aspect ratio problem

Hi ,

I render for PAL anamorphic ( 16:9 - 720x576 ) video.
I use option -pa 1.42 in both rpict and pfilt for setting the proper aspect ratio. The output resolution is 720x576.
The resulting image indicates (using getinfo) an aspect ratio 1.42 as expected.
But the result of a radiance image converted using ra_tiff seems to be a tiff file with square pixels (aspect ration 1:1).

When I import such a tiff image into a video editing program (final cut pro) I get a black borders left and right of the image (the image does not fill the entire video screen).
What am I doing wrong here ??

-Iebele

I render for PAL anamorphic ( 16:9 - 720x576 ) video.
I use option -pa 1.42 in both rpict and pfilt for setting the proper aspect ratio. The output resolution is 720x576.
The resulting image indicates (using getinfo) an aspect ratio 1.42 as expected.
But the result of a radiance image converted using ra_tiff seems to be a tiff file with square pixels (aspect ration 1:1).

I think you should not use -pa at all (or at zero). You can use it if one pixel is not like a "square", to correct. But in your case, the pixels have an aspect ratio of about 1/1, so the -pa setting results in a picture stretched just as far as to get a square picture. In other words: the aspect ratio of the image is determined by vv and vh (or the resolution), the pa setting is for the pixel aspect ratio. That is at least how I understood -pa....

CU Lars.

Hi Lars,

Thanks for your suggestion using image aspect ratios instead of pixel aspect ratios.

Anamorpic Pal is (as I understood) a 720x576 image stretched to 1024x576 by the displaying device.
This suggests a pixel of which width is 1.42 taller as its height.

I tried your suggestion and rendered my image at 1024x576 square pixel with a the vv value calcualted like this:
vv = vh / 1.42 (where 1.42 is 1024/720 )

Rendering square pixels at 1024x576 works fine, at least the image fits the videoframe.
But I really wonder if the way I calculated the vv value is correct.

-Iebele

Lars O. Grobe wrote:

···

I render for PAL anamorphic ( 16:9 - 720x576 ) video.
I use option -pa 1.42 in both rpict and pfilt for setting the proper aspect ratio. The output resolution is 720x576.
The resulting image indicates (using getinfo) an aspect ratio 1.42 as expected.
But the result of a radiance image converted using ra_tiff seems to be a tiff file with square pixels (aspect ration 1:1).

I think you should not use -pa at all (or at zero). You can use it if one pixel is not like a "square", to correct. But in your case, the pixels have an aspect ratio of about 1/1, so the -pa setting results in a picture stretched just as far as to get a square picture. In other words: the aspect ratio of the image is determined by vv and vh (or the resolution), the pa setting is for the pixel aspect ratio. That is at least how I understood -pa....

CU Lars.

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

iebele wrote:

Hi Lars,

Thanks for your suggestion using image aspect ratios instead of pixel
aspect ratios.

Anamorpic Pal is (as I understood) a 720x576 image stretched to 1024x576
by the displaying device.
This suggests a pixel of which width is 1.42 taller as its height.

I tried your suggestion and rendered my image at 1024x576 square pixel
with a the vv value calcualted like this:
vv = vh / 1.42 (where 1.42 is 1024/720 )

Rendering square pixels at 1024x576 works fine, at least the image fits
the videoframe.
But I really wonder if the way I calculated the vv value is correct.

-Iebele

vv and vh are view angles. To fit them to a given image aspect ratio
(height over width) make sure that the quotient of the tangents of the
angles fit the image aspect, i.e. tan(vv)/tan(vh) = 576/1024 in your
example.
This gives you undistorted square pixels at the given resolution, I
don't know however if this is what you need, 'cause I don't know how
that stretching in anamorphic pal is done..

-cb

Hi Carsten,

I have a real problem in understanding goniometry :frowning:

Do you know how to solve your equotation when the horizontal view angle is known?
In other words, how do I get the value of vv when:

tan(vv) / tan ( 84 degrees ) = 576/1024

Thanks a lot!

···

iebele wrote:

Hi Lars,

Thanks for your suggestion using image aspect ratios instead of pixel
aspect ratios.

Anamorpic Pal is (as I understood) a 720x576 image stretched to 1024x576
by the displaying device.
This suggests a pixel of which width is 1.42 taller as its height.

I tried your suggestion and rendered my image at 1024x576 square pixel
with a the vv value calcualted like this:
vv = vh / 1.42 (where 1.42 is 1024/720 )

Rendering square pixels at 1024x576 works fine, at least the image fits
the videoframe.
But I really wonder if the way I calculated the vv value is correct.

-Iebele
   
vv and vh are view angles. To fit them to a given image aspect ratio
(height over width) make sure that the quotient of the tangents of the
angles fit the image aspect, i.e. tan(vv)/tan(vh) = 576/1024 in your
example.
This gives you undistorted square pixels at the given resolution, I
don't know however if this is what you need, 'cause I don't know how
that stretching in anamorphic pal is done..

-cb

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

Hi Iebele,

If you have a specific pixel aspect ratio, ra_tiff preserves this information in it's horizontal/vertical density tags. If your application doesn't understand or interpret these tags, your image won't be displayed correctly. Oddly enough, Photoshop doesn't correct the aspect ratio on display. Photosphere does, as does Preview. I don't know what tool you are using when you say that the aspect ratio is lost in the TIFF.

-Greg

···

From: iebele <[email protected]>
Date: September 7, 2006 5:45:54 AM PDT

Hi Lars,

Thanks for your suggestion using image aspect ratios instead of pixel aspect ratios.

Anamorpic Pal is (as I understood) a 720x576 image stretched to 1024x576 by the displaying device.
This suggests a pixel of which width is 1.42 taller as its height.

I tried your suggestion and rendered my image at 1024x576 square pixel with a the vv value calcualted like this:
vv = vh / 1.42 (where 1.42 is 1024/720 )

Rendering square pixels at 1024x576 works fine, at least the image fits the videoframe.
But I really wonder if the way I calculated the vv value is correct.

-Iebele

Hi iebele.

i think i may for the first time be able to offer assitance on a radiance post!

from

tan(vv) / tan ( 84 degrees ) = 576/1024

rearrange to

tan(vv) = (576/1024)*(tan(84))

then too

vv = tan-1 [(576/1024)*(tan(84))]

stick that in a calculator and u should get an answer.

chris

p.s. the -1 is meant to be superscript as it represents tan to the power of -1 (inverse tan).

···

-------------------------------------------------------------------
This e-mail and any attachments may contain confidential and/or
privileged material; it is for the intended addressee(s) only.
If you are not a named addressee, you must not use, retain or
disclose such information.

NPL Management Ltd cannot guarantee that the e-mail or any
attachments are free from viruses.

NPL Management Ltd. Registered in England and Wales. No: 2937881
Registered Office: Serco House, 16 Bartley Wood Business Park,
                   Hook, Hampshire, United Kingdom RG27 9UY
-------------------------------------------------------------------

If you have "Rendering with Radiance," look up "aspect ratio" in the index.

-G

···

From: iebele <[email protected]>
Date: September 7, 2006 8:09:56 AM PDT

Hi Carsten,

I have a real problem in understanding goniometry :frowning:

Do you know how to solve your equotation when the horizontal view angle is known?
In other words, how do I get the value of vv when:

tan(vv) / tan ( 84 degrees ) = 576/1024

Thanks a lot!

I have a real problem in understanding goniometry :frowning:

Do you know how to solve your equotation when the horizontal view angle is known?
In other words, how do I get the value of vv when:

tan(vv) / tan ( 84 degrees ) = 576/1024

In 2 steps:
tan(vv) = tan( 84 degrees ) * 576 / 1024
vv = atan( tan( 84 degrees ) * 576 / 1024 )
which gives vv = 79.4 degrees.

Siegbert

···

--
Dr.-Ing. Siegbert Debatin
Development
Relux Informatik AG Dornacherstr. 377 CH-4053 Basel
Tel. ++41 61 333 07 70 Fax: ++41 61 333 07 72

Hi Iebele and others:

Subject: Re: [Radiance-general] aspect ratio problem

Hi Carsten,

I have a real problem in understanding goniometry :frowning:

Do you know how to solve your equotation when the horizontal
view angle is known?
In other words, how do I get the value of vv when:

tan(vv) / tan ( 84 degrees ) = 576/1024

Thanks a lot!

I'm not repeating the other add ons to this equation
just one thing everyone seems to miss (and which is
a bit of nitpicking):

tan(2*a) != 2*tan(a)

therefore the correct equation is

tan(vv/2) / tan(hv/2) = 576 / 1024

Might make the amazing difference of 2 pixels for
an image of that size!

Thomas

···

-----Original Message-----

***********************************************************************************
This e-mail, (and any attachments) is confidential and may be privileged. It may be read, copied and used by the intended addressee only. If you have received this in error please contact BDP immediately.

If you have any queries, please contact the sender.
***********************************************************************************
Building Design Partnership
Registered in England No 2207415:
Registered Office: Building Design Partnership Ltd, Sunlight House, PO Box 85, Quay Street, Manchester, M60 3JA, http://www.bdp.co.uk
***********************************************************************************

Hi,

Thinking about the geometry, I believe that you actually need the following:

    tan(vv/2) / tan(hh/2) = (height/2) / (width /2)

···

==

    tan(vv/2) / tan(hh/2) = height / width

solving for vv:

    vv = 2* atan( height / width * tan(hh/2))

given:

    hh = 84
    height = 576
    width = 1024

    vv = 53.7224 degrees

-Jack

Siegbert Debatin wrote:

I have a real problem in understanding goniometry :frowning:

Do you know how to solve your equotation when the horizontal view angle is known?
In other words, how do I get the value of vv when:

tan(vv) / tan ( 84 degrees ) = 576/1024

In 2 steps:
tan(vv) = tan( 84 degrees ) * 576 / 1024
vv = atan( tan( 84 degrees ) * 576 / 1024 )
which gives vv = 79.4 degrees.

Siegbert
--Dr.-Ing. Siegbert Debatin
Development
Relux Informatik AG Dornacherstr. 377 CH-4053 Basel
Tel. ++41 61 333 07 70 Fax: ++41 61 333 07 72

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

--
# Jack de Valpine
# president
#
# visarc incorporated
# http://www.visarc.com
#
# channeling technology for superior design and construction

Hi Folks,

I love you all.

Jack has given the right answer as far I can judge now:
vh 84 vv 53.7224 renders an image of exactly 1024x576 pixels.

Thanks a LOT.

And Greg, the application that does not interpret the horizontal/vertical density tags of the tif file is Final Cut Pro (Apple).
I tried ra_t16 also, same thing.

-Iebele

···

Hi,

Thinking about the geometry, I believe that you actually need the following:

    tan(vv/2) / tan(hh/2) = (height/2) / (width /2)

    ==

    tan(vv/2) / tan(hh/2) = height / width

solving for vv:

    vv = 2* atan( height / width * tan(hh/2))

given:

    hh = 84
    height = 576
    width = 1024

    vv = 53.7224 degrees

-Jack

Siegbert Debatin wrote:

I have a real problem in understanding goniometry :frowning:

Do you know how to solve your equotation when the horizontal view angle is known?
In other words, how do I get the value of vv when:

tan(vv) / tan ( 84 degrees ) = 576/1024

In 2 steps:
tan(vv) = tan( 84 degrees ) * 576 / 1024
vv = atan( tan( 84 degrees ) * 576 / 1024 )
which gives vv = 79.4 degrees.

Siegbert
--Dr.-Ing. Siegbert Debatin
Development
Relux Informatik AG Dornacherstr. 377 CH-4053 Basel
Tel. ++41 61 333 07 70 Fax: ++41 61 333 07 72

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

--
# Jack de Valpine
# president
#
# visarc incorporated
# http://www.visarc.com
#
# channeling technology for superior design and construction

------------------------------------------------------------------------

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

Bleicher, Thomas wrote:

I'm not repeating the other add ons to this equation
just one thing everyone seems to miss (and which is
a bit of nitpicking):

tan(2*a) != 2*tan(a)

therefore the correct equation is

tan(vv/2) / tan(hv/2) = 576 / 1024

sorry, yepp, I've forgotten about that (funny enough, as I've
implemented exactly the above equation long ago..)
Its not nitpicking, it's simply correct..
-cb

···

***********************************************************************************
Building Design Partnership
Registered in England No 2207415:
Registered Office: Building Design Partnership Ltd, Sunlight House, PO Box 85, Quay Street, Manchester, M60 3JA, http://www.bdp.co.uk
***********************************************************************************

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

--
---------------------------------------------------------

"Der Deckel muß zugehen." (P.Kölsch)

---------------------------------------------------------