Instructions to use Bl4ckSpaces/Ilustrious-Nova-V3-TIES with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use Bl4ckSpaces/Ilustrious-Nova-V3-TIES with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("Bl4ckSpaces/Ilustrious-Nova-V3-TIES", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
Ilustrious-Nova-V3-TIES (God Tier Merge)
Ilustrious-Nova V3 represents the pinnacle of merging technology using Task Arithmetic (Delta Merging).
Instead of mixing models like paint, we extracted the "knowledge vectors" (deltas) from IllumiYume and NoobAI relative to SDXL Base 1.0, and surgically injected them back together.
The V3 Recipe (Gene Editing)
- Canvas: SDXL Base 1.0
- Vector A (Style): IllumiYume v3.5 Delta (High priority in Output Blocks)
- Vector B (Anatomy): NoobAI XL v1.0 Delta (High priority in Input Blocks)
Why V3 is Superior?
- No "Washed Out" Colors: Since we add vectors instead of averaging them, the saturation remains pristine.
- Distinct Knowledge: The model distinctly knows "How to Pose" (from NoobAI) and "How to Color" (from IllumiYume) without blurring the two concepts.
Recommended Settings
- Sampler: Euler a / DPM++ 2M SDE
- CFG: 6.0 - 8.0
- Steps: 30+
- V-Prediction: ENABLED
- Downloads last month
- 18