MoP Mixes are a series of checkpoints that merge the MoP (Mixture of Perverts) model with one other model that I find interesting. These are completely separate from the normal MoP releases because those are a much more complicated merge that requires a lot of testing. MoP Mixes are much simpler to produce with the goal of creating a MoP variation that hopefully some people may find useful or just fun.
To be clear, these MoP mixes should in no way be considered better than or a replacement for the checkpoint being merged with MoP. I strongly encourage you to download and test out the checkpoint for yourself and use it in the way that its author intended.
Each checkpoint will be set up in the manner that I think works best. So some of them will have the DMD2 LoRA fully integrated, some partially and some not at all. The recommended settings for each mix will most likely be different so check the About This Version section for all the details, as well as the recommended workflow.
Note: If you want to convert any of these DMD checkpoints to a non-DMD checkpoint then all you need to do is add the DMD LoRA and set the strength to a negative value. I would start with -0.85 and adjust from there. That will allow you to use a higher CFG and different samplers than LCM.
Fooocus Setup Instructions
1) Click the Advanced checkbox.
2) On the Settings tab, leave the Preset as 'initial' and the Performance as 'Speed'.
3) On the Models tab, set both the Base and Refiner models to MoP.
4) Change the Refiner Switch At setting to anywhere from 0.75 to 0.90. I like 0.85.
5) On the Advanced tab, change the Guidance Scale to 1.
6) Change the Positive ADM Guidance Scaler to 1.
7) Change the Sampler to 'lcm'. You can leave the Scheduler as Karras or Exponential.
8) Set the Forced Overwrite of Sampling Step to 7 to 11. I like 9.
Description
This is a merge between MoP v5.1 DMD and Switch Mix Toon V2, which is a very cartoonish checkpoint. Unfortunately, I can no longer find Switch Mix Toon on CivitAI so I'm not sure what has happened to it.
You can use this MoP Mix workflow for best results.
This is a DMD-enabled checkpoint so the recommended settings are:
Sampler: LCM
Scheduler: karras, exponential or beta
Steps: 8-14
CFG: 0.8-1.2
FAQ
Comments (25)
What'about upscale, are there blisters? After 1.5-2
Not sure what you mean by 'blisters'. As far as upscaling, I tend to only upscale about 1.25 because I generate at a higher initial resolution. If you use my workflow, it has upscaling built in, https://civitai.com/models/2130186?modelVersionId=2411015.
sorry for my bad, i use fooocus, after enhance upscale some results in some models have blisters artifacts. Now testing yr model, results not bad. Very good!
@ArtdeO Did you follow my Fooocus instructions for using DMD models?
@GBRX not yet, but interesting, where workflow? i see only images in MoP Mix Workflows, simple descriptions under image. Me need seen structure UI settings setup (Fooocus UI).
i tried yr last settings with my adv settings of upscale into Fooocus enhance - results good, no blisters on skin, body, no bad eyes, impressive!
so you'll get a phat like from me)
i added 0.3 in Forced Overwrite of Denoising Strength of "Upscale"
omg your bringing out some amazing models, im quicky running out of space for them all lol. Keep up the amazing work
Thanks!, I'm running out of space too!
Okay, this one I really like!! Natvis... Nice bodies, faces, lighting, faces, bodies... :-P See posted images. Thanks!
It refuses to follow even the most basic prompts, like "in a jewelry store" could be any store, or if you type "braid-waterfall," a waterfall is guaranteed to appear.
Which specific model are you referring to?
@farengete22487 It definitely doesn't understand that hairstyle. I couldn't get it to work. To be fair, I've never heard of that hairstyle before and had to google it ;). I did get the jewelry store to work simply by adding 'jewelry case filled with expensive diamond and gold jewelry' to the prompt, https://civitai.com/posts/24544840. So being a bit more descriptive in the prompt may be necessary but that is the hallmark of NatViS. It's supposed to work well with natural language prompts that are more descriptive.
DMD2 in this paired with decent CLIP L models thats used should do it, but you might need to let LLM (Gemini or Grok) write you prompt. If you need it NSFW, try to figure SFW version and modify it later.
It should chew rather lengthy prompt, DMD2 if merged right (it is merged right here) allows that.
As for hair style, LLM should use description instead of name of actual hairstyle, which can possibly work, if CLIP(s) manage to "get it".
In general, unless there is LoRA or model that actually understands "braid-waterfall", you are out of luck and either need a) train LoRA or b) use Gemini to prompt Chroma via image input, that will work with rather high degree of probability. Thats if you can/want to run Chroma.
Good ideas. For me a realistic model must be good at emotions, and therefore it can provide good lifelike eyes.
Natvis is great at this, but forgot the inbuilt celebs, while Jugger lacks emotions. I use celebmix to get different faces. Original MOP is still too yellow tinted.
I will wait for your next mixes. Thanx.
Hi. I mainly use SDXL models to generate celebrity images, sometimes with limited success. I'd like to know if you can advise me on which model generates the most accurate celebrity images. Thank you very much!
Sorry, I've never really generated anything for celebrities.
@MirandoPaCuenca The usual ones are great, like Juggernaut and Cyberrealistic. More porn finetune, less celeb likeness(Bigasp, Lustify and their merges)
@ronitmikias846 Thanks for the recommendations!
@MirandoPaCuenca You can still download celeb loras(not all)
Search for deleted loras. They are good on every model.
Nice model and handles upscaling well. Skin color stays true to OG render.
Great! I'll try it out this weekend.😋Let's get the babes machine running.😍BIG THANKS
Dammit, another new one?? I was just shutting down my rig to do something else... GRR... :-P