➕ BAS+ Better Anime Style Plus Model for Anima ➕
BAS+ (Better Anime Style Plus Model) is a merged checkpoint based on "Anima Base v1.0" and "Anima Turbo LoRA", designed to generate high-quality modern anime-style images. Since "Anima Turbo LoRA" is merged into this checkpoint, there is no need to load it separately using a "Load LoRA" node.
Effect of Different Step Counts
(8 steps is recommended.)
Effect of Different CFG Values
(CFG 1 is recommended.)
Recommended Settings
CFG Scale: 1 (IMPORTANT!)
Steps: 8
Sampler: er_sde or euler
Use the following prompts to make images look more like modern anime.
anime, anime_coloring, anime_screenshotWhat's New?
Unlike the previous BAS model, "Anima Turbo LoRA" is included in BAS+.
Previously, users needed to load Anima Turbo LoRA separately. With BAS+, there is no need to load it through a "Load LoRA node".
Improvements
Produces a stronger anime-style look than the previous BAS model.
Thanks to the integration of "Anima Turbo LoRA", BAS+ can generate high-quality images with fewer sampling steps.
Optimized for fast generation while maintaining strong anime aesthetics.
Differences from the Previous BAS
The previous BAS model offers greater flexibility and tends to follow the base characteristics of Anima Base v1.0 and your prompt instructions more closely.
Depending on your goals and preferred art direction, the previous version may still produce better results in some cases.
X ( Follow me on X for updates on what I'm working on!🙇 )
License
This model follows the licenses of "Anima Base v1.0" and "Anima Turbo LoRA".
Please refer to the original license terms of each model for detailed usage rights and restrictions.
Description
FAQ
Comments (15)
What's the difference between this one and the one released yesterday?
Thanks so much!🙇
For some reason, there was an issue where the model wasn't showing up in the search results, so I re-uploaded it.
I’m not sure why, but the re-uploaded version is now appearing in the search results. Maybe I accidentally got something wrong in the upload settings...
@news_v_com If it happens again, check whether the first image in your showcase is still being analyzed. This happened to me before, and I fixed it by reuploading the image. I think the whole upload process (for loras or checkpoints) gets held back to be shown on profile or explore page because of it. Nice checkpoint btw!
@ZaeY Thank you for your kind comment!
That’s reassuring to hear.
I'll keep an eye on the status of the first image. Thanks again! 🙏
I'm surprised no one's done a turbo merge before I think, it makes a better model I think absolutely.
Thanks for your comment!🙇
Yes, Turbo LoRA improves images surprisingly well!
The first time I heard about merging Turbo into one of the previous models, it was mind-blowing.
Gotta say, it's an interesting concept to merge the turbo LORA, and welp, it does a great job tbh, nice model and results too, crisp results at high resolutions and what's better? Slashing down the generation times by a significant margin!
Congrats, you're a real one.
Thank you for your comment!🙇
I'm really happy to hear that!🙏
Very nice model, believe or not I can generate an image of 1152x896 in 1 minute and 22 seconds with a GTX 970
Hey guys, could you please check my workflow? The lighting and shadows on the images generated directly from the model aren't as good as the ones in the demo.
大佬能看看工作流吗？模型直出图的光影没有演示的好啊
quick question, this still works for custom lora styles?