Welcome to Perfection Amateur!
This is an Illustrious XL merge with a bunch of hand-selected checkpoints and LoRAs to improve the base model's realistic capabilites.
This is a special branch-off for cinematic flavor and for when you need your images to be extra juicy and well-lit!
It's focus is on creating high quality, photorealistic but amateur style images with a focus on women.
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.
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.
Recommended Settings
Base Generation
Steps: 30
CFG-scale: 3
Sampler: DPM++ 3M SDE
Scheduler: Simple (beta upscale scheduler)
Upscale/Refine
Steps: 16
CFG-scale: 4
Sampler: DPM++ 3M SDE
Scheduler: Simple
Denoise: 0.35
Description
Keywords that could help or give different results:
Amateur photo
Amateur
DCIM
onlyfans
snapchat photo
instagram photo
raw photo
FAQ
Comments (8)
Hi. I am having issues generating good Images with this checkpoint. I am new to XL models so i am pretty sure i am doing something wrong or miss some thing. Do I need a specific VAE or something? I hope you(or someone) can help me.
Here are some examples of the output I get
https://civitai.com/posts/16365960
Hey, yeah it looks like you need to tweak some settings there!
There are more complex workflows you can adapt, but if you want to keep it very simple, you'll want to tweak some settings.
First:
Change the "Scheduler" from Normal to Simple or Beta.
This will fix the extreme noise distortions you are seeing.
Second:
Change the sampler from "euler" to one of the following:
euler_ancestral
dpmpp_2m_sde
dpmpp_3m_sde_gpu
Third:
Play around with the CFG. Higher CFG, will give more darkness and depth/punchyness to your images. Lower will make the image "lighter", more washed out, and less detailed.
After this, you'll want to start working on the prompt. You should probably prompt for a background for the character. Otherwise you might get very plain, white, or wallpaper backgrounds.
More details in prompt means less generic outputs:
```A pretty woman in a dress holding an apple. She is standing in an apple orchard on a sunny day. Green grass. She is smiling and have blue streaks in her hair. He holds one arm out. Close-up portrait photo.```
Here's a post with sample images:
https://civitai.com/posts/16367064
For more advanced, and higher quality outputs, you can download the images from the main model post. It has workflows built into it. It might be some special nodes used, which you can download, or you can skip. Like you don't need to use the fancy "save image with metadata" node etc. Replace it with what you have and know.
Good luck!
@6tZ Thank you very much for your quick response and detailed tips. Changing the scheduler and the sampler did the trick ;-)
I switched from A1111 and 1.5 to comfyui and XL ... so I have to relearn a lot of stuff!
@Mr_Monster Nice! Welcome to the dark side! This is where shit really takes off!
Stay away from Flux for a bit. Get into SDXL, Pony and Illustrious and stay there until you are comfortable.
Then consider starting to tame the beast that is Flux :D
This doesnt know anything about poses. You merged with some amateur XL model?
Could you give some examples?
It includes this one for example, to get less "perfect poses":
https://civitai.com/models/347310/amateurs-cant-pose
sideview doggy, 1girl 1boy i tried. Result two solo girls.
@jamesclark171 Oh,yes the model is definitely overfit on women! This is just lack of prompt coherence. It's something I'm trying to improve constantly. I need to "wash it out" with less biased models.



















