Hi All,
I hope i have put this in the correct place.
I am trying my hand at the camera tracking in blender and I have followed a couple of tutorials and keep coming to an error. In the attached image you can see that the markers in the clip editor do not match those in the 3d view.
Any help would be greatly appreciated.
For info the camera used was a HTC One X and using Blender 2.63a.
Thanks
Jon
Ok first of your awesome because your using ubuntu instead of windows :D. Ok there is “Error” you will always get that when camera tracking that is a measurement of how much the trackers slide around the higher the number the less accurate the track is the lower the more accurate so hope that helped there is no bug or anything wrong with blender 
Thats super good news huh
Thanks for the thumbs up for Ubuntu. You just can’t knock an is that works and doesn’t crash.
I have run a solve on the tracking and its around 1.6 which is ok. But when I add the camera constraint all the points are bunched up next to the camera not spaced out like. What’s in the clip editor.
I was about to post just before the current went out. Oh well…
In short, your focal length and sensor size values are incorrect. I do not know the correct focal length for that particular phone, but it tends to be roughly between 3.5-4.5 mm for most smartphones. The sensor size, which can be found here is sqr(4.54^2+3.42^2) = ~5.68 mm - given the initial values (4.54 x 3.42 mm) on that page.
Thanks for the info. However when I run the solve again it comes to around 9.5 which is a bit bad. Not only that the camera seems to be placed somewhere else in the scene. and I can’t seem to be able to move is back to the object.
Do you have any perspective shift a.k.a. parallax in that shot? I suppose if you’ve already watched a few tutorials, it must have been mentioned that a non-tripod shot must have a certain degree of parallax. Basically, there must be visible translation between points in the footage for Blender to do a proper solve.
@Daccy: hrm… if the sensor width on your link is stated as 4.54, why would you give him a diagonal measurement to use when the tracker wants width?
anyway, that link also gives a lens fl of 28 ( which is probably a 35mm equivalent…) I would suggest trying a sensor width of 36* and focal length of 28
*the sensor width of a 35mm camera is actually 36mm, 35mm refers to the width of the film -> a 35mm negative is 36mm x 24mm
edit: have you tried using the refine options?
My bad! I’m not sure what I was thinking at the moment… 
Is it really possible to get a correct solve if the equivalent values are used, such as those you mentioned, and not the actual ones? If so, does it mean that the ratio between sensor width and focal length is more important than the actual values?
Daccy,
the last tut i saw was the one that andrew price did and this did mention about parallax but the footage is a sweeping movement from left to right and there is no obvious change in movement.
gpaprmh,
I will give the sensor and focal length settings above a go and see what happens. No I have not tried the refine option yet but i’ll give the above options a go then try refining it.
Thanks all for your help.
EDIT:
Ok the new values gave a better average of 1.5 
but the points don’t match up with the background image in the camera.
@Daccy: I believe it does… if you change the sensor width in the camera data tab, just watch what happens to the focal length…
You’re right! I haven’t considered that before. Thanks!
@Daccy: your welcome… but it goes even deeper than that… my camera doesn’t even use the whole sensor width when in movie mode so using my spec sensor width is completely invalid - I have to use the 35mm equivalents. I will be posting soon on a new way to determine a good sw/fl pair to use for any camera… it seems to work well( I still don’t get real good solves due to other probs with the camera) but I need put together a coherent way to present it on BA.
BTW:
No worries, I do that way too often myself… I call them “brain farts” :o … just read my sig
