Highly recommend using with my positive and negative embeddings, Lazy Embeddings, for your prompts, so you don't have to enter them both! https://civitai.com/models/1302719/lazypos-and-lazyneg-embeddings-illustriousxl-noobai-models
Please like and share this model with friends! ♥️ Would love to see your creations as well please tag the model! If you like this model, please consider donating at https://ko-fi.com/janxd or tipping buzz! Thank you for your support!
v2.1 released!
Red eyes fixed.
Better eyes. I did not use detailer to fix any eyes in my sample post.
Less anime style, in order to improve body of characters, but can be adjust with prompting.
My goal is the model is always first try generation without detailer and extra loras! This version puts it at that for me for all the samples I tried.
v2 released! Changes below:
Improved hands. I noticed issue with hands with my new Carried sex LORA, so I research how to adjust it.
More lewd/nsfw. The checkpoint like to remove clothing and make the character position themself in a lewd way.
Improved coloring. Colors are bright and less dull.
More details for background and character. There is some issues with it adding too much details like tattoo and symbols compared to v1.0. I will play around the merge blocks more to see if it can be fixed. Recommend putting symbol in negative prompt to help.
I have tested with my LORAs and no issue!
This is a mix of my favorite illustrious and NoobAI checkpoints along with the LORAs I always use in my generation. This will help make my generation easier to follow when importing into ComfyUI.
Here is what the checkpoint offers:
No need for LORAs to follow my generation unless it is a style/concept.
NoobAI checkpoint has more updated knowledge, new outfits on existing characters.
Way more NSFW ;) More positions which means better anatomy (messed up bodies) and better framing of the male.
No more stability issues.
More details in background and on character.
Better anatomy you can see the differences if you use my arch back sex and kissing lora.
No need to upscale can go generate higher resolution. I don't use upscalers in any of my pictures. I use 1024 x 1536 on all my images and that's how I post them.
Less issues with eyes, though you can use my LORA to fix it if the image is NSFW. You can run Adetailerr after too, but I rarely do this.
Responses more strongly for prompts. This could be a negative for some.
Recommended settings (see my images below for settings I'm testing!)
steps: 25
cfg: 3
sampler: euler a
Or you can import ANY images into ComfyUI it will also import the workflow I use. All my images have my prompt and settings. I'm playing around with euler a, but I prefer res_multistep for NSFW.
VAE baked in!
Please provide feedback and any issues that you encounter. I will fix it all as I will use this model from now. I will continue to make adjustments.