3D ARTSTYLE LORA is on NoobAI-XL V-pred now!
For NoobAI-XL Epsilon / Illustrious 3D ART LORA visit this page.
LoRA specializes in creating artistic 3D images by adding lighting effects and knows a lot of characters.
Recommended model for this LoRA - NoobAI-XL V-pred 1.0.
If you can't use rescaling CFG features better use my 3D art lora for Epsilon.
VERSION 2 IS RELEASED!
What's new:
Replaced about 50% of dataset for better quality images;
More effects, more characters, more stable 3D style!
Now optimized for vanilla NoobAI-XL V-pred because Cyberfix V-pred models are not so good as Cyberfix NoobAI Epsilon models (CyberrealismXL has epsilon-prediction while NoobAI-XL V-pred has v-prediction). So I noticed that Cyberfix makes my pictures more like 2.5D photo than 3D.
I don't recommend for this LoRA Cyberfix versions anymore. Use pure NoobAI-XL V-pred 1.0 (Everything will be fine with anatomy, my dataset is big enough for good anatomy).
NoobAI-XL tags:
For version 2:
3d \(realistic\)
Optional:
shiny skin, wet, oiled, shiny clothes, shiny face
sidelighting, backlighting, underlighting, bloom, shadow, light paricles, sunlight, light rays, moonlight
No special negative.
For version 1:
Main tags:
3d
optional:
shiny skin, wet, oiled, shiny clothes
sidelighting, backlighting, underlighting, bloom, shadow, light paricles, sunlight, light rays, moonlight
helping tags:
photorealistic
negative prompt:
flat color, painting \(medium\), painterly, 2d, film grain
To prompt known franchise characters use their names. Don't add franchise anime / game names! It can harm 3D style!
If franchise character has too agressive baked-in style you can put his franchise game / anime in negative and 3D will be applied!
All images in my LoRA are tagged with danbooru tags! So use tags from this site: https://danbooru.donmai.us/
My preview can help.
NoobAI-XL 3D quality settings:
Put "masterpiece, best quality" in positive and put "worst aesthetic, worst quality, low quality, bad quality, lowres, extra digits" in negative.
Recommended generation settings:
Euler A SGM Uniform (Also you can try Uniform, Simple, Normal, DDIM, beta schedulers but I prefer SGM).
30 steps, CFG 5-6.
Rescale CFG 0.5 (Or other rescaling features. Use my guide abot V-pred if your pictures are oversaturated).
CFG++ samplers are not recommended this time. According to my test it harms 3D volumetricity. Better use Euler A and rescale CFG For V-pred.
Highres fix:
0.25-0.3 denoising strength, 12-15 hires steps,
Upscalers: Foolhardy Remacri 4x or DAT 2 x2 (VERY powerful but slow. His main advantage - saving many small details. It is not so critical for anime but very critical for realistic 3D).
Adetailer:
Denoising strength 0.25 - 0.3. Detection model confident threshhold: 0.7-0.8.
Inpaint only masked padding pixels - 70-90 pix for face models, 156-256 for hands models.
Please don't forget to rate my LoRA. This motivates me to do more!