Gladas' Workflow for Krea 2 Turbo — v1 — v11

Use this Krea 2 Turbo workflow to turn a text prompt into an image, with wildcard and multiple LoRA support, then optionally upscale and save the result.

An optional second-pass latent upscale has separate sampling settings; when latent upscaling and USDU are both enabled, the latent result takes priority.

SeedVR2 is an optional final cleanup pass; the face and eye detailers and final cleanup are bypassed by default.

Set a Width × Height preset and separate scale values for initial generation and latent upscaling; saved images can include metadata in a CivitAI-readable format.

The download is an editor workflow file named gladasWorkflowForKrea_v11.json.

The workflow lists these node packs: bkwildcards, comfyui_ultimatesdupscale, comfyui-custom-scripts, comfyui-image-saver, comfyui-impact-pack, comfyui-impact-subpack, comfyui-kjnodes, and rgthree-comfy.

Listed model files include Anzhc Face seg 1024 v2 y8n.pt, bbox/Eyeful_v2-Individual.pt, betterDaysIllustriousXL_illuXLV02BAKEDVAE.safetensors, qwen_image_vae.safetensors, qwen3vl_4b_fp8_scaled.safetensors, sam_vit_b_01ec64.pth, seedvr2_3b_int8_convrot.safetensors, seedvr2_ema_vae

Initial generation defaults to Euler + Simple, 8 steps, and CFG 1; USDU defaults to Euler + Simple, 3 steps, and denoise 0.24.

USDU defaults to 1536×1536 tiles with 128 padding; connecting image dimensions to the tile-size inputs enables full-image single-tile processing. The latent upscale target follows the resolution preset independently of the initial-image scale.

Before setup, verify that these exact model filenames are available: Anzhc Face seg 1024 v2 y8n.pt, betterDaysIllustriousXL_illuXLV02BAKEDVAE.safetensors, and vxpKrea2Nsfw_beta3INT8.safetensors.

Suggestion · not verified

This workflow is not set up for models using CFG above 1; upscale, detailer, and cleanup settings also vary with your finetune and hardware.