Beta: OpenCapture

Which Codec did you choose to save the video?

Merci Patrice! J'espère que tu pourras faire tout ce que tu voulais faire.
Je vous prépare une version 0.7 avec la possibilité de tracker un objet (typiquement le visage).
Et ensuite je m'attaquerai à la prise en charge des caméras 3D (depth map). Et pourquoi pas la reconnaissance des gestes si je suis en forme?

Hello,

Just a little video to show the work in progress for this extension:
[video=youtube_share;RYdR9PdszOY]Please login to see this media element.]

It shows an example of 2 playing cards tracked one after each other. The feature tracking is not as robust as I wished but it is already a first step for Augmented Reality.
I finished this version at least 2 weeks ago but I did not release it for 2 reasons:
- I was not completely satisfied how robust the feature tracking is.
- I still have a bug with the [Make object invisible] action in Standard mode. It does not hide the camera image and keep the last frame captured. In Direct3D, it works perfectly. If some of you have any idea?

I can upload this version if some of you want tor try Face Tracking and Feature tracking or want to see the known bug to help me debug this.

For the bug, I already asked Yves but his first suggestion did not work. He is certainly overbooked, that is why I am asking other developper on the forum if they already faced to such a problem.
I hope it is not a concurrency problem. The trick with the action [make object invisble] is that I have no control over it and I can't debug this properly with the SDK.

You can put any active objects behind or over the camera. It is not shown on the video but with my example file, 2 active objects are moving with the corresponding playing card. With 3D objects, you can calculate the perspective transform of the card (with the 4 corner positions of the green polygon) and display them in the corresponding orientation.

Fanotherpg: The way I implemented the camera display does not require any HWA functionalities, therefore I let the Standard mode available. In HWA mode, the camera surface should be converted in a HWA texture so it takes more time. I can enable the ink effects but I am not convinced that it is a "must have" feature for this extension. On the other hand I can let the gpu do more calculation (for instance histogram equalization) through the availbale Opencv functions. But it is a lot more work and I do not want to begin with something like this at the moment.
copperbob and Ham: I will compile the extension in release mode and upload today or tomorrow.

Here is the version 0.7: Please login to see this link.
Short Release Notes for v0.7:
- refactored code for a full multithreaded extension (all camera processes are performed in parallel). Therefore the framerate of the application is not directly linked with the framerate of the camera anymore.
- Added face tracking with not patended cascade classifier file
- Added feature tracking with not patended algorithmus
- Added mirror image function in real time
- Added some filtering in color tracking to increase performance
- Fixed the bug with [Make Object Invisible] action

Since I did not make it at the beginning of this thread, I want to thank Yves for his great help on this extension. He helps me a lot at the beginning and at the end with this tricky bug.

The extension is 100% royalty free and can be used for commercial applications. The only limitation is the license of OpenCv (see help file) which requires a license chapter to add in your documentation.