💖 My STUDIO1911A2 Workflow is available to generate on Tensor.art
https://tensor.art/template/941106256993613248
This baby (ting-ting!) is a two pass, hand-tuned, Illustrious-based workflow, my attempt at a "zero-shot gatcha" to gen FOUR (count em, FOUR!) 1024x1024 hentai images, with accurate anatomy and body posture! The more hentai you add, the more freaky morphs you get 🙀
🥇 A2Detailer adds "ADETAIL" nodes from ComfyUI Impact! It also includes (beta) img2img input plus detailed guides on dynamic prompting and use! After the initial generation image is saved, these new nodes take the finished image and isolate the faces, hands, and feet to regen and repair! I've also includes "mask sights" which will show the white BBOX selection during the repair, so that you can visually inspect if the models are working together! Sometimes there is interference, and you may need change out your BBOX model! Many thanks to the community members who helped me test and refine this workflow!
🥈 A2 Master versions are released and far superior! I have finished the final post-release polish and testing, and I have added extra example prompts, generation seeds, engineering and technical article, and example images to the be included with either the .zip file or intrisically listed in the JSON workflows.
These Master editions use both Wai-Illustrious v. 14 and 15! LoRAs and noise buffering! Upscaling! The Zip includes text READMEs, Prompt Guide, and JSON workflow. README includes default KSampler settings, enhanced prompt examples, and more readable labels and instructions. For A2 Anime Akimbo, Ksamplers euler_ancestral and normal schedule is preferred for the first pass and Dpmpp_2s_ancestral or Euler_ancestral with simple or karras scheduling is preferred for the second pass. 😘
Use A2 Master Bounty for 2k accuracy.
Use A2 Master Big Game for 2k, 4k, and 8k.
Use A2 Master Machine Gun for 1024x1024 rapid fire!
🥉 A1 is still a great and straightforward one checkpoint choice! READMEs include LoRA lists, prompt samples📋
Just unzip ( the JSON), and shove it in (to your ComfyUI) 😉💖
Use the Bareback or Zero Edition if you don't want to bother with finding and downloading LoRAs! Bareback is just ComfyCore nodes and the checkpoints, with my tuning and adjustmnets. Zero is A1 with the LoRA's bypassed and zero'ed for your ease. These workflows are a bit more unstable in the anatomy though!
Here's some resources to help start ComfyUI locally! I recommend using the .exe Windows install, or the github Python source install if your tech savvy. The portable version is unstable, difficult to update, and crashes often 😭 They have a web-only service now too!
https://www.comfy.org/download
https://github.com/comfyanonymous/ComfyUI
https://docs.comfy.org/installation/desktop/windows
Here's the github for the ComfyUI Manager which is a mandatory add-on to manage and update custom nodes: https://github.com/Comfy-Org/ComfyUI-Manager
https://www.comflowy.com/preparation-for-study/install-comfyui-manager
I take no responsibility, liability, or accountability for any content or images made using this workflow. This is a completely uncensored AI tool, and the user is burdened with using it responsibly (as is intended). This includes illegal uses such as deepfakes and content that will get you banned in most places. You have been warned and advised.
Description
Adjusted for more facial accuracy, added Power LoRA for assisting with first KSampler Pass, and included output Latent upscaling to 2K.