TRUALITY ENGINE (PRO)
THE BRAND NEW PRO VERSION IS A MAJOR ENHANCMENT OF V3
More Contrast, More Sharpness, More Depth and More Warmth and all your older and more current prompts from Version 1-3 will still work in the PRO VERSION. DON'T USE VAE
PRO VERSION ADDED JANUARY 6TH 2024
click here if you want to support me and donate to our common cause
SAMPLE IMAGES
Everything you see is coming straight out of the engine, i am using no Loras to create effects, cameras or moods. I am using women and faces from this checkpoint and of course my very own Vicious Girls Series
HOW TO PROMPT?
Prompt as you wish, depending on what you are going for it doesn't need many words.
If you want to create a particular scene with particular assets, put those in brackets or double brackets. If that doesn't do the job use the weighting method (+/-) from 0.5 to 1.5
The higher the weight, the lower the quality.
SETTINGS
Steps: 35
with this model, too high means it takes longer to load, test it for each scene individually, keep it at 35 if you don't mind waiting a few seconds longer.
CFG: 3.5 to 11.5
The higher the value the higher the contrast and the amount of colors. The engine will listen more subtly to your prompt up till CFG 7. Everything above could be described as over-exaggerating your prompt to adding new features. A Lens Flare can become something entirely else. Definitely interesting if you feel like experimenting.
When using TI's and Lora's you should not go higher as 4.5
Best Size 512 x 768
Best Samplers: DPM++ 2M SDE KARRAS
HIRES FIX
Upscale a 512x 768 by 2x
UPSCALER for V3 4x UltraSharp
Clean and less grainy results, yet very detailed
(Resize Steps 15 to 25, Denoise 0.23-0.33)
UPSCALER for V1 & V2 & Pro: 4x_NMKD-Superscale-SP178000_G.pth
(Resize Steps 10 to 15, Denoise 0.17-0.33)
This upscaler provides a special layer of depth for photo realistic images, slightly grainy.
LORAS
Most Loras will distort your results, you have to weigh them in, 0.7 seems to work best by default, each Lora is different and behaves differently. Many... if not the most will mess with your models face because they were trained with a different face, sometimes you can fix it with clip skip and if that doesn't work just use roop or face swap and write your girls face over the trained face. Most Loras are designed for 512x512, when bodies or objects get strethed or appear twice then lower the output size.
WEIRD FACES
Especially from side angles and long distances you will keep running into weird faces, Hires Fix will 9/10 times fix this.
Please share your favorite results, consider rating this model if you like it.
Get your own custom and exclusive Influencer---> contact me in chat
or see my Fiver Page
DISCLAIMER: All Models released under the Vicious Family are fictional, I can not know every person, every actor, every influencer in the world and resemblances are purely coincidental. While it is possible that I take inspiration from a look that I like, I have not, not will I ever use any of their original fotos to create a resemblance on porpuse. All my models are either entirely created by prompts and trained on the results or mixed with body parts, facial features of 2 female friends who have consented in writing for their images being used in my process. If anyone feels that one of my models resembles a celebrity, actress or influencer and feels concerned about it, please leave a comment, I will take it to the person, management or representative directly.
Description
DON'T USE VAE
SETTINGS
Steps: 35
with this model, too high means it takes longer to load, test it for each scene individually, keep it at 35 if you don't mind waiting a few seconds longer.
CFG: 3.5 to 11.5
The higher the value the higher the contrast and the amount of colors. The engine will listen more subtly to your prompt up till CFG 7. Everything above could be described as over-exaggerating your prompt to adding new features. A Lens Flare can become something entirely else. Definitely interesting if you feel like experimenting.
When using TI's and Lora's you should not go higher as 4.5
Best Size 512 x 768
Best Samplers: DPM++ 2M SDE KARRAS
HIRES FIX
Upscale a 512x 768 by 2x
UPSCALER for V3 4x UltraSharp
Clean and less grainy results, yet very detailed
(Resize Steps 15 to 25, Denoise 0.23-0.33)
UPSCALER for V1 & V2 & Pro: 4x_NMKD-Superscale-SP178000_G.pth
(Resize Steps 10 to 15, Denoise 0.17-0.33)
This upscaler provides a special layer of depth for photo realistic images, slightly grainy.
FAQ
Comments (55)
Magic! the Pro Version is a superior!
Thank you so much :-D
@Tower13Studios聽I've been using it too with LCM (since i only have entry level gaming laptops, aka GTX 1650). And so far it's all working good 馃憣
Wow just great work on everything. I don't think I even bothered using any hand negatives, it draws the vast majority with no issue. keep doing what your doing.
Your content is BRILLIANT. Thank you.
thank you very much :-D
please make it in sdxl turbo
This superscalar that you recommend gives me very very odd results and kinda butchers the image any idea why?
No not at all to be honest, HighRes Fix is supposed to stabilize the image. I need more information. Can you upload an image so i can read out the metadata?
@Tower13Studios聽Im using vladmadic i don't really see a high res fix option
@jkermdawg聽Vladmandic is basically an alteration of Automatic1111. Vlad did a great job but people don't contribute as much to his version as they do for Automatic1111, therefore it's always behind. I don't mean that in a negative sense. He only has 2 hands.
If you use exotic Software, it's likely to get exotic results. If I was you, I would install the most recent version Automatic1111, Highres Fix is a standard implementation and is part of the generation process, it makes faces more stable, less distorted. What you are doing is upscaling an already problematic image that has not received the stabilizing treatment.
@Tower13Studios聽I found the issue it was a weird bug with one of my extensions! now im getting insane results
@jkermdawg聽awesome, glad to hear that :-D
@Tower13Studios聽just posted one shit looks on point 0 inpainting as well
Civit seems to have finally fixed the octo problem and I can use the truality to its full potential again.
Thanks for all you do and look foward to playing with it again and keeping myself amused at work. Keep up the great work!
that's great to hear
Thank you for The Truality Engine! It does some fantastic dark stuff when you let it loose (cfg<2, >1024x768 resolution). Love it!
Thank You, it was originally created to be used on a broad range of content. Dark works :-D
Just a great realistic model with good flexibility and huge variability. Easily and efficiently copes with any prompt, no matter how difficult it may be!
Thanks!
thank you :-D
is there an inpainting version?
No, you can use any inpainting model that you like
Great results with this model, tho the hands are worse than ever, even the texture changes into something more alien. Anyone got some tips for the damn hands? Thanks all.
The pro versions delivers some of the best hand results, trust me :-D the issue is in your settings or prompt.
I can't analyze based on no information. Basically what you are saying could be anything, too high CFG Scale would be my first guess when you are talking about alien textures. try a CFG Scale of 3.5 and see if that fixes it. Don't use a refiner or VAE, it's baked in already
Hope that helps if not respond with an image
Hyper Version please for us Med systemers
I would never sacrifice quality for speed, just work around it with smaller highres fix upscales and upscale the images you want to keep later, preferably with an external software like Gigapixels from Topaz Labs where you could make 10K images that look better as the original.
The checkpoint does not slow down your system, it's the features you chose and how you use them.
Limited versatility. Tried to change age or ethnicity with no luck.
Versatility is literally the hallmark of the checkpoint. It's all about the prompt and based on what I have seen in your portfolio, you treat this SD 1.5 Checkpoint like an XL or Pony Checkpoint with flowing text and some brackets here and there. A consistent SD 1.5 Prompt contains weights in brackets. "African Girls" is not a trigger for black skin tones for example, Africa is a continent with 3000 ethnic groups including white people.
Loras such as flashing "b枚枚bies" can also affect your results because they are generally trained with white women which overwrite your prompt which is too weak when you don't put weights on it.
There's a billion compatible loras for this by the same creator, don't give up
Any plans to do a SDXL variant or nah
not even a little bit :-D
The whole point of Truality is to create reality based content.
SDXL is a lame overblown nonsense when it comes to create real images.
Hello, I recently installed Fooocus and I'm learning how to use it, I'm quite new to stable diffusion. Could you tell me how to make this model work on Fooocus? Thanks.
I have no idea, never looked into one of these exotic SD Clients
If it is operating based on the original file structure the model goes into the folder models\Stable-diffusion
Fooocus is for SDXL. You can use DeFooocus (https://github.com/ehristoforu/DeFooocus) which is a Fooocus fork with limited SD1.5 support.
@iwasrobert聽thanx for sharing that
I struggle to get face consistencies on low resolution. Prompting on detailed face and all features just does not produce the same face as shown on the model page. Looking at the gallery I see the same effect.
This changes when doing Hires-Fix. All fixed images in Hires work fine (this has it's price on this and other platforms, one reason probably why base models based on SDXL are more popular.
The image quality from higher resolution picture even with negative prompts and 1.5 based textual embeddings sometimes destroys hands or even adds limbs. One reason is the limited context window of 1.5.
Overall I can recommend the model within the limits of 1.5. While Flux looks promising I prefer nowadays backports of Flux to models based on SDXL (base SDXL can be debated. I have not tried to get a good workflow because there are a lot of models providing it.
Creating consistent characters is not so difficult and the quality gain with additional embeddings or Lora make me question this 1.5 model.
Commenting on your first part
HighresFix is an essential part in SD 1.5
ADetailer improves results as well and I haven't created an image in the past 12 months without it
In the first Adetailer Tab goes the embedding name with inpainting denoising strength of 0.3 to .035
The second tab you can use an eye color model or a hand model, in case of a hand model leave the prompt empty, if you are using an eye model prompt the eye color. If you want the model to appear entirely without make-up and real human pores and the occasional imperfection you use a face model in the 2nd tab and leave the prompt empty.
Embeddings don't destroy hands. Hands have never been good or even remotely useable.
Embeddings are the least destructive addition to a checkpoint because it doesn't overwrite any quality features while Loras hijack an image and therefore dictate the quality.
Literally my favourite model to use and make custom merges with, incredibly versatile. Hope v5/ Pro v2 is coming soon :)
Thank you very much, i am not planning to release a new model even when they already exist. They are excusive to my client base so they have something different. Never say never but at this point i have no idea how to get more out of it.
@Tower13Studios聽Client base? You have more models?
@MassivetART聽yeah but that's exclusive to the business side of what we do. They get custom models and some of the larger clients get custom checkpoints that work particularly well together with their orders.
@Tower13Studios聽Do you have a patreon or something where you host this stuff for a premium subscription or is it like individuals reach out to you for a product just for them? Also we? you're more than one person?
@MassivetART聽it's not a subscription based thing, it is custom orders for Influencers or Female Models
Excellent model. I just found it and it swiftly rose to my top 3 models to use. Excellent work.
thank you
Why is The Truality Engine no longer available to generate images here on the site? 馃槶馃槶馃槶
I have no idea, I have stopped questioning what they are doing. I have really now idea. After 4 million generations you would think this is a very established check point. They do things on a daily basis only they understand. They are living in a bubble.
this might explain it
they didn't roll back, they excluded previously existing check points such as mine.
https://civitai.com/articles/12689
@Tower13Studios聽But this Checkpoint was existing before the new changes were rolled in. Why have they excluded it now? Its really sad.
@CapTab聽because they didn't roll back 1:1 and added checkpoints manually. It is what it is, I have learned to cope with the hundreds of things I disagree with on this site.
This model is not available to me. How are people still able to use it?
it's only available for download and local use, Civit wants us creators to bid on it (pay money) in order for people to be able to use it.
good checkpoint, responsive to prompts and giving me very nice results. 馃憤
Details
Files
theTrualityEngine_trualityENGINEPRO.safetensors
Mirrors
theTrualityEngine_trualityENGINEPRO.safetensors
theTrualityEngine_trualityENGINEPRO.safetensors
theTrualityEngine_trualityENGINEPRO.safetensors
theTrualityEngine_trualityENGINEPRO.safetensors
theTrualityEngine_trualityENGINEPRO.safetensors
theTrualityEngine_trualityENGINEPRO.safetensors
theTrualityEngine_trualityENGINEPRO.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.





