experiment model
start from here:
https://civarchive.com/articles/33117/experimental-with-sdxl
the result: XD
Juggernaut version
Since the concept involves so many constraints, locking the canvas size before entering the prompt is likely a necessity.
medium shot portrait of bla bla bla,
medium shot of man selfie on japan street on market
close up portrait of bla bla bla
medium shot of bla bla bla
etc,
Thank you to everyone who provided feedback and test.
Sample workflow:
https://civarchive.com/api/download/models/3212463?fileId=3097585
Known Issue:
https://civarchive.com/articles/33674/experimental-with-sdxl-or-part-2
========================================================
in here:
didn't train the base model or merge to any model, just tuned the base model by shifting it backward and increasing the some weight value 🤫.
work with any lora lighting step like..
https://civarchive.com/models/350450/sdxl-lightning-loras
https://civarchive.com/models/1608870/dmd2-speed-lora-sdxl-pony-illustrious
vae:
https://huggingface.co/madebyollin/sdxl-vae-fp16-fix/blob/main/sdxl.vae.safetensors
Personal/Non-Commercial only. Commercial use requires written permission.
This model is a modification based on
Juggernaut XL (Ragnarok) created by RunDiffusion.
All credit for the base model goes to them. 🔗
Original model: [https://civarchive.com/models/133005/juggernaut-xl]
Description
this Juggernaut version tunned by text + cross attention on low cosine drift..
text: incredibly lifelike, masterpiece, exposure mapping midtones tone curve, clean color grading, iso noise, film grain, harsh shadow, harsh lighting, ambient occlusion, ultra imperfect framing, aesthetic amateur candid, poor lighting, dutch angle view. etc XD
FAQ
Comments (2)
thats something new:)) when its good, its marvel(really photolook).
Prompt adherence not existing:) 10/5 picture is some random something, nothing to do with prompt. Its like you would use it on 0 cfg.(girl with sunglasses give a pair of sunglasses:)))
But its the only sdxl model, what worth tune further. Good luck.
Thanks a lot for the feedback—it's really helpful. It's pretty chaotic, isn't it? XD... I'm really sorry; the bucketing feature is still under development, so I just bundled all the keywords together and threw them in. That bug definitely pops up around the 8th generation in here. T^T...
I really appreciate the feedback...
Maybe you could try using this: "medium shot portrait of girl with sunglasses give a pair of sunglasses at paris"
Thanks again! :)