RLY KREA 2 (Turbo) - Basic Photorealism Workflow

Just testing out some preliminary KREA 2 workflow parameters this morning for photorealism. Sharing to help others get off the ground quickly.

Text to Image w/Basic Latent Upscale.

--

Model

• Winnougan/Krea-2-Base-Turbo-NVFP4-FP8-INT8

Text Encoder — Load with CLIPLoader, type krea2:

• qwen3vl_4b_fp8_scaled.safetensors

VAE — Same as Anima:

• qwen_image_vae.safetensors

**Download links from Winnougan

Cheers,

RLY