cobb7127 wrote:

Hi,

I have only started using APP in the last few weeks and am still quite new to it so I am not sure if what I am trying to use it for is possible.

For work I have been given videos of concrete pylons. The video is taken sweeping side to side, to the top of the pylon using a JVC action cam. I then have split up the video in to stills, leaving 200-300 pictures. The idea is to then use APP to generate a picture of the whole pylon using these images.

I have not been able to get any decent results yet. Im not sure whether the similar pattern in the concrete confuses APP when it tries to make links, giving a 'mashed' up image.

Are there any settings I should use to try an overcome the pattern similarity or other issues? am I using too many images?

I have used the program to generate landscape panoramas so im beginning to think that what I'm trying to do might not be reasonable.

Thanks for any help, much appreciated.

Hi, cobb7127 (is that your name? )

What you´re talking about is named "object-vr" - and that is very different from a panorama. But you might give it a try using the KRPAno-tools´ object-vr functionality.
Or you might try a planar projection-mode in AutoPanoGiga.
But most unlikely you´ll get an acceptable result because of the very limited resolution of the action-cam and because you didn´t use an appropriate camera-mount.

best, Klaus

PS:

I remember Microsoft provides a application where you can put in dozens or hundreds of images you shot from an object - without any correct mounting or shoot-strategy -
and it combines them all to a 3D or 3D-like object. Don´t remember the name - but i´m sure it can be found.

Last edited by klausesser (2013-02-05 13:00:36)

klausesser wrote:

I remember Microsoft provides a application where you can put in dozens or hundreds of images you shot from an object - without any correct mounting or shoot-strategy -
and it combines them all to a 3D or 3D-like object. Don´t remember the name - but i´m sure it can be found.

You are probably thinking of PhotoSynth, or maybe Microsoft ICE (which apparently supports stitching from video with Windows 7)?:

Photosynth:  http://photosynth.net/default.aspx

Microsoft ICE - :

http://research.microsoft.com/en-us/projects/ice/

http://en.wikipedia.org/wiki/Microsoft_ … ite_Editor

http://hdview.wordpress.com/2011/04/05/ … -blending/

Last edited by mediavets (2013-02-05 13:26:37)

cobb7127 wrote:

Thanks for your replies.

I have uploaded a sample video to https://www.dropbox.com/s/laye8eeqhgysb … orama..MP4 which hopefully works. Our process has been to crop the video appropriately then to break up into stills, delete 9 out of every 10 frames so to make it manageable for APP to upload/detect etc. Hopefully this will give you a better idea on what I am trying to do and whether it is possible.

I looked at Microsoft ICE but could not get any decent results unfortuanely.

Thanks, cobb7127

Ah, the videio material is not quite what I expected.

Do you have a complete image of the concrete pylon so I can better unederstand the 'context' of the video?

What is the goal of this activity?

I wonder if it has any similarities to this project which involves monitoring the structural integrity of a very large concrete structure?:

http://www.kolor.com/forum/t10563-using … n-cpc-1100

cobb7127 wrote:

I have drawn up a diagram (click the link below) that tries to show the process.

I don't seem to be allowed access to the diagram: