Anima-Preview2 Turbo-LoRA (14-Step) and Anima-Preview2 Turbo-LoRA (8-Step)
🚀 Overview
This is the 14-Step Turbo-LoRA and 8-Step Turbo-LoRA for Anima-Preview2.
This versions are fun and experimental.
This "fun version" demonstrates the turbo-training capabilities within the Anima architecture. 14-step LoRA was distilled on 227 Anima-Preview-2 latents over 2000 steps.
🛠 Recommended Settings
To get the best results with 14 step lora, use the following configuration:
| Setting | Recommended Value |
|---|---|
| Steps | 14 |
| CFG Scale | 1 |
| Samplers | er_sde or res_2m |
To get the best results with 8 step lora, use the following configuration:
| Setting | Recommended Value |
|---|---|
| Steps | 6-8 |
| CFG Scale | 1 |
| Samplers | er_sde or res_2m |
14 step example
8 step example
⚖️ License
Usage is governed by the Anima-Preview2 license terms.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support