Welcome to Perfection Super Model!
This is an Illustrious XL merge with a bunch of hand-selected checkpoints and LoRAs to improve the base model's realistic capabilites.
And wow, does it deliver!
It's focus is on creating high quality, photorealistic characters with model-like beautiful perfect faces, with a focus on women.
Do not use the Normal scheduler. Use Simple or Beta!
Use euler_a, dpmpp_2m_sde or dpmpp_3m_sde as your sampler.
To use the faster low-step settings, you need to use the DMD2 LoRA. This allows you to use 8 or 16 steps and generate very fast. Remember to generate in 2 passes, either using "Hires fix" in Forge/A1111, or by creating and chaining a second sampler node in ComfyUI. A sample workflow if included in the sample images of the model. Use this as a reference.
Each sample/preview image contains the used workflow. Here's a quick article with simpler more beginner-friendly workflow. This is a recommended starting point.
Recommended Settings
Base Generation
Steps: 24
CFG-scale: 4
Sampler: DPM++ 3M SDE
Scheduler: Simple (beta upscale scheduler)
Upscale/Refine
Steps: 8
CFG-scale: 6
Sampler: DPM++ SDE (dpmpp_sde)
Scheduler: Beta
Denoise: 0.35
Description
FAQ
Comments (7)
Man, your models looks good. But they dont work as Illustrious models.
Booru prompt adherence sucks, you cant make simple sex pose.
Illustrious prompting has nothing to do with gpt.
I haven't had a chance to try it yet, but probably these models can be used as a refiner, and another model can be used as a base, more flexible. I don't think we can yet make a perfect model that will be good in all aspects
It's a pain that the best realistic models are no longer available directly on the site's image generation...
It's because of the new bidding system.
It's the most fair, but it sure is inconvenient!
With the buzz "economy" being down to 10% from last month, it's not realistic for me to cover the cost of putting checkpoints on the generator.
It's now entirely a community thing. People need to use the "Bid" system to contribute to their favorite models being on the generator, fair and square.
It resets Thursday nights (GMT). Which one were you hoping to have on the generator?
Thanks for all of the models you've published! What tools do you primarily use to assemble these checkpoints? Diffusers? Do you also have a changelog listed somewhere between v1 and v1.5?
I use CivitAI training or local training to train LoRAs when I'm solving specific bugs in the models, which I then merge into the checkpoints using ComfyUI. Just basic nodes.
Between version 1 and 1.5 here I don't really have a change log. I trained a LoRA whihch adds details and noise to the images, and blended this in with 25% weight, as well as some other LoRAs to improve the skin textures.
I hope that helps ^^
@6tZ Definitely helps, thank you. I've been setting up a full training pipeline with ClearML to learn AI engineering more formally. Do you source your datasets from somewhere or do you build out small ones yourself? It seems like there are a lot of SDXL models/LoRAs but a lot to be built on the FLUX side still.