Experiments/3D: Difference between revisions

From Open Food Facts wiki
Line 19: Line 19:
=== 3D scanner we could buy ===
=== 3D scanner we could buy ===
* https://www.artec3d.com/3d-scanner/artec-eva
* https://www.artec3d.com/3d-scanner/artec-eva
=== Cool demo we could hide in the OFF app ===
* With just spinning, you could make a Pepper's ghost: https://www.youtube.com/watch?v=zkgy-WcflWE (but that's silly)
It works much better with cylinders like cans than with more complex shapes, because it's still just four images. @aleksejrs
===Photogrammetry===
* Making 3D models out of photos is https://en.wikipedia.org/wiki/Photogrammetry
* The only open source program for that I've seen was abandoned and so had annoyingly incomplete UI and crashed often.  Not a very popular activity.  http://insight3d.sourceforge.net/ (edited)
* There was also a page about doing it using many different programs (IIRC, that included one used for making panoramas with Hugin), but that was complex (also, the most difficult part was removing the incorrect dots).  An example was making a 3D model of a part of the earth surface from ISS photos.


==Generating usable data==
==Generating usable data==

Revision as of 10:30, 15 June 2017

<<Back to Experiments

This project aims to prepare Open Food Facts for the advent of 3D and virtual/augmented reality.

Inspiration

3D models of food in a dystopian future

Google launches the tango phone - allows for 3D scanning

3 models of food

3D scanner we could buy

Cool demo we could hide in the OFF app

It works much better with cylinders like cans than with more complex shapes, because it's still just four images. @aleksejrs

Photogrammetry

  • Making 3D models out of photos is https://en.wikipedia.org/wiki/Photogrammetry
  • The only open source program for that I've seen was abandoned and so had annoyingly incomplete UI and crashed often. Not a very popular activity. http://insight3d.sourceforge.net/ (edited)
  • There was also a page about doing it using many different programs (IIRC, that included one used for making panoramas with Hugin), but that was complex (also, the most difficult part was removing the incorrect dots). An example was making a 3D model of a part of the earth surface from ISS photos.

Generating usable data

Getting flat photos of products we can map on preexisting standard models

Using photos from Daniel's machine

Straightening existing photos to maps them onto models

Storing the data

We need to be able to differentiate 3D ready photos, photos that can be transformed on the fly, and unusable photos.

Visualising data in 3D