You can get this model through Civitai Early Access, or grab it along with many others by joining The Tinkerer on Whop. Membership gets you early releases, private tools and members-only pages. There are free pages too, no membership needed.
š Join on Whop
š¬ Join the community for support, free tools and early news on Discord
CyberIllustrious blends the artistic flair of Illustrious-XL with a big dose of CyberRealisticās style. You still get all those cinematic, dreamy strokes, but now they come through with crisp, realistic detail.
If you want that painterly, elegant look but still want your images to feel sharp and lifelike, this is your model. Itās great for editorial vibes, soft glam, or even fantasy art that looks like a real photo. Basically, it just nails it.
Itās powerful, easy to use, and honestly? Prompting it is seriously fun.
š Get the most out of this model with the Illustrious / NoobAI Prompting: Master Guide.
āļø Recommended Settings
Sampling method: DPM++ SDE Karras / DPM++ 2M Karras - SGM Uniform / Euler a
VAE: is already Baked In
Sampling steps: 30+ Steps
Resolution: 896x1152 / 832x1216
CFG: 4-5About (DMD2)
The special DMD2 edition of CyberIllustrious is created for faster, cleaner generations while requiring fewer steps.
DMD2 takes the balanced improvements of the standard release and optimizes them for speed, precision, and responsiveness, making it ideal for both SFW and NSFW workflows.
Expect tighter results, smoother control, and a model that reacts beautifully to the classic Illustrious prompting style.
Recommended Settings
Sampler: LCM
Schedule Type: Exponential
Sampling Steps: 8 (12 for extra fine detail)
CFG Scale: 1.5
ā” These settings are optimized for DMD2 and are different from my standard Illustrious workflow!
š§ Prompting Made Easy
Positive Prompt:
masterpiece, best quality, high quality, ultra-detailed, realistic, ... rest of promptNegative Prompt examples (optional):
cartoon, illustration, anime, painting, CGI, 3D render, low quality, watermark, logo, labeldeformed, deformed face, low quality, bad quality, worst quality, (drawn, furry, illustration, cartoon, anime, comic:1.5), 3d, cgi, extra fingers, deformed teeth, watermark, logoš©ŗ Optional ADetailer Settings
Adetailer model: face_yolov9c.pt
If you only want the main face being refined set 'Mask only the top k largest' to 1.š Resources
Backup location: HuggingFace
ā Support the Madness
Liking what you see? Grab me a digital espresso (or a couple!). It keeps the creative energy going and helps bring more cool prompts and models your way.
ā ļø Heads Up
This model can spit out some mature stuff. Use it wisely, and please, no cursed creationsāweāve really seen enough of those.
š¬ Want Better Prompts?
Use this custom ChatGPT to auto-generate polished prompts for CyberIllustrious:
š [Try it now on ChatGPT]
Description
This an alternative edition of CyberIllustrious V8.0 Redux, built on a different foundation.
FAQ
Comments (18)
Ngl, I think v8 is a step down from v7, v7 creates more realistic results.
Yes, but at what cost? CyberIllustrious V7 may look more realistic, but it loses some of the key characteristics (prompt) that define Illustrious.
I could simply push the realism sliders all the way to maximum and end up with a highly realistic model, but at that point it no longer feels like Illustrious.
@CyberdeliaĀ Tbf mate, I just go for whatever models creates the most realistic characters. v7 was great for that. As for what 'feels' like Illustrious I couldn't say because I'm not up to speed on differences when compared to, let's say, Pony for example.
@JD1997_Ā I'm releasing a LoRa soon (Tomorrow/Monday) that gives a flare of realism to these models. Maybe it is something for you :)
If you want realistic then use CyberRealistic XL (and yes it can do NSFW)
@CyberdeliaĀ No worries mate, sounds good :)
@JD1997_Ā it's aweaome :) will release it tomorrow
@CyberdeliaĀ Thank you bud
he has another model named:
CyberRealistic CyberIllustrious Realism LoRa
(cyberrealistic_illustSemiRealV11.safetensors)
and this model:
cyberrealistic_v80Redux.safetensors
for some reason its exactly the same file size:
meaning it seems to be just a copy/paste.
both file sizes are:
6.61 GB (7,105,347,814 bytes)
The file size doesnāt mean anything here.
Most of my checkpoints end up the same size because theyāre all correctly pruned - thatās literally how pruning works. SD models have fixed tensor shapes, so when you prune them properly, the final filesize often lands on the same number down to the byte.
That doesnāt mean theyāre ācopy/paste.ā
If you actually load the two models or generate with them side-by-side, itās immediately obvious they behave completely differently. Different datasets, different training runs, different style, different output - the only thing they share is that theyāre both pruned to the same structure.
So yeah, the size being identical is completely normal. Most of my checkpoints are the same size for exactly this reason. File explorer ā model similarity.
If someone really wants to check, they can open the weights or just run a test prompt - the difference is very clear.
Could you elaborate regarding differences between both of your Redux v8 checkpoints? Is one better at X but not as good at Y compared to the other Redux v8 version?
Both are good. Just check the sample images and compare them.
@CyberdeliaĀ Yes, both are good, that's for sure. But that's not what i'm asking. ^^
The showcased images don't answer my question, because it's not a set of 1000 images, but only a few. So i can't say whether what i'm observing is specific to the seed used or to the checkpoint itself.
For example, if i look at the images you presented, the Alt version of Redux v8 seems to produce images with a tab bit more natural look and more anatomicaly correct hands/fingers, while Redux v8 seems to favor characters that look more like magazine covers.
However, what i just said can't be stated with certainty based only on so few images. Thus why i'm asking for more details when you describe changes in new checkpoint releases. So when you type "This an alternative edition of CyberIllustrious V8.0 Redux, built on a different foundation." as a checkpoint description it doesn't explain anything regarding use cases.
This is not a criticism, but i'm sure you understand that people who don't use checkpoints randomly need to understand the strengths and weaknesses of the checkpoints they use.
@antarekĀ Both versions are solid - the difference is mostly a slight shift in style because theyāre trained on different bases. It really comes down to preference, so the best way is to try both and see which vibe fits your project better.
Itās basically like two good coffees from different roasters. Same category, slightly different flavor notes. Best way is to test both and pick what fits your style.
I have found both versions of V8 to give great results. It is just a matter of your preference. Both give more realism than I had believed was possible with an illustrious model. The one negative I found was that human eyes all have an "uncanny valley" look to them, even with inpainting or at high rez. This is the area I would focus on for the next version
Was the name of this model changed at some point?
I have a model file named "cyberillustrious_v60" (I don't rename them) and I'm trying to figure out if it's an older version of this model or some different model. The new ones are prefixed with "cyberrealistic" so I'm not sure but I can't find any other model matching that name.
Civitai renames the files. The original filename was CyberIllustrious_V8.0alt.safetensors, but it looks like itās been changed to something else now. No idea why they do that, but itās really inconvenient.
Ok so that means the "cyberillustrious_v60" I have is probably a version of this one from before civitai decided to change the name. Thanks!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.