(2024/02/02) OhMen Toontastic is online ! Link here
>>> Please, make sure to check out the model description, especially the "best settings" and "to go prompts" sections <<<
OhMen Lifelike is the enhanced version of OhMen Origins, designed to create realistic and lifelike images
Check my previous mixes here.
The OhMen Lifelike mix is expertly crafted to enhance male characteristics by combining different base models and my trained mix "OhMen Origins".
It simplifies the creation of lifelike / photorealistic images of handsome and beefy males, without depending solely on LoRAs.
However, if preferred, LoRAs can still be utilized as the mix offers great versatility.
The OhMen Lifelike mix also provides a wide range of sub-styles offering even more options for artistic expression.
If you're interested in bringing OhMen Lifelike to life through merging, you can refer to the "Merge a OhMen model" section. This will guide you on how to combine and blend different elements to create your desired result.
╭───────────────────────╮
│My to go prompts
╰───────────────────────╯
Positive prompt : photo (1man), muscular, brown eyes, casual, bulge,
Positive prompt (NSFW) : (1man),flaccid penis, saggy testicles, muscular, brown eyes,nude
Negative prompt : (render, cartoon, cgi, render, illustration, painting, drawing,deviantart), FastNegativeV2, deformityv6, bad-hands-5,╭───────────────────────╮
│Best Settings
╰───────────────────────╯
Clip 1 is highly recommended, while Clip 2 will likely generate pics with less details.
DPM++ 2M Karras, DPM++ 2M SDE Karras, or my personal favorite, DPM++ 3M SDE Exponential Sampler. Restart works well too
For sampling, try using 35-60 steps.
To get the best results, set your CFG scale at 6. It works great between 3.5 and 8.
Make sure to uncheck the Face restore option.
Recommended resolutions fall between 512x512 and 768x768.
You can go for 1024x1024, but sometimes it can generate some weird pics.Keep in mind that upscaling can sometimes reduce the level of details instead of increasing it.
To counter that, I recommend using a denoising strength of 0.35 or 0.4.
Also, try using 15-20 Hires steps and either "NMKD-Superscale-SP_178000_G" for most of your gens or "8x NMKD Faces" for close-ups and portraits as your upscaler.If you want your faces to look super clean and even more realistic, you should totally give the AfterDetailer extension a try. It's not mandatory since OhMen usually does a really great job with portrait, but using it can add that extra touch of realism to your composition. Same for penises, but you can check out this Penis model for the AfterDetailer extension ;
The Vae-ft-mse-840000-ema-pruned is included but you may need to set your VAE on 'automatic';
If you wanna gen some NSFW pictures, add the "flaccid penis, nude, saggy balls, ..." tags at the beginning of your prompt.
╭───────────────────────╮
│Known Issues
╰───────────────────────╯
V1
Lack of color in my gens: add the following in your negative prompt as suggested by @boomerang: "(colorless, HDR, flat, lens flare)."
Skin appears 3D generated: (render, cartoon, CGI, illustration, painting, drawing, deviantart).
Feet aren't properly generated : SD doesn't generate feet properly most of the time. However, you can solve this issue by adding the "Neg Feet" embedding in your negative prompt or using a LORA such as the "Male Feet Pose" one.
You don't like beefy or super hairy guys ? Add "(Muscular:1.5) or "(hairy:1.5)" in your negative prompt, I'm gonna try to fix that in the V2. This issue has been fixed in V2.
V2 (No known issues reported yet, please report them in the "Discussion" section)
╭───────────────────────╮
│Merging a OhMen model
╰───────────────────────╯
A OhMen model represents the culmination of countless hours dedicated to make it stable and easy to work with. So, I recommend the following:
Set OhMen at 0.5-0.6 as Checkpoint A (if you plan to use it as a base model) or 0.35-0.5 if OhMen is your Checkpoint B.
If you plan to merge LoRAs, merge them with higher values before adding OhMen to your mix.
As OhMen already includes HQ built-in penis, set a lower value if you are merging one.
In case you are block merging and only want a specific feature from the model (as checkpoint B) :
Strong overall style, textures affected: 0,0,0.5,0.5,1,1,1,0.5,1,0.5,0.5,0.5,0,0,1,1,0,0,0,0.5,0.5,0.5,0.5,1,1
Strong overall style and composition, genitals untouched (from a Rentry post): 1, 0.9, 0.7, 0.5, 0.3, 0.1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0.1, 0.3, 0.5, 0.7, 0.9, 1
Alternative overall style and composition: 0,0.9,0.7,0.5,0.3,0.1,1,1,0.35,1,1,1,0,0.45,0.25,0.25,0.5,0.35,0.5,0.25,0.3,0.5,0.7,0.9,1
strong overall style and mid-weak composition: 0,0,0.7,0.5,1,1,1,1,1,1,1,1,0,0,1,1,0,0,0,0.1,0.15,0.3,0.5,0.9,1
Mid overall style and composition, genitals slightly affected : 0,0.5,0.5,0.5,0.5,0.5,0.5,0.5,1,0.5,0.5,0.5,0,0,1,1,0.3,0.5,0.3,1,0.5,0.5,0.5,0.8,0.8
weak overall style and composition, genitals untouched: 0.2,0.5,0.5,0.5,0,0,0,0.3,0.45,1,1,1,0,0,0.45,0.4,0,0,0,0.25,0.35,0.4,0.35,0.7,0.7
Genitals, slightly affects facial features: 0,0,0,0,0,0,0,0,0,0,0,0,0,0.3,0,0,1,0.55,1,0.25,0,0,0,0,0
Strong genitals, skin textures and faces affected : 0,0,0,0,1,1,0,0,0.5,0,0,0,0,0.5,0.5,0.5,1,0.55,1,0.45,0,0,0,0,0
hand repair (based on a Rentry post)(may requires a top-rated or well trained model): 1,0.6,0.40,0.20,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
╭───────────────────────╮
│Credits
╰───────────────────────╯
A HUGE thanks to my beta testers from the Men-Only community (Unstable diffusion discord ) :
DigitalGhosts https://civarchive.com/user/digitalghosts
Resonance https://civarchive.com/user/resonance_ai
Boomerang https://civarchive.com/user/b00merang
Revelancy https://civarchive.com/user/Relevancy
Checkpoints :
- Stocked Reality : https://civarchive.com/models/86526/stoked-reality
- OhMen Origins V2 (coming up soon) : https://civarchive.com/models/207007/ohmen-origins
- Elegant Realism : https://civarchive.com/models/115920/elegant-realism
- Absolute Reality : https://civarchive.com/models/81458/absolutereality
- Humans : https://civarchive.com/models/98755/humans
- CyberRealistic : https://civarchive.com/models/15003/cyberrealistic
V2 was 95% reworked upon the updated base models from V1
Loras:
- DPO Lora : https://civarchive.com/models/242825?modelVersionId=273993
- Skin HQ : Personal Lora trained on skins
- Furthermore : https://civarchive.com/models/157105/furthermore-detail-enhancer
- Contrast Control : https://civarchive.com/models/155496/contrast-control-sd15
- Film Grain control : https://civarchive.com/models/155526/film-grain-control-sd15
- Advanced Enhancer : https://civarchive.com/models/106717/advanced-enhancer-lora
- Stock Male MJ : Personal Lora , trained on male stock-like photos sourced from my Midjourney RAW prompts
- Anymen RD : Personal Lora , from a combination of Anymen and Midjourney pics
Please note that with OhMen Lifelike, you have the ability to sell the images generated using this model. It provides you the opportunity to monetize your creations and AI artworks :)
Airfuck , KingApollo and Muserato ❤️💙
Description
This update is the result of months of fake successes, trials, errors, and epic block merging.
Tbh with you, it hasn't been an easy ride.
It's been tough, not because of the countless hours poured into my models, but also because I've had to push through despite some health issues.
But it's fine 🙂
*The Ohmen lifelike v1 brings some awesome features to the table. Here's what you can expect :
- Wayyyyy more stability, so no more frustrating glitches or wtf hands (unless you are very unlucky).
- I've upgraded the goods! Impressive package, both on and off duty, if you know what I mean hehe 😜 ;
- Say goodbye to the overexposed look of the 0.5 version, that is now a thing from the past.
- The level of definition in the faces has been seriously amped up, ciao jewellery faces.
- Wanna sell pics online ? No problem!
We're throwing in a permissive license, so you can turn your Ohmen Lifelike v1 gens into cash.
- For all the merge masterminds out there, I'll include block merging recipes, which will make it easier than ever to incorporate Ohmen Lifelike v1 into your merges.*
See ya in a couple of hours 😏
(small announcement, awaiting early-mid 2024 to remix and train OhMen once again, but this time for SDXL)
FAQ
Comments (24)
Meh. Extremely homogenous and not super flexible. I do hope for more development, but it's presently too samey for me. It's always good to see more male-oriented models, though.
Idk how to respond because this model is way more flexible compared to many other gay models, as confirmed by multiple individuals who have tested it against popular alternatives.
However, I'm curious if you have read the model description as it may provide additional insights that you probably have missed. Is your SD install up to date ?
Tbh, I know the model isn't perfect at all but... extremely homogeneous ?
If reading the model description was helpful, that's great!
In case it wasn't, please, reach me out on Discord : j_wolf_
Your comment is constructive, I'll keep in mind what you said
And fortunately, the next update will be more realistic.
But do you think the model has a lack of variety or looks monotonous ?
Hey @hushicho :) V2 is out
Hmm, v1 looks quite different from v.5, the style seems more like rendered material and less realistic.
And I can guarantee you that v1 does perform better in many ways.
Tbh, v0.5 had heaps of issues like jewel face effect, low quality hands, pics artifacts...
But to get the most of the model, I highly recommend you to read the model description and the "to go prompt" and "best settings" sections.
But I do keep your comment in mind for the v2
@ewolf Thanks for the comment. Yes the issues with v.5 were apparent. v1 is more consistent. A good mix for v2 would be cool ;)
@nik1_09534 Hey :) V2 is out :D
As I have in a previous comment 0.5 Beta is incredible but V.1 is a drastic change from the beta and seems like a big downgrade? It looks less realistic. Also if I rerun a previous image of a twink I did with the 0.5 beta using same seed and prompt V.1 gives me an almost sepia tone image and turns my twink into a bear!? lol
Hey :)
V1 is far from perfect but is way more stable, someone said that the textures look like photorealistic 3D materials. I keep it in mind for the next version, dw ;)
Have you read the model descriptions ? Especially the "To go prompt " and "best settings" sections
@ewolf Is there a way I can contact you to show you some comparison image between all 3 of the versions of this model? I don't see a private message feature on this website. I looked at your profile and I see X listed but I deleted my Twitter out of anger for Elon lol. If there is a PM feature on here I can't seem to find it. I'm getting such drastic differences compared your first beta and I don't understand what is going wrong. I still adore your first beta of this model over the rest!
@StecFX V2 is out ;)
@StecFX I'm "j_wolf_" on Discord :)
Also, did the "best settings" section helped you ?
@ewolf thank you. I'll take another look at the best settings section and hit you up on Discord to share some examples of what I'm experiencing if no improvement not to complain more just out of curiosity and I would like you see if my situation is unique to the platform I am using which is MageSpace.
@StecFX okay :)
Just wanna let you know, the model had a silent update yesterday as the VAE has been corrupted during the upload (which never happened to me before).
Try with the re-uploaded model but hit me up on discord anyway :)
@ewolf That was the problem once I got the fixed version of the V.2 model it was instantly so much better problem solved!!! V.2 Model looks and works amazing thank you!
@StecFX Cool ! Glad to hear that :D
The "To go prompts " and "best settings" sections will help you to get the most of the model :)
Hit me up on Discord, so I can give you additional tips ;)
Very excited to try this out!! <3
very nice model, sometimes virilereality is more realistic but this model is better at generation especially on penises. keep up the good work.
Thanks heaps for your constructive comment, much appreciated man :)
And fortunately, the next update will be more realistic, wanna test the experimental models ? ( "j_wolf_" on Discord )
hey :) V2 is out
@ewolf I am gonna check it out thanks.



















