Z-Image Turbo Controlnet

Fast 6B model generating images from text and control images using edge, depth, or pose guidance.

Overview

Z-Image Turbo is a super-fast 6 billion parameter model designed for image-to-image generation. It creates images conditioned on both text prompts and control images such as edges, depth maps, or pose outlines using ControlNet techniques. The model supports various preprocessing options (none, canny, depth, pose) to condition the generation process and can produce multiple images per request.

Strengths / What it does well

• Generates detailed images conditioned on text prompts and control images
• Supports multiple preprocessing modalities: edge detection, depth maps, and pose
• Enables fine control over generation through parameters like control scale and inference steps
• Can produce up to 4 images per request with configurable output formats (PNG, JPEG, WebP)

Best use cases

• Creating images that closely follow user-drawn outlines or poses
• Enhancing text-to-image generation with additional structural or spatial guidance
• Applications in art, design, and creative workflows requiring controlled image synthesis