🔥 ZIT+XL Workflow by Stable_Yogi🔥

✨ Z Image Turbo → DMD2 Hybrid Workflow

A fast 3-pass workflow combining Z Image Turbo for the base image + Stable Yogi DMD2 models for detail, realism, and upscaling.

How It Works

1. Pass 1 – Z Image Turbo:
   Builds a clean base (pose, anatomy, colors).

2. Pass 2 & 3 – DMD2 Realism Models:
   Enhances structure, adds texture, and performs latent upscaling.

3. Final Face/Eye Detail + Tile Upscale:
   Sharpens the final image and boosts resolution.

Produces high-res, detailed images in ~15 sec on a 4090.

---

Model Setup

Place models like this:

ComfyUI/models/
├── text_encoders/qwen_3_4b.safetensors
├── diffusion_models/z_image_turbo_bf16.safetensors
├── vae/ae.safetensors
├── loras/(Realism LoRAs + qinglong_detailedeye_z-imageV2)
└── checkpoints/(Your SDXL / DMD2 checkpoints)

Use up to 4 LoRAs.
CFG = 1.

---

Recommended Models

• Realism By Stable Yogi Pony V6 (Patreon Pro C42)

• Realism By Stable Yogi Pony V6 – CivitAI

• Realism LoRA BSY V3 Pro or CivitAI version

---

Customization

Edit only the yellow-boxed nodes on the top-left.
Changing upscale/step values may slow the workflow heavily.

---

Notes

• Designed specifically for DMD2 models.

• SDXL checkpoints go in models/checkpoints.

• Lower-VRAM cards not tested—share results in comments.

🆘 Having Issues?

🎮 Join Discord : https://discord.gg/DZEenb5wGc

🔗 Get the Pro Version of This Workflow with More tweaks and enhancements on My Patreon.