Hey everyone!
This is my first try at merging the model. Please be gentle, but let me know what you think. :)
Recently, I discovered Illustrious, which has opened a new door for me in terms of creativity.
I don't have any recommendations for key settings, but I usually use the next option:
CFG - 6.0
Steps - 60
Sampler - dpmpp_2m
Scheduler - Karras
To make the primary image more realistic, use a latent upscaler with denoise level = 0.6
Merged checkpoints:
Merged loras:
Midjourney mimic - (0.6)
Darker Than Dark - (0.35)
Penis Size Slider - (-0.5) (It's weird, but it gives a sweet effect to the style)
The model's default drawing style is semi-realistic, and it's quite creative. However, I think it's very sensitive to quality tags (realistic, 2D, 3D, anime). I'm not sure if that's a positive or a negative. And it's suitable for creating any type of content (SFW/NSFW).
I would appreciate feedback, posting images, criticism, praise, etc.
All of the images in the preview are created using ComfyUI in my workflow. And without any loras.
Description
New version of Rillusm · Realistic Illustrious has merged. Weights are changed.
FAQ
Comments (2)
The first one was much more enjoyable than the second one. The newer model just doesn't have much of its own style unless strongly guided, and there are too many things it doesn't know very well compared to the first one. The first release was enjoyable and fairly flexible by comparison, and I enjoyed using it. I hope you'll continue developing it though, because there's plenty of potential there!
Thank you for your feedback. I'll take note as I further develop the model.