⚡ Anima Workflow:
🛠️ Purpose & Design Philosophy
This workflow is designed for quality and autonomy, not speed. It follows an "all-in-one" philosophy: configure your settings, hit queue, and let the workflow handle everything from initial generation to high-res detailing in a single pass.
Not for Speed: If you want rapid-fire generations, this is not the tool for you. A solid, much faster alternative created by darksidewalker can be found here.
Personal Use: This was built for my personal production needs. It is not intended to be a "one-size-fits-all" solution, but I am sharing it for those who value the same high-fidelity results. Please adjust the settings to your preferences!
Heavy Duty: Due to the multi-stage processing, this workflow can be resource-intensive. In my experience, the detailers are not usually needed. YMMV.
On v1g and after: If the results are too blurry for you after USDU, you can try using the RTX nodes after it to help.
v2 is compatible with AIO versions of Anima models. Note: You might have to select a random model in the Checkpoint Loader node even if you are not using it. This is because it references your models/checkpoints folder. The opposite may also be the case if you are using the Checkpoint Loader node but not the Model Loader in the Diffusion Model Loader group which references the models/diffusion_models folder.
🚀 Key Features
Beyond standard generation and upscaling, this workflow integrates:
Power LoRA Loader: Efficiently manage multiple LoRAs without spaghetti wires.
Global Controls: Centralized Seed, Sampler, and Scheduler nodes for a unified experience.
Bypass Control to toggle features on/off.
Visual Validation: Integrated Image Comparer nodes to see exactly how your image evolves at every stage.
Upscaling: 2-stage upscaling using standard image upscaling and Ultimate SD Upscale (optional).
Triple Detailer Groups: 3-stage detailing using standard BBOX and SEGM detection models for faces, hands, and clothes.
CivitAI Ready: Images are saved with full metadata (Model, LoRAs, Prompts) for easy site parsing.
⚠️ Disclaimer & Compatibility
Install at Your Own Risk: Updating ComfyUI or adding custom nodes can break your environment. I am not responsible for any installation issues.
Portable Version: This was built and tested on the ComfyUI Portable version. Desktop app users may require additional troubleshooting.
"Your Mileage May Vary": Your environment is almost certainly different from mine. I do not guarantee 1:1 compatibility.
Nodes 2.0: I strongly recommend disabling Nodes 2.0. It causes unpredictable behavior; I will not provide support for any issues arising from its use.
🤝 Support & Boundaries
No DMs: Direct messages are disabled due to high volume. Please use the Discussions tab below. Check previous comments first (unless there aren't any yet), as most common questions may have already been answered.
Custom Requests: I do not take private requests for custom workflows. If you need a specific solution built, please post a Bounty on CivitAI. There are many talented creators here who will be happy to assist you for a fee.
Modifications: You are free to add or remove nodes as you see fit. However, if you change the internal logic, you are responsible for your own troubleshooting.
The only place I am actively maintaining this workflow is here on civitai. If my workflows are being posted and monetized elsewhere, whoever posted them is obligated to provide support.
Description
🌸 Anime Workflow v1a: Release Notes
This update focuses on efficiency and prompt precision. I’ve integrated new nodes to give you better control over your generations without cluttering the workspace. This workflow was tested on the Anima Official Preview 2 model and AnimaIka.
🚀 Key Improvements & Features
RescaleCFG Integration: Added to help maintain image integrity and prevent "cooked" looks at higher CFG scales.
Fast Groups Bypasser (rgthree): Cleanly categorized bypassers are now included. You can now toggle entire features (like Detailers or Upscalers) on or off with a single click.
Advanced USDU Control: Added an optional logic gate for Ultimate SD Upscale.
You can now use an additional CLIP Text Encode + Conditioning (Average) for the positive prompt during the upscale phase.
Note: Check the sticky note inside the workflow for a breakdown of how this influences fine details.
Expanded Documentation: Added generalized notes regarding USDU settings to help you troubleshoot tiling or blurring issues.
⏱️ Performance Benchmarks
Avg. Generation Time: ~3 Minutes.
Hardware Tested: NVIDIA RTX 5060 Ti.
Process: This includes the full pipeline—initial latent generation, 2x upscale, and the final USDU pass.
💡 Pro-Tips
Load & Go: The settings saved in this JSON are the exact parameters used to generate the sample images.
Flexibility: If you want a faster workflow, use the new Fast Groups Bypasser to disable the USDU pass for quick drafting.
FAQ
Comments (6)
I appreciate your work as usual!
ConditioningAverage is something I hadn't considered for upscale conditioning adjustment. I previously tried ConditioningConcat but couldn't get it to work and ended up just concatenating the prompt strings lol. Will have to give it another shot after seeing your workflow.
You might be interested in this Anima Enhancer (reddit) node. I haven't tested it thoroughly yet but it seems to be decent without adding overhead. The more pertinent part for me personally is its implementation of Spectrum which cuts quality slightly but gives a noticeable speed increase. There are already a few ComfyUI implementations for Spectrum but this one has been the most customizable and stable for me so far.
I don't think Spectrum makes sense for a quality-oriented workflow but it is nice to have enabled for rapid iteration before turning it off for final generation. It can also speed up detailing but I haven't done a side-by-side there. Theoretically, the minor quality loss should be a non-factor. Combining it with SageAttention and TorchCompile brought my Anima gen times pretty close to Illustrious (without boosts). And there's no annoying setup / debugging like those two have.
Thanks, the Conditioning Average idea came about with me trying stuff and bouncing ideas off of Gemini. One of the things it suggested was to give USDU it's own prompts, but for me that generally made things worse or caused USDU to hallucinate even more.
It's been a while since I have tried sageattention and the quality loss was notable enough for me to not want to use it. This was especially annoying since it was a real pain for me to get sageattention working. Not sure if they have made it easier since then, but I'm not inclined to try it again.
Thanks for the suggestion about Anima Enhancer. I'll take a look at it, but I'm not making any promises to implement it.
Tested with Anima Enhancer added to the Initial Image and USDU. Having it on with Spectrum enabled was around 30 seconds faster at 146 seconds. Not using the node on the same settings/seed took 172 seconds. Note: using the node with Spectrum definitely affects the composition of the image giving a different result.
Further testing using the node without Spectrum enabled resulted in a less saturated image (which is a big plus to me), doesn't drastically alter the composition, and took about 181 seconds, so it is slightly slower. The differences might be worth it for some folks. I would probably keep it to use if I feel the output is too saturated or want a softer result.
Installing sageattention was relatively painless using this installer but I think it's better to leave it out of the workflow since it's likely to explode things and have everyone asking you for help. Will have to double check the quality now that you mention it! I've been using the "--use-sage-attention" launch flag and disabling it with the KJNode doesn't seem to change my output for some reason.
EDIT: The flag was forcing SageAttention to always be enabled. Apparently, TorchCompile is also affecting the composition even though it shouldn't. I can definitely see a difference now so I might disable some of these for final generation and just keep them on while iterating for huge speed gains. Thanks for mentioning that!
Spectrum does certainly change some larger details around. I've settled around w = 0.3, m = 16, lam = 0.5, warmup_steps = 6, window_size = 2 for Anima (all within the settings recommended by the node creator) and used the gains to raise my step count from 30 to 50, which in my flawed book was a net gain for the same time spent. With your focus on quality (and better GPU), I don't think it's worth it.
Regarding saturation, I just slapped on the Image Filter Adjustments node from the WAS Node Suite (which I believe is in your workflow?) before the Color Corrector node for quick/minor adjustments. Logically, it'd go after the Color Corrector node, but enabling "use_gpu" there moves the image to the GPU = kaboom. I rarely use it in practice but it's nice to have for when the Color Corrector node simply refuses to cooperate.
I used to have the Image Filter Adjustments node in an older workflow, but got enough questions/complaints about it to not keep it.
Adding your preferred nodes to any workflow is something I strongly encourage for anyone.
I was going to try torch compile but then got the error saying I needed Triton and I was like "NOPE, not going through that headache again!"
I'm pretty satisfied with the results I get now and the Anima Enhancer node is useful enough to probably add in a future update since it is just a single node.
v1b now has this implemented for the initial image only. I think it works best there for my usage. Thanks for suggesting it!