3
 


 
 Sampler: Euler a

UwazumiMix_Anima_v3.0

( Not my Model )


 <a href="https://civitai.red/models/2677866/uwazumimixanima?modelVersionId=3036058" rel="noopener noreferrer nofollow" target="_blank">
 https://civitai.red/models/2677866/uwazumimixanima?modelVersionId=3036058
 </a>




 About this version


 ( Google Translate )


 I don't think it has changed all that much.


 It feels a bit sharper and perhaps leans slightly more towards a Western aesthetic.


 It operates much like the SDXL family—you can use it with the same workflow as Pony or Illustrious.


 Personally, I think "Anima" feels pretty good.


 It’s been a while since I’ve had fun experimenting with prompt engineering.


 Above all, it’s lightweight.


 I think it could well become a mainstream choice for low-VRAM setups.


 I’m not really looking for high-art aesthetics or spectacular VFX; as long as models like "zit" or "F2K" deliver solid NSFW performance, I’m happy.


 I’d love to see a high-performance, uncensored model that runs on standard VRAM (without distillation) and offers decent generation speeds.


 I wonder if someone might create one.




 I decided to try out Anima's photorealistic-style model to see what it was like.


 As always, I mixed together various high-quality models created by others and distilled the best elements from them.


 Because I’ve been tinkering with it so freely, I don't have a full grasp of the specific source materials involved.


 Please enjoy using this model for personal use only.


 The creator assumes no responsibility for any issues arising from the use of this model.


 Lately, I’ve been wishing for a model that combines SDXL-class size and image quality with the kind of prompt comprehension found in modern text encoders.




 It may not match Zit or F2K in terms of realism, but it seems capable of producing results on par with Pony or Illustrious.


 Anima has better prompt comprehension, and it’s fun to see poses and compositional patterns that feel different from those of Pony or Illustrious.




 Also, I prefer wearing my tops untucked, and since Anima seems particularly strong at rendering illustrative styles, I find it great that prompts for that look are relatively easy to get right.


 However—perhaps a side effect of being a merged model—there isn't much facial variety. It feels like I have to rely on LoRAs for that aspect.




 I think it's great that models which evolved from illustration-style to photorealistic ones excel at NSFW content.


 I look forward to seeing what comes next.