ReconstructMe SDK 2.0 released

QReconstructMe2.0_scan2 Hello everyone,

one week after the launch of ReconstructMe 2.0, we are happy to announce that we’ve just made ReconstructMe SDK 2.0 publicly available. ReconstructMe SDK is the reconstruction engine used by ReconstructMe to perform 3D scanning in real-time.

The SDK update brings tons of new features and many new examples. Make sure you don’t miss the following sections!

3D Color Capture

The new ReconstructMe SDK enables you capture geometric and color information in real-time. The colorize example shows how you can capture a colorized 3D model with just a few lines of code. A resampling algorithm automatically optimizes the surface while maintaining its appearance as the following video shows

Multiple sensor handling

With our latest release we made using multiple sensors as easy as never before. Have a look at our body-scanning example that uses 2 sensors that work together and is written in less than 100 lines of code including calibration. Here’s a sneak peak at the result

Point and Shoot

Using ReconstructMe’s ability to perform global alignment, its easy to write an application that uses only key-frames for reconstruction. This is especially interesting for lower-powered devices such as note-books and tablets where real-time scanning is computationally too expensive. Our point and shoot example shows how trivial it is to implement

More Features and Bug-Fixes

Besides the features mentioned above here’s a short list on new things you’ll find in ReconstructMe SDK 2.0

  • Sensor Positioning We’ve added some intelligent sensor positioning techniques which can automatically detect the floor and construct an aligned volume around it. This technique was also for the experimental hidden feature in ReconstructMe 2.0
  • Setup Tuner The setup tuner will automatically adjust reconstruction parameters to the chosen volume size.
  • Performance Improvements The performance of alignment routines greatly improved by a module that allows a position forecast based on the tracked camera path.
  • Alignment Stability We’ve added an option to allow tracking even if the sensor is not looking at the volume at all. This feature is best used for static scenes.
  • Bugfixes We’ve fixed many sensor related issues. Intel HD series and AMD 79xx series are now supported.

The ReconstructMe SDK can be downloaded from our developer page.

13 thoughts on “ReconstructMe SDK 2.0 released

  1. RSP

    Tried ReconstructMe 2.0 and while some features are very good (larger field of view, color, mesh density optimization), overall the meshes are much rougher. There is a noticeably lower quality when using my PrimeSense RD1.09 sensor resulting in less detail. I’m reverting back to the old version ReconstructMeQt 1.2.95, hopefully the next build will boost the mesh quality.

    Reply
    1. Christoph Heindl

      Hi,

      this shouldn’t be the case. Maybe ReMe picked up the wrong GPU? Could you try to switch off auto-select GPU and select your device. Also try to adjust the scan volume as tight as possible to your needs, and try to adjust the tuning.

      Best,
      Christoph

      Reply
  2. Mangap

    Good job. I hope the developer keep improve the software.
    so we can use to create 3D models for animation, for technical drawing and others.

    Reply
    1. Christoph Heindl

      Thanks for the feedback :) Yes, we are receiving an increased interested in our SDK. Also from the 3D movie industry. Anything in specific you would like to see in ReMe?

      Best,
      Christoph

      Reply
      1. Mangap

        Have you prepare for new kinect ? which have better cameras/ resolution. please prepare for this and release it when ready. including the ability to reduce the details if we want less detail models. because too details sometimes is heavy on the animation.
        http://spectrum.ieee.org/automaton/robotics/robotics-hardware/heres-microsofts-new-kinect-sensor
        http://techcrunch.com/2013/10/02/how-microsoft-built-the-cameras-in-the-upcoming-kinect/
        http://www.xbitlabs.com/news/multimedia/display/20130815221115_Microsoft_s_Own_Chip_Enables_Ultimate_Precision_of_New_Kinect_Sensor.html

        Reply
  3. Dougal Harris

    Hi,
    I have installed version 2.0 and am finding it works much better than the original version – thanks. I am using it for a volunteer program scanning historical boats for a museum. Currently i need to do between 5 and 20 scans to get the entire hull of the vessel (as some boats can be up to 50m). Do you have plans for large scale scanning for a future release? Any idea on the time frame?
    Kind regards,
    Dougal

    Reply
    1. Mangap

      maybe new coming soon kinect will help. It has higher resolution. so you can scan from further distance and still get more parts scanned. CMIIW

      Reply
  4. Kim Lindaberry

    Please bare with me a minute . . . I really want to get setup so I can try this new version of ReconstructMe me. I haven’t used it before so I’m a bit uncertain exactly what input capture devices I can use. I’ve got a brand spanking new Lenovo with more than enough cpu & gpu power to handle it to do real time scanning. I’m prepared to buy a Kinect but I see that there is a new Kinect 2 on the horizon so I’m hesitant to go out and buy a brand new Windows Kinect until the release of the newer model comes out. Does this new version of ReMe support the MICROSOFT – KINECT FOR XBOX 360 or any xBox Kinect device? I can find dozens of slightly used Kinect 360 xBox units for sale for $20-$50 to get me going if they will work? Do you know what the limitations might be for the XBox 360 if it does work? Will the new Kinect 2 be so much better in final resolution? In your opinion currently what is the best capture device for scanning full figures with the best detail? . . . Thanks . . .

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *