Replace outdated WAI Illustrious image model with Anima Turbo

For anime workflows, the version of WAI currently with by Venice pro is outdated and produces subpar images. Anima Turbo 1.1 is similarly trained on booru datasets, but has a more recent cutoff and accepts natural language prompts alongside standard tag-based prompts. It’s also more performant than WAI, able to create high quality output in only 8-12 steps. Prompts are transferrable from WAI to Anima with very slight alteration, so it could easily be used as a drop in replacement.

Anima huggingface page: circlestone-labs/Anima · Hugging Face

Features supported:

• Positive and negative prompts

• Tag list and natural language prompts

• LoRAs and artist styles

Key points for Turbo:

• CFG 1.0

• 8-12 Steps

• Prepend positive and negative quality prompts

Models:

• anima-turbo-v1.1.safetensors (Diffusion model)

• qwen_3_06b_base.safetensors (Text encoder)

• qwen_image_vae.safetensors (VAE)