Z Image Turbo Controlnet

Z Image Turbo Controlnet

Playground

Z-Image-Turbo ControlNet generates images guided by structural control signals (depth, canny edge, pose) for precise composition control. Ready-to-use REST inference API, best performance, no cold starts, affordable pricing.

Features

Z-Image Turbo ControlNet is a powerful image generation model that gives you precise control over composition through structural guidance signals. Unlike standard text-to-image models that interpret prompts freely, ControlNet lets you define the exact structure, edges, depth, or pose of your output by analyzing a reference image.

Think of it as a blueprint system: you provide a reference image, choose how to analyze it (depth map, edge detection, or pose estimation), and the model generates a new image that follows that structural blueprint while matching your text prompt.

Why Choose This?

• Precise composition control Define exact layouts, poses, and spatial relationships instead of hoping the model interprets your prompt correctly.

• Multiple control modes Choose depth mapping for 3D structure, canny edge detection for outlines, pose estimation for human figures, or none for standard generation.

• Reference-guided generation Use existing images as structural templates while completely changing style, content, and appearance.

• Flexible strength control Adjust how strictly the model follows the control signal — from loose inspiration to exact replication.

• Fast and affordable Turbo-optimized for quick generation at just $0.05 per image.

Understanding Control Modes

The mode parameter determines how the model analyzes your reference image:

Parameters

How to Use

1. Upload reference image — The image that defines your composition structure.
2. Choose control mode — Select depth, canny, pose, or none based on what you want to preserve.
3. Write your prompt — Describe the style, content, and appearance you want.
4. Adjust strength — Higher values follow the control signal more strictly.
5. Set output size — Define your target dimensions.
6. Run — Submit and download your controlled generation.

Pricing

$0.012 Per image. Simple flat-rate pricing regardless of control mode or image size.

Best Use Cases

• Architectural Visualization (depth mode) — Maintain spatial relationships while changing style or materials.
• Character Posing (pose mode) — Generate characters in specific poses from reference photos.
• Style Transfer with Structure (canny mode) — Apply new styles while preserving exact outlines and shapes.
• Product Photography (depth mode) — Generate product images with consistent composition across variations.
• Comic and Illustration (canny/pose modes) — Convert sketches or poses into fully rendered artwork.

Pro Tips

• Depth mode works best with images that have clear foreground/background separation.
• Canny mode is ideal when you have line art, sketches, or want to preserve exact shapes.
• Pose mode requires images with visible human figures — it won’t work on landscapes or objects.
• Start with strength 0.6 and adjust: lower for loose interpretation, higher for strict adherence.
• The prompt matters more at lower strength values; at high strength, structure dominates.
• Use the same seed to compare different control modes on the same reference image.

Notes

• Reference image quality affects control accuracy — clear, well-lit images work best.
• Pose mode only detects human poses; it won’t extract structure from other subjects.
• At strength 0, the control signal has minimal effect (similar to standard generation).
• At strength 1, output will closely match the reference structure regardless of prompt.

Related Models

• Z-Image Turbo Text-to-Image — Standard text-to-image without structural control.
• Z-Image Turbo Image-to-Image — Transform existing images with variable strength.
• Z-Image Turbo Inpaint — Edit specific regions of images with masks.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "prompt": "A cinematic ocean wave at sunrise, highly detailed",
  "image": "https://interactive-examples.mdn.mozilla.net/media/cc0-images/painted-hand-298-332.jpg",
  "mode": "depth",
  "size": "1024*1024",
  "strength": 1,
  "seed": -1,
  "output_format": "jpeg"
}
JSON
)

# 1. Submit the prediction.
SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  -X POST "https://api.wavespeed.ai/api/v3/wavespeed-ai/z-image-turbo/controlnet" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}")

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id
' >&2
  exit 1
fi
RESULT_URL=$(printf '%s' "${TASK}" | jq -r '.urls.get // empty')
if [ -z "${RESULT_URL}" ]; then RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"; fi

# 2. Poll until the prediction finishes.
while true; do
  RESPONSE=$(curl --silent --show-error --fail-with-body \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}")
  RESULT=$(printf '%s' "${RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout) printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    created|processing) sleep 2 ;;
    *) printf 'Unexpected status: %s
' "${STATUS}" >&2; exit 1 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Response Parameters

Result Request Parameters

Result Response Parameters