I think you are proceeding under a false assumption.
The "front" cube face and the center (left-right) portion of the
equirectangular image MIGHT correspond to the Initial View values of
a QTVR movie, but there is nothing that says they MUST.
Further, neither the exacted cube faces, nor an extracted
equirectangular image contain information that corresponds to the
Initial View values for a QTVR movie. [It is true that the data
inside a QTVR movie, probably tiles, can be reordered so that tiles
"in front of" the viewer occur earlier in the file, but trying to
unravel THAT to figure out the initial view seems like far more work
than is necessary, especially since the initial view info is ALREADY
IN a QTVR movie.]
That is, you CANNOT reliably derive the initial angle of view (and
zoom, or window size) from the extracted cube faces. Assuming that's
what you're hoping to do. That information is not in there!
On Jun 3, 2007, at 5:24 PM, Pat Swovelin wrote:
> Hans Nyberg wrote:
>> --- In PanoToolsNG@yahoogroups.com, Pat Swovelin <Panoramas@...>
>>> Hans Nyberg wrote:
>>>> You can only do this on Mac with Cubic Converter.
>>> *Aarrrgh!* There's got to be a way to be able to do it on a PC.
>>>> But it is easy to just open it in photoshop and change the view
>>>> the offset filter.
>>> Not if the initial view is tilted up/down because that will wrap the
>>> distorted top and bottom edges to someplace in the middle of the
>> I can not see the problem why you can not use your original
>> equirectangular image and
>> change the view with the offset filter. I do that all the time with
>> extracted images from the
>> movies I get.
>> If I understand you right you want the center of the equirectangular
>> image to be at the
>> same initial center as the movie. I am not sure how Pano2QTVR does it
>> but I assume you
>> have done some settings in the conversion which have changed the
>> initial view.
>> That is just a setting in the movie and has nothing to do with the
>> Your extracted cubefaces will always be at zero tilt so you do not
>> any changes like you
>> assume. Your reassembled equirectangular will be exactly as your
> Exactly. The problem, as it turns out is that when you have a
> equirectangular image (based on some pan and tilt numbers) the
> bottom of
> the pano is now off to the side and the thing looks like it was shot
> with the new NauseaCam (it made me ill just to look at it). So the
> concept of having a new equirectangular image that's pre-tilted and
> panned won't work (darn it).