We are happy to announce that version 0.5.1 of Matherix 3Dify is now available for download to all of our beta users. Thanks to all our beta testers who used version 0.5 and gave us this feedback. The changes from the previous verison are as follows:
- Freezing/not responding problem on some computers – fixed.
- Changed shortcut for Tools->Delete Selection to “Del” instead of the previous “Ctrl+T, D”.
- Fixed a bug in the “Save mesh” dialog – where the file was being saved to a ply file even though a different extension was selected in the drop-down. Now, there is no need to specify the extension explicitly in the file name.
- Added option to delete/crop selected regions while in the point selection stage using the “Tools->Delete Selection” and “Tools->Crop” options.
- Changed the naming convention of files stored by 3Dify to more readable ones.
This version can now be downloaded from our downloads page.
We are launching our second stage of Private-beta to a limited number(about 100) of users from today. We are looking for enthusiastic users to try our product in various scenarios and help us improve it. System and accessory requirements are as follows:
System Configuration: Intel Core2Duo (or) above with at least 2GB of RAM.
Operating System: Windows 7. Please try it if you have Windows XP or above.
Accessory: Kinect for XBOX.
Your feedback as early users will help us a lot in improving Matherix 3Dify in future releases. If you are interested, please fill this form. You will receive a copy of Matherix 3Dify v0.5 (private-beta) in your inbox. We will be rolling it out slowly, fixing bugs on the way. So, please be patient as we send you a copy of Matheirx 3Dify.
Please check our tutorials on how to get started.
If you have already recorded the Kinect stream and want to run 3Dify on this stream, you need to use “Data project”. As the name suggests, it uses already captured data for 3Dify. In case 3Dify failed to build a good model in the first attempt due to some reason, you can simply rerun by changing parameters like “interval” using the “data project” approach. Check out this video for more details:
Here is a simple video explaining the typical workflow and usage of Matherix 3Dify. There are five stages in 3Dify:
1> Capturing: the process of capturing data stream from Kinect.
2> Alignment: aligning multiple images’ data into a single model.
3> Noise removal: remove any stray noise in the model.
4> Meshing: building a triangular-mesh from the clean-up point-cloud.
5> Post-process mesh: lighting changes, cropping, filling holes in the mesh etc.,
For each of these stages, just follow the links to tutorials with more explanation and best-practices.
In this stage, after all the frames are aligned into a single model, we try and remove any noise in the model. This is done as explained in this video.
Some of the good practices are as follows:
1> Try and remove most of the noise for good meshing results.
2> Select only those regions where noise is present while thresholding so as to avoid removing some clean data points from other areas of the point-cloud.
This is a pretty simple and easy step to perform.
This is one of the most crucial stages of 3Dify. In this stage, 3Dify aligns the captured data frames into a single model. Chek out this video-tutorial for more info.
Some of the good practices for Alignment stage are as follows:
1> Choose start and end frames well
While choosing start and end frames, only select that region of the data you captured in which the subject is stationary. Any movement will greatly hinder alignment. Any area out of start and end indices will simply be ignored from processing. Also, make sure you avoid any frames where the subject is out of the capturing region either beacuse you have moved too close or too far from the subject.
2> Setting the interval
The default value of interval is 10. Interval, in alignment stage, is the maximum number of frames 3Dify would jump to look for alignment. So, if you moved slowly enough while capturing and want faster results, you may increase it to 20. If your Start-Index and End-Index differ by more than 1000 frames, you may choose to increase the interval for faster processing. If 3Dify failed to process the data when you tried it earlier, simply set the interval to a lower value like 3 or 5. This will slow down 3Dify but could result in more accuracy in the final model. So interval is a trade-off between speed and accuracy.
Capturing is the process of grabbing data from Kinect sensor to be used by Matherix 3Dify to build 3D model. Check out this tutorial video for more info.
Some of the good practices while capturing video are:
1> Segment the object of interest from the background using the slider.
At the start of the capturing process, you are given a slider to adjust the depth. By adjusting it, you can segment away the unnecessary background(shown in white) . Try and keep only the object of interest in the scene as shown below.
2> Avoid flat regions like a wall or floor.
It is always preferable to avoid flat regions like a wall or floor. In case, you cannot avoid it for some reason like a flower pot placed on the floor, a simple trick is to place some crumpled cloth around the object. The crumpled-ness creates more depth discontinuities (or) bumps which are actually beneficial in the Alignment stage. We can remove these crumpled-cloth sections from our final model at the end in the post processing stage. A rule of thumb is that more the non-flatness of the scene, better would the result be.
3> Avoid direct light at the Kinect.
Always try and keep any form of lights (natural or otherwise) behind your Kinect and not in front of it. Direct light at Kinect introduces lot of noise which can hinder in alignment and surely require more work at Noise-removal stage. It is a good practice to keep a good ambient light with no direct light at Kinect cameras. So, avoid scenarios shown below:
We will shortly put up tutorials for good practices for various types of objects like a face, shoe, furniture, plant etc.