Note: For optimal results, it's recommended to use this model with a VAE. While it can still function without one, incorporating a VAE will significantly enhance its performance.
IMPORTANT: Due to this seed breaking change [ 2023-02-19 - deterministic DPM++ SDE across different batch sizes],
The example images cannot be reproduced unless you toggle the: Do not make DPM++ SDE deterministic across different batch sizes setting, found in the Compatibility tab of A1111's settings.
Meet a new general-purpose model designed for creating realistic images of the human form against different backgrounds. Lazy Mix is focused on capturing natural beauty, giving you photorealistic results every time.
While it takes inspiration from the popular Ares Mix, this mix is unique in its own way. I have made changes to some elements, to suit my own personal preferences, making it a one-of-a-kind tool. Unlike Ares Mix, this model doesn't include the anime core, which has pros or cons, depending on your use case.
Whether you want male or female subjects, this model is up to the task. It excels in capturing every detail, use a desired embedding to push it even farther, making your images look as close to reality as possible. While photorealism is the main focus of LazyMix, there is also capabilities for creating NSFW content.
This model is all about bringing your vision to life, whether you're an artist, photographer, or just someone who appreciates the beauty of the human form. It's a perfect tool for anyone who wants to create stunning and realistic images, without having to go through a lot of technical hassle. Relax, and use LazyMix.
This model includes DreamLike PhotoReal v2, here is it's license.
~ Description Written by Chat GPT (Revised by creator)
Description
v2:
New Lazy Base
Replaced DreamLike PhotoReal V2 with epiCRealism Pure Evolution v5
New Lazy Base is more NSFW, use a negative prompt if you wish to avoid NSFW
VAE is now baked in
fix:
included the final model merge which was previously missing
FAQ
Comments (13)
Really, 5 days in early access? I have over 500 models to mess with, including the previous version of this, I don't need this one that bad.
This version includes many of the improvements in the latest LazyPlus. It's definitely a lot better but more biased towards NSFW.
As for early access, it wasn't meant to be seen as an inconvenience, I just prefer a scheduled release and I was under the impression I was supporting the CivitAI community by doing early access.
@kaylazy Fair enough, but it's not usually 5 days, I'm not sure I've seen it that long before. I'm not bothered, I have plenty to mess around with, thanks for the update!
@MysteryWrecked you can probably filter out seeing early access items, so just do that and you won't be triggered.
Early access is simply a means to get people to ($Pay$) to not have to wait, that is all that "early access" is, [no offense], there is no other reason for it. This only works for people with disposable income and very low self discipline however.
@pursuit_of_beauty ok, and maybe it's just good business to utilize users with disposable income who want a model early and it's not that deep? do 'anti-capitalism' sentiments have to rule the entire GenZ/Milleninnial population, or can creators just try to get ahead in a system that isn't going away in any of our lifetimes? ffs, not everyone is broke and bitter
I totally get the drive to monetize the viral AI craze around this technology, it actually bothers me a little that I'm not industrious enough to find a way to make a buck off of it myself. I'm absolutely sure that if I dusted off that old dark web knowledge I had in the ought's, I could get some real shady deals going with some of the stuff these models can do, but that's the best idea I've had, and it ain't a good one. Kudos to whoever gets rich off this stuff, and shame on 'em too.
I'll just have my fun making goofy and spicy pictures to keep on an encrypted drive nobody will ever see. This model is actually pretty good by the way, thanks for bringing it up in my notifications so I could grab the fixed version!
When can we expect the pruned Safetensors? Pickles are slow as heck.
it turns out i just now realized I uploaded the wrong model file. It was only the base mix and was missing one more model merge. Wooops. Safetensors is being uploaded now.
holy shit this model is just perfect. Resoults are insanely great!
can u tell me how to install this model?
Any thought to doing a Flux version of this model? I quite liked it when I was messing around with SD 1.5. I think a Flux version would be great.
The lazy mixes should have way more likes than they have. They still turn out some of the most realistic stuff I make with the least effort and the most consistency.