My RI-MIX Style
Lora based on this model - Ri-mix - Style LORA
Settings for CIVITAI On-site Generation
Workflow: Text-to-image Hi-res fix
CFG Scale: 7
Sampler: Euler a
Steps: 30-40
Denoise: 0.35-0.40
My settings for ForgeUI
Generation
Sampling method - Euler a
Schedule type - Automatic
Sampling steps - 20
Dimensions - 768 x 1152
CFG Scale - 6
Highres. fix
Upscaler - None (works for me)
Hires steps - 10
Denoising strength - 0.1
Upscale by - 1.5
ADetailer
ADetailer detector - Anzhc Face seg 640 v2 y8n.pt / face_yolov8n.pt
Inpaint denoising strength - 0.35
ADetailer steps - 20
ADetailer VAE - sdxl_vae.safetensors
ADetailer sampler - DPM++ 2M
Description
Positive and Negative prompt I'm using on ForgeUI works on on-site generation as well
POS - masterwork, masterpiece, best quality, detailed, depth of field, high detail, very aesthetic, 8k, dynamic pose, depth of field, dynamic angle, adult, aged up
NEG - lowres, worst quality, low quality, bad anatomy, bad hands, jpeg artifacts, signature, watermark, text, logo, artist name, extra digits, censored, patreon username, loli
FAQ
Comments (22)
Using the prompts of examplary images it gives me basically exactly that. I'm looking forward to experiment with this model's quite distictive styles.
Definitely one of the best models i've tried recently, don't even have to add too many loras to achieve very consistent and good result
Thanks for the buzz BTW
Very impressive. Beautifully generated characters.
Omega is currently my favorite IL model for offline use. Very, very strong and unusual compositions, good booru tag knowledge. Even when using other checkpoints for the final look, I will often use Ri-mix to generate a controlnet input for composition - it's my "concept artist" step.
Thank you
Amazing model! What pleases me the most is that she perceives direct text with some limitations! And if you write promt with tags, it generally depicts everything almost verbatim!
So, does this model actually combine base Pony and base Illustrious? Or is it just base Illustrious? Using the Pony score tag doesn't make much difference for me.
i NEED a comfyui workflow for this, including all the bits and bobs for the hi-rez fix and all please
Give this one a shot. Has a 3 pass manual hires fix for maximum quality, plus detailers for pretty much everything.
https://civitai.com/models/1911850/ultimate-comfyui-sdxlpdxlillustrious-workflow
Hi Phinjo, I really admire your work with A-Mix and Ri-Mix checkpoints. I’ve been trying them in ComfyUI, but my results (especially face, eyes, and lips) don’t look as sharp and natural as yours. I mainly use your checkpoints for concept creation and storytelling. Could you please share or explain your complete ComfyUI workflow settings so I can better understand how you achieve that high-quality outcome? Thanks a lot for your guidance.
try 0.5 denoise? i didn't even need adetailer for it, it works really well with Eular A, karras and denoise 0.5
are you comparing your results to his post detailing, face fixing + upscaling? Also be careful of your LORA's. Messed up eyes and lips are a classic sign of over cooking with too many LORA's or too high of weights. If I recall correctly his examples include no LORA's whatsoever.
@toyuyosuke are u sure? denoise 0.5 means img2img, but what I'm actually doing is txt2img, how could 0.5 denoise works?
Love this model. Assimilates LoRAs well but barely needs them with style prompts.
So, does Pony Loras work with this checkpoint?
It requires more attention and post processing than A-mix, but it's beautiful and adheres very well to prompts. The art style, imo, is the best around without LORA's. Thank you for this gem.
I've tested many different models and this one seems to be the most flexible and understands a wide variety of prompts. I especially like the style of this model and the overall image quality.
one of the best 2.5D Illustrious-compatible checkpoints. especially without excessive chinese cartoon oneme style.
*looks at top posts*
riiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiight
Great model with great features. Definitely has the ReMix feel, Very flexible. I highly recommend following the included settings. (And thank you for including Forge settings!)
Nice model. Sometimes I have to work harder to get the best results, but I liked the base style of the model. Although I haven't had much time to experiment, the initial results are quite good. 👍
Details
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.