RizVision-V1 – Enhanced ZImage Turbo Checkpoint

V1 is a refined checkpoint built on the powerful ZImage Turbo base model, designed to deliver vibrant colors, balanced lighting, and natural-looking details while maintaining the fast generation speed of turbo models.

This version integrates DPO-based optimization, improving color depth and lighting dynamics — areas where the base ZImage Turbo model can sometimes lose visual richness, especially when stacking multiple LoRAs.

The goal of V1 is to produce visually striking yet natural images with improved clarity and reduced artifacts.

Key Improvements

Enhanced Color Vibrancy V1 improves overall color saturation and depth, making images look more lively and visually appealing without becoming overly artificial.
Better Lighting & Contrast Optimized light balance helps create more cinematic and eye-catching outputs.
Cleaner Skin Texture Special attention was given to reducing the common “liquid” skin effect often seen in low-step turbo models. Skin now appears slightly more matte and natural, closer to real-life textures.
Stable with LoRAs The model maintains color stability even when multiple LoRAs are applied, preventing washed-out outputs.

Workflow Compatibility

V1 works smoothly with:

ZIT Simple Workflow • ZIB + ZIT Workflow

These workflows allow you to fully utilize the speed and efficiency of the ZImage Turbo ecosystem.

Recommended Settings

Sampler: DPM++ SDE / Euler Steps: 4 – 8 CFG: 1.5 – 2.5 Resolution: 1024×1024 or higher

Turbo models perform best with low steps, so avoid using high step counts.

Notes:

If you want even stronger color vibrancy, you can experiment with additional DPO LoRAs. However, stacking too many enhancements may lead to over-saturated outputs, so adjust carefully.

Final Words

V1 is designed for creators who want fast generation, vibrant visuals, and natural-looking details without sacrificing image quality.
Thank you for trying the model and supporting the project ❤️

[ZIT] RizVision - V1-DPO-bf16