Calculating Angle of View from stitched images
Rusty
together. Any suggestions on how to calculate this?
Erik Krause
> I need to determined the angle of view after stitching 2 to 3 images
> together. Any suggestions on how to calculate this?
You don't need to calculate this. Go to Panorama Settings tab
(eventually switch to advanced interface on Project Assistant tab) and
read "Field of view". These are the values for the entire view visible
in panorama editor.
best regards
--
Erik Krause
http://www.erik-krause.de
Rusty
PTGui Support
the field of view.
Joost
Rusty
Andre Regini
> (eventually switch to advanced interface on Project Assistant tab) and
> read "Field of view". These are the values for the entire view visible
> in panorama editor.
doing a QTVR, how nice it would be to have a window that you size and
drag to set the field of view and initial pan and tilt positions, and
then have the min and max pan/tilt values taken from any crop lines.
(Most of my clients do not want full pan tilt 360 QTVRs)
Andre Regini
www.kingbridge.co.uk
www.regini.com
Erik Krause
> Great until you use the crop lines. Or have I missed something? When
> doing a QTVR, how nice it would be to have a window that you size and
> drag to set the field of view and initial pan and tilt positions, and
> then have the min and max pan/tilt values taken from any crop lines.
> (Most of my clients do not want full pan tilt 360 QTVRs)
I don't understand you completely. You have two possibilities to make a
QTVR of less than 180° vFoV. You can either create a cylindrical QTVR
(currently not supported in PTGui: use Pano2VR or similar) or create a
cubic QTVR with limited vFoV in PTGui. In both cases you don't need to
use the crop lines (and in fact you can get a bent panorama if you don't
use them exactly symmetrical). Simply use hFoV slider (and eventually
vFoV slider) to crop the image in panorama editor and read the values in
pano editor lower left corner (or Panorama Settings tab).
Andre Regini
you are sugegsting and what I want cannot be achieved that way. But I
fail to understand why you cannot comprehend that. Let me explain in
(may be) simpler language. Many of my clients want to show an area of
a location, not a full 360 and not a full pan tilt
(www.kingbridge.co.uk/pano/pano1.mov and ..pano2.mov). With me so far?
I have been successfully producing these by doing a partial image
capture of the location (note partial), producing an equirectangular
projection and then producing a QTVR output from PTGui. Because the
capture is not a full panoramic you cannot always successfully use the
horizontal and vertical fov sliders to crop the image. The issue then
is where to set your initial, min and max tilts and pans and FOV
values to limit the QT viewer and not show blank areas. At the moment
this achieved by trial and error. My suggestion therefore is to have a
scalable and moveable box (aspect ratio defined by "window size") to
define the "Initial" values of pan tilt and FOV and then have limit
lines to define "Min" "Max" values for pan and tilt. Two other boxes
can then set the "Min" "Max" FOV (if need be). (If you say the limit
lines have to be symmetrical (eh?), so be it. Moving the one crop line
also moves the oppossing one an equal amount.)
Andre Regini
Erik Krause
> (www.kingbridge.co.uk/pano/pano1.mov and ..pano2.mov). With me so far?
Yes. Your example could be displayed as a cylindrical QTVR but it works
as an equirectangular as well. However, as an equirectangular you have
to put black space around the image to get it converted correctly. A
cylindrical could simply be cropped to the content area.
> I have been successfully producing these by doing a partial image
> capture of the location (note partial), producing an equirectangular
> projection and then producing a QTVR output from PTGui. Because the
> capture is not a full panoramic you cannot always successfully use the
> horizontal and vertical fov sliders to crop the image. The issue then
> is where to set your initial, min and max tilts and pans and FOV
> values to limit the QT viewer and not show blank areas. At the moment
> this achieved by trial and error.
This is why programs like Pano2VR exist. You have a WYSIWYG FoV editor
there. You simply pan as far as you want and store the respective value
as min or max pan. Same for tilt.
> My suggestion therefore is to have a
> scalable and moveable box (aspect ratio defined by "window size") to
> define the "Initial" values of pan tilt and FOV and then have limit
> lines to define "Min" "Max" values for pan and tilt. Two other boxes
> can then set the "Min" "Max" FOV (if need be). (If you say the limit
> lines have to be symmetrical (eh?), so be it. Moving the one crop line
> also moves the oppossing one an equal amount.)
No, the crop lines have not to be symmetrical (you automatically crop
symmetrical if you use the FoV sliders).
However, it would be easy indeed to use the yellow crop lines as min and
max values for the QTVR output. Perhaps Joost can take this as a feature
request.
For the time being you can use a workaround instead of try and error:
Save the project file and open it with a text editor. Search for the
line starting with
#-outputcrop
The values behind are the relative positions of the crop lines in the
order left right top bottom (no crop is 0 1 0 1). Use these numbers to
calculate the angles (works only for equirectangular): Multiply the left
and right value with 360 and the top and bottom value with 180. With me
so far ;-)
The general problem using the pano editor to get these values is that in
the equirectangular image the view you later see in QT is by no means
rectangular - which makes it somehow unintuitive to set the initial view...
Bjørn K Nilssen
>
> Now I don't understand what you don't understand. I understand what
> you are sugegsting and what I want cannot be achieved that way. But I
> fail to understand why you cannot comprehend that. Let me explain in
> (may be) simpler language. Many of my clients want to show an area of
> a location, not a full 360 and not a full pan tilt
> (www.kingbridge.co.uk/pano/pano1.mov and ..pano2.mov). With me so far?
What I don't understand here is why you feel the need to be patronizing because you
didn't understand Eriks reply, which BTW was much easier to understand than the original
questions, and which gave a good and fully adequate solution to your problem.
If you need further, more detailed explanations on how to do it, or on how
spherical/cubic panoramas work, I'm sure you can get that too, if you ask.
--
Bjørn K Nilssen - http://bknilssen.no - panoramas and 3D
Andre Regini
langauge issue here.
The phrase "I don't understand you completely." which preceeded Erik's
post, is not very good grammer and is ambiguous. But in the context of
the rest of the post, I interprteded it as arrogant and patronising.
Hence the tone of my reply. If it wasn't intended in that manner, I
apologise. The phrase "I don't completely understand you." which is
what may have been intended has a very different tone and meaning in
English. (Only the English language could get be that twisted)
However contrary to what you say, "..you didn't understand Eriks
reply.", although I may have read the tone at face value, I did
understand what had been proposed and I also understood why it wasn't
a solution.
Thank you Erik for the suggestion, just a bit long winded when trying
to process as fast as possible. I've now found a method where I set a
grid with say 18 divisions in the verical axis (equal to 10 deg/div)
and then count the squares. So... as a halfway house, it should be
possible to have set grids (5/10/15 degrees?) annotated in degrees in
the editor window? Or maybe a cursor read out?
Andre Regini
www.kingbridge.co.uk
www.regini.com
On Jul 21, 9:42 pm, Bjørn K Nilssen <b...@bknilssen.no> wrote:
> On 21 Jul 2009 at 11:48, Andre Regini wrote:
>
>
>
> > Now I don't understand what you don't understand. I understand what
> > you are sugegsting and what I want cannot be achieved that way. But I
> > fail to understand why you cannot comprehend that. Let me explain in
> > (may be) simpler language. Many of my clients want to show an area of
> > a location, not a full 360 and not a full pan tilt
PTGui Support
> The general problem using the pano editor to get these values is that in
> the equirectangular image the view you later see in QT is by no means
> rectangular - which makes it somehow unintuitive to set the initial view...
Indeed, that's exactly the problem. The yellow crop lines are straight
in the equirectangular image, but when projected into the QTVR they
would be curved line.
The QTVR export in PTGui is deliberately kept simple; if you need more
advanced settings, hotspots or a better GUI interface I would recommend
dedicated software such as Pano2VR.
Joost