Hunyuan3d V2 Multi View

Hunyuan3d V2 Multi View

Playground

Hunyuan3D V2 Multi-View generates accurate 3D reconstructions from multiple images. Tencent-developed and available on WaveSpeedAI. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

Hunyuan3D V2 Multi-View is part of Tencent’s open-source Hunyuan3D-2 series — a state-of-the-art 3D generation system that creates high-fidelity 3D models from multiple reference images. Provide front, back, and left views of your subject for accurate 3D reconstruction with detailed geometry.

Why It Stands Out

• Multi-view input: Uses front, back, and left images for more accurate 3D reconstruction.
• High-precision geometry: Multiple viewpoints enable better shape capture and detail.
• High-fidelity output: Produces detailed 3D models with accurate geometry and high-resolution (4K) textures.
• Fast generation: Completes model generation in as fast as 30 seconds.
• Decoupled architecture: Separates geometry generation and texture synthesis for improved quality.

Technical Highlights

The Hunyuan3D-2 system adopts a separated process of geometry generation + texture synthesis:

• Geometry Generation (Hunyuan3D-DiT): Based on a flow diffusion model that generates untextured 3D geometric models, with 2.6B parameters, capable of precisely extracting geometric information from input images.

• Texture Synthesis (Hunyuan3D-Paint): Adds high-resolution (4K) textures to geometric models, with 1.3B parameters, supporting multi-view diffusion generation technology to ensure realistic textures and consistent lighting.

By decoupling shape and texture generation, the system effectively reduces complexity and improves generation quality.

Performance and Efficiency

• Fast Generation: Completes model generation in as fast as 30 seconds.
• Accelerated Inference: The optimized version shortens inference time by 50% through guidance distillation techniques.
• Multi-modal Support: Compatible with various integrations including Blender plugins and Gradio applications.

Parameters

How to Use

1. Upload front view image — the front-facing view of your subject.
2. Upload back view image — the back-facing view of your subject.
3. Upload left view image — the left side view of your subject.
4. Click Run and wait for your 3D model to generate.
5. Preview and download the result.

Best Use Cases

• Character Modeling — Create 3D characters from reference artwork or photos.
• Game Assets — Generate game-ready 3D models from concept art.
• Product Visualization — Generate 3D models of products from multiple angles.
• 3D Printing — Generate printable models from multi-view references.
• AR/VR Content — Create 3D objects for immersive experiences.
• Animation — Build 3D characters and props from design sheets.

Pricing

Pro Tips for Best Quality

• Use consistent lighting across all three reference images.
• Ensure the subject is centered and at similar scale in each view.
• Use clean backgrounds (ideally transparent or solid color) for better results.
• Align viewpoints accurately — front, back, and left should be 90° apart.
• For characters, include clear details of face, clothing, and accessories in each view.

Notes

• Ensure all uploaded image URLs are publicly accessible.
• Processing time varies based on current queue load.
• Please ensure your content complies with usage guidelines.

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'
{
  "front_image_url": "https://interactive-examples.mdn.mozilla.net/media/cc0-images/painted-hand-298-332.jpg",
  "back_image_url": "https://interactive-examples.mdn.mozilla.net/media/cc0-images/painted-hand-298-332.jpg",
  "left_image_url": "https://interactive-examples.mdn.mozilla.net/media/cc0-images/painted-hand-298-332.jpg"
}
JSON
)

# 1. Submit the prediction.
SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  -X POST "https://api.wavespeed.ai/api/v3/wavespeed-ai/hunyuan3d/v2-multi-view" \
  -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