Illustrious Renaissance v6_NS Update
First things first: all the preview images you see here are pure Text-to-Image (t2i). No LoRAs, no Hires.fix, and no Face Detailer were used. What you see is exactly what the model can do out of the box.
Sorry for the silence! This update took a bit longer than I expected, but it’s because v6_NS is a total game-changer for the model's future.
I’ve been spending a ton of time in the "lab" experimenting and training text encoders from various models. The goal? To bake in enough flexibility so you can use your favorite LoRAs without headaches.
What does this mean for you? The model is now a bit of a hybrid. I’ve tuned it to be compatible with almost everything: SDXL, Illustrious, and Pony. I’ve also started adding support for Noob—some things already work, and full compatibility is next on my list.
It’s a work in progress: Is it 100% perfect? Not yet. But it’s finally at a point where it plays nice with a huge variety of LoRAs, and I wanted to get it into your hands as soon as possible.
I need your eyes on this! Try it out, break it, and please share your feedback and images. I really need to know how it’s performing for you so I can keep polishing it in the next updates.
Enjoy the new versatility! I hope you like it. —DeViLDoNia
⭐ Illustrious Renaissance
Quick Start: For v5 and earlier versions
Recommended Sampler: euler, euler_ancestral, dpm++ 2m
Steps: 30+
CFG: 3.5–7 (5 recommended for most cases)
Clip Skip: 2
Hires Fix / FaceDetailer: Strongly recommended
Resolution: 1152x1536 / 1024x1536 / 832x1216 / 896x1152
Recommended Upscaler: 4x_NomosWebPhoto_RealPLKSR.pth
Upscale Factor: ×1.5–×4 (depending on your hardware)
Trigger Words: For v5 and earlier versions
realistic_skin_texture, realistic, realistic_detail, detailed_background
Positive Prompt Example:
masterpiece, best quality, YOUR PROMPT, realistic_skin_texture, realistic_detail, detailed_background
Negative Prompt:
worst quality, low quality
Enjoy and give me all your love: like, follow, review, post, it helps a lot.
~ DeViLDoNia
📧 Business Contact:
For commercial inquiries, collaborations or licensing, please contact:
📩 email: [email protected] / Discord: hades5826
Description
v6_NS
LoRa support from SDXL, Illustrious, Pony and NoobAI partially implemented.
FAQ
Comments (8)
Version 5 is my go to, but for some reason this new version seems to love to generate hats, crowns, animal ears, branch ears, or masks for my renders. Is this a known issue?
One of the text encoders is very sensitive and it seems that if too many qualitative tags are loaded, it starts to hallucinate.
Haven't noticed anything weird like that for v6_NS, and that's with using way too many overly descriptive words (likely well over the token limit plus a fair amount of negative prompting).
When I get something strange or unexpected like that, it's usually the model misinterpreting a particular word or concept (ie, describing a specific camera or photo settings used and then it gives you your subject holding a camera). Wait a minute! Was just looking back through my results and noticed one of them had some small crazy cone-like sparkly cat ear/antennas on her head. That was only one out of fifty results, but apparently a thing it does!
And very nice job @DeViLDoNia! v6_NS is capable of some stunning results and is very fun/enjoyable to work with, with very little effort tweaking prompts and settings. Faces are maybe just a tiny bit too cartoonish/CGI still (not by much and love the 3D depth it gives to everything else) and wish LoRA character likeness was just a little stronger. Otherwise, v6_NS is an amazing model like each of its predecessors. Honestly can't decide if I like this one or v5 or v4 the best. Think we need those 3 thrown in a blender ;)
Thanks for your continued work on Illu Renaissance, really appreciate it!
@AFD_0 Thank you very much! V6_NS is very different from the previous two versions because it has a very different prompting system. V5 is a direct evolution of v4, while V6_NS has exactly the same Unet as v5, but uses two different text encoders to be compatible with Pony and SDXL LoRas, in addition to Illustrious. One of the text encoders is also a modified Pony one who has been retrained to be compatible with Illustrious and not distort images at 1536p resolution. The change is a bit jarring at first, but once you get used to it, it does some really cool things and produces high-quality results. It clearly needs improvement in some areas; it's far to be perfect, of course. Especially regarding quality tags, it currently prefers Pony "score" and SDXL quality tags in some cases. In other cases, it even yields better and more realistic results without including any quality tags. And you shouldn't overload the negative prompt. But it lays the groundwork for future updates that I hope will be more stable and, I believe, even higher quality.
Any guidance for a huge fan @DeViLDoNia ? The model is amazing, but i need pointers on adjusting the prompting to prevent the daydreaming. I just posted an example of my output using version 4, 5, and 6ns using the exact same prompt.
"(simple dark background:1.5), ,Cyberdelia - illustrious_positive_v1,,
<lora:Size_Examination:0.8> upper body, close up, smug expression, 1girl, 1boy, standing male, kneeling female, veiny penis, long penis, handjob, 2 hands on penis, holding penis, penis over shoulder, penis covering mouth, obscured mouth, penis in front of mouth, nude male, naked male,nude"
@danewman
1. Start without a negative prompt, it usually gives better results. Then add what you don't want to see in the image. (I usually use "worst quality, low quality" as a negative).
2. If you overload the negative prompt, Renaissance v6_NS has a tendency to hallucinate, not always, but very often.
3. Do not use embeddings, I see that in your prompt you use a positive and negative embed (Cyberdelia's in this case). These files are optimized to perform better in their own models, in others they may work well, halfway or not at all.
4. Example prompt structure using yours as an example to get better results with Renaissance_v6_NS:
Quality & Realism: masterpiece, best quality, ultra detailed, photorealistic, realistic, solo focus, depth of field, film grain,
Characters & Pose: 1girl, 1boy, heterosexual, standing male, kneeling female, height difference, upper body, close-up, smug, expressionless,
Action & NSFW Tags: nude, male nude, handjob, grabbing penis, holding penis, two hands, large penis, veiny penis, long penis, penis on shoulder, penis in front of face, obscured mouth,
Technical & Environment: (simple background:1.3), dark background, <lora:Size_Examination:0.8>
I have uploaded a comparison with your same prompt and seed with the optimized prompt: Without negative prompt, with a sober negative prompt and a more aggressive negative prompt: https://civitai.com/images/120150702
I hope it helps you.
Turns out it was me character LORA that was causing it. When i removed it and kept the prompt the same PERFECT Results!
Nice, but it doesn't seem to do anything with lora's and completely ignore any IL lora loaded