■This is an experimental ground for Anima LoRAs.
The LoRAs I create serve more of a role in stylistic tuning rather than reinforcing single concepts.
■I would like to share the possibilities of anima with everyone.
My wish is for many people to discover basemodels with potential and to see their possibilities unfold even further. I would be happy if I can help make that happen.
■The anima architecture has been heavily modified and retrained from scratch—almost like a radical overhaul—but the base architecture is Cosmos-Predict2-2B. To put it simply, you could compare it to a lightweight version of Flux 1.
I think it’s a great architecture capable of generating high-quality images despite its small size.
I have a very good impression of Anima when it comes to color vibrancy, lighting, and prompt adherence.
Its knowledge regarding NSFW concepts, characters, and styles is at least on par with SDXL. The overall feel of using it is also quite similar to SDXL.
Furthermore, it is lightweight and easy to train. I believe large-scale training is also a viable option.
Since we are provided with a base model that already contains a vast amount of knowledge, we rarely need to teach the model completely unknown concepts from scratch.
Basically, our work mostly comes down to adjusting styles or reinforcing minor concepts and characters. You can achieve great results without having to do any heavy lifting.
The developer has also ensured transparency by sharing the training settings, which means you are less likely to stumble over black-box issues.
Because of this, the burden of training is low, making it an option well worth nurturing together as a community.
■The LoRA has a stronger influence when used with shorter prompts.
Since longer prompts often carry their own inherent style, the LoRA will primarily take on a supporting role to refine and unify the overall image.
■https://github.com/gazingstars123/Anima-Standalone-Trainer
This is a great tool that allows for easy training, even on Windows. It has everything you need for LoRA training.
Even if you run into any issues, you should be able to get your LoRA training up and running by troubleshooting with ChatGPT or Gemini.
If you find the tool useful, you might want to consider making a donation to him. It allows the developer to focus more on development and deliver even better tools, which ultimately benefits you in the long run.
■If you have any questions, please feel free to ask!
日本語での質問も大丈夫ですのでご気軽にお声がけください～
Description
This is an aesthetic LoRA for Anima-Base-v1.0, trained on 1,618 aesthetic 2D-style images.
Style Tendencies:
This LoRA has a strong tendency to generate a flat, low-contrast 2D style. The dataset itself heavily leans toward this aesthetic, and I deliberately excluded any 3D, volumetric, or realistic images. As a result, you will often notice a distinct hand-drawn vibe in the generations.
Backup Versions (Different Steps):
While the main version is the 1,500-step LoRA, I have also included a few backup versions trained up to 3,250 steps. These backups are trained further, which significantly shifts the style—resulting in a much cleaner and more subdued/calm aesthetic. If you find you aren't a fan of the main LoRA's style, I highly recommend giving these a try.
The step count is low, but with an effective batch size of 32, it's actually trained quite a bit.
Dataset Size:
The reason the dataset is relatively small this time is that I had a very specific 2D style goal I wanted to achieve, so I intentionally curated only images with similar stylistic tendencies. Normally, my goal is to train the model on a wide variety of aesthetic styles, which explains the difference in dataset size compared to my usual releases.
You can find the workflow either inside loras.zip or inside anima_workflow.zip located under the anime_aesthetic_v2_pre3 tab.
FAQ
Comments (7)
The workflow attached to the example image looks like some kind of cult summoning ritual
Sorry about that! I always tend to mix a bunch of things together, so my workflows usually get pretty messy lol.
Also, just a heads-up: the workflow embedded in the images uses refactored versions of old custom nodes, so you might not be able to recreate it.
You can find the organized workflows either inside loras.zip or in anima_workflow.zip under the anime_aesthetic_v2_pre3 tab. I'd appreciate it if you could check those out.
The simple workflow doesn't use any custom nodes, so it should be really easy to use as a reference.
I highly recommend the "tipo" workflow that uses wildcards. However, the nodes used in it are getting outdated and tend to cause issues with the latest ComfyUI updates, making it harder to recreate the environment.
Because of that, I'm currently in the middle of refactoring my nodes.
I'm also considering sharing a self-contained ComfyUI portable build in the future so everyone can just run it out of the box.Well, I don't know if things will actually go that smoothly, but let's hope for the best.
@hjhf thank you for your work! very good lora btw
@CH4AcKO3 Thanks for trying it out! I'm happy to hear it was useful for you.
On your newest example images, about half of them the prompts don't match the images at all. Maybe something went wrong carrying the workflow over from comfy?
Ah, sorry about that! I made a human error and pasted different prompts for each image, but I've fixed it now.
@hjhf thanks! Keep up the amazing work!