3D scanning by photos taken with a simple smartphone

Example of composed scene (scaned objects + modeled objects)

3D scanning with photos by Structure-from-Motion (SfM) and Image Based Modeling (IBM) has provoked the admiration and has pointed the curiosity of the people.

Some of these people want only to test randomly the technology, but other have objective ambitions, like to reconstruct the architectural heritage, art for advertising and even forensic research that already has been shown here, in other post.

The objective of this blog is share the knowledge of all people that want to learn about free software technology. Thinking about a great number of individuals, today will be shown the result of scanning with the camera of a simple smartphone, the Galaxy Y Duos, from Samsung.

The next phase of the researches will be the collect of photos of other cell phones and smartphones more and less sofisticated than this presented here.

Configuration

Before anything else, is necessary know that the scanning is not made inside the smartphone, but in a personal computer with PPT-GUI installed. A few months ago a post was published here with a tutorial of 3D scanning by photos, that you can follow to make your own test.

The objective of this post is prove that the result is literally at your fingertips. Even the hardware you have not be the most sophisticated (sorry, the version of screenshots are only in Portuguese).

 

In this case, the model used is a GT-S6102B, with Android 2.3.6. It is a quite simple smartphone sold in Brazil.

All the photos was taken with default configuration of illumination.

Evidently, anyone scanning was made with photos taken during the night.

Some of the scenes was in sunny environment. Some of them in environment with shade of sun, and other in internal of houses.

The only change in the default configuration was use the 3.2 megapixels the total power of the camera, with 2048x1536 pixels.

The result was shown in the video on top of this post.

I hope you enjoyed.

See you in the next!

Tweet