SentiSculpt SDK

Software information
Windows 7/Vista/XP/2000/Server
File size:
151.01 Mb
Date added:
March 20, 2015
Product page:
Description from the Publisher

SentiSculpt is designed for software developers who want to use computer vision-based 3D object model reconstruction from images in their applications. Object models are created automatically from sets of photos, which were captured from different angles using regular cameras.

SentiSculpt is available as a software development kit that provides for the development of 3D object model creation systems for Microsoft Windows.

  • Series of object photos are used for automatic creation of 3D object model.
  • Off-the-shelf cameras or smartphones used instead of 3D scanners.
  • Reconstructs point clouds, generates meshes, performs texture mapping.
  • Resulting models can be stored in widely supported file formats.
  • Available as an SDK that supports multiple programming languages.
  • Reasonable prices, flexible licensing and free customer support.

SentiSculpt technology is designed for rapid creation of 3D object models from series of 2D images. The technology is intended to work with common cameras instead of expensive 3D scanners. The generated models are suitable for wide range of applications, including virtual or augmented reality, video games creation, 3D printing, computer aided design, architecture etc. The object size may range from a small toy to a large building.

The SentiSculpt technology has these capabilities for 3D object model reconstruction:

  • Converts series of photos into 3D model. SentiSculpt needs series of object photos, which are made from different angles. Each photo is analyzed by advanced computer vision algorithms, and certain features of the object are captured for the point cloud creation. See the recommendations and specifications for the photography-related details.
  • Common camera required. The photos for the 3D object model reconstructions can be made with an off-the-shelf camera or a smartphone. A tripod can be used for improving photography quality. The optimal model creation performance needs 2-5 megapixel camera resolution. The recommendations and specifications contain more detailed information about camera setup.
  • Automatic images processing. Common 3D model creation may need many hours of work by a trained designer. SentiSculpt algorithm provides simple 3D model creation workflow and does not require advanced skills, as its input is usually a folder with images. The algorithm analyzes each image, recreates the shooting angle, extracts the necessary features and adds them to the object model.
  • Advanced 3D model creation. The technology performs several steps of object model creation. A point cloud is created at first. A mesh (wireframe model) is generated based on the point cloud. A texture is recreated from the captured object surface and applied to the generated model, so the final model has realistic look.
  • 3D model export into widely supported formats. SentiSculpt allows to save the generated object model in widely used file formats so the model can be used by most 3D graphics and CAD applications.

SentiSculpt SDK contents

SentiSculpt SDK is designed for developers who want to use computer vision-based 3D object model reconstruction in their applications. The SDK allows rapid development of computer vision-based 3D object model reconstruction systems using functions from the SentiSculpt library for Microsoft Windows platform (see also the system requirements).

SentiSculpt SDK distribution package contains these components for Microsoft Windows (64-bit):

  • SentiSculpt installation license (1 single computer license)
  • Device manager library
  • Programming samples for C, C++ and C# languages
  • SentiSculpt SDK documentation

Recommendations and Specifications

  • 2 - 5 megapixel camera resolution is recommended for optimal reconstruction performance and 3D model quality. Larger resolution will slow down model creation performance, so users should consider if it is acceptable for them. Model quality can be also improved by using larger number of photos from different angles.
  • Tripod usage is highly recommended to avoid blurry images.
  • Camera settings should be adjusted before making photos of an object:
    • The auto focusing should be disabled.
    • Avoid zooming in/out between shots of the same session, constant focal length should be used. Digital zoom should be also avoided in the same session and used with care in general, as it leads to all kinds of distortions, noise and reduction in the actual image resolution. If you need more details, you can go closer to the object.
    • Flash should be disabled to avoid reflections on the object's surface.
  • Constant lighting condition are recommended, as variations in light during the same session can lead to an inaccurate reconstruction results, or even to a failure.
  • Circular trajectory is recommended for object capture. The camera angle can change by about 15 degrees in horizontal plane for each shot, resulting in about 24 shots of the object. Images must overlap by approximately 2/3. Same features of the object should be visible on at least 3 adjacent photos.
  • Object model reconstruction may be problematic in these cases:
    • objects with shiny surfaces;
    • partially or fully transparent objects (glass, windows, water);
    • periodically textured objects (like chess board);
    • objects without any texture (plain painted walls).

System requirements for SentiSculpt SDK

  • PC with x86-64 (64-bit) processor:
    • SSE2 support is required. Processors that do not support SSE2 cannot run the SentiSculpt algorithm. Please check if a particular processor model supports SSE2 instruction set.
    • SSSE3 support is recommended, as the SentiSculpt algorithm provides higher performance using this instruction set. Please check if a particular processor model supports SSSE3.
    • 32-bit architecture is not supported.
  • At least 1 GB of free RAM should be available for the SentiSculpt-based application. Additional RAM may be required for working with larger images and larger sets of images, as well as larger object models.
  • Microsoft Windows XP / Vista / 7 / 8 / Server 2003 / Server 2008 / Server 2008 R2 / Server 2012, 64-bit only.
  • Microsoft .NET framework 3.5 or newer (for .NET components usage).
  • Microsoft Visual Studio 2008 SP1 or newer (for application development under C/C++, C#, Visual Basic .Net)
Users reviews & testimonials

This software is not reviewed yet.

Other downloads from this publisher

Real-time 3D face motion capture and tracking SDK for development of augmented reality apps, digital character facial expression control or mapping animated avatars on users face. Works on common PC with webcam or Android smartphone.

Real-time biometric face recognition and pedestrian/vehicle tracking SDK for surveillance systems and networks. Supports up to 10 cameras on one PC. Automatic identification against watch list, gender/emotions detection. See:

Eye movement tracking with webcam and PC, no specialized hardware required. The technology demo app shows real-time gaze tracking and heatmap generation for research of single and multiple subjects. A software development kit (SDK) is also available

NCheck Bio Attendance is an automatic end-user time-tracking system with fingerprint, face and iris biometrics. Protection against errors or user fraud. Client-server architecture. Reports generation. Webcams, 90+ biometric scanners models supported.