Composing multiple radiance images

radiance-general
1

Hi All,

I have what I think might be a simple question.

I am trying to recompose a series of images rendered using the -vs and
-vl options of rpict (horizontal and vertical view shift).

I thought that pcompos would have done the job just right but
unfortunately it seems it changes the exposure in each image, having the
final result to look rather patchy.

Has anyone got round this (maybe some pcomb command? but (quote) 'Pcomb
combines equal-sized RADIANCE pictures and sends the result to the
standard output')

thanks in advance for your help,

best regards,

Giovanni

2

Hi Giovanni,

Are the images being filtered in any way before pcompos? If so this is where your variation is coming from....

-Jack

···

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

On 4/17/2012 3:30 PM, Giovanni Betti wrote:

Hi All,

I have what I think might be a simple question.

I am trying to recompose a series of images rendered using the -vs and
-vl options of rpict (horizontal and vertical view shift).

I thought that pcompos would have done the job just right but
unfortunately it seems it changes the exposure in each image, having the
final result to look rather patchy.

Has anyone got round this (maybe some pcomb command? but (quote) 'Pcomb
combines equal-sized RADIANCE pictures and sends the result to the
standard output')

thanks in advance for your help,

best regards,

Giovanni

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

3

Hi Jack, thanks for the quick reply,

I am not performing any filtering but maybe there is something happening
behind the scenes...
I simply set up a job generating 9 images using rad and then a simple
pcompos line to tile them together

  pcompos -a 3 test00*.pic > composite.pic

is this offering more insights?

Thanks,

G

···

----Original Message-----
From: Jack de Valpine [mailto:[email protected]]
Sent: 17 April 2012 20:55
To: Radiance general discussion
Subject: Re: [Radiance-general] Composing multiple radiance images

Hi Giovanni,

Are the images being filtered in any way before pcompos? If so this is
where your variation is coming from....

-Jack

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

On 4/17/2012 3:30 PM, Giovanni Betti wrote:

Hi All,

I have what I think might be a simple question.

I am trying to recompose a series of images rendered using the -vs and
-vl options of rpict (horizontal and vertical view shift).

I thought that pcompos would have done the job just right but
unfortunately it seems it changes the exposure in each image, having

the

final result to look rather patchy.

Has anyone got round this (maybe some pcomb command? but (quote)

'Pcomb

combines equal-sized RADIANCE pictures and sends the result to the
standard output')

thanks in advance for your help,

best regards,

Giovanni

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

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

4

The other thing that happens is that the images lose their original
luminance values and seem to be remapped between 1 and 179cd/m2... this
means that there is definitely some filtering going on. Is this the
correct pcompos behaviour or there is something I am doing wrong?

Best,

Giovanni

···

-----Original Message-----
From: Giovanni Betti [mailto:[email protected]]
Sent: 18 April 2012 10:07
To: Radiance general discussion
Subject: Re: [Radiance-general] Composing multiple radiance images

Hi Jack, thanks for the quick reply,

I am not performing any filtering but maybe there is something happening
behind the scenes...
I simply set up a job generating 9 images using rad and then a simple
pcompos line to tile them together

  pcompos -a 3 test00*.pic > composite.pic

is this offering more insights?

Thanks,

G

----Original Message-----
From: Jack de Valpine [mailto:[email protected]]
Sent: 17 April 2012 20:55
To: Radiance general discussion
Subject: Re: [Radiance-general] Composing multiple radiance images

Hi Giovanni,

Are the images being filtered in any way before pcompos? If so this is
where your variation is coming from....

-Jack

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

On 4/17/2012 3:30 PM, Giovanni Betti wrote:

Hi All,

I have what I think might be a simple question.

I am trying to recompose a series of images rendered using the -vs and
-vl options of rpict (horizontal and vertical view shift).

I thought that pcompos would have done the job just right but
unfortunately it seems it changes the exposure in each image, having

the

final result to look rather patchy.

Has anyone got round this (maybe some pcomb command? but (quote)

'Pcomb

combines equal-sized RADIANCE pictures and sends the result to the
standard output')

thanks in advance for your help,

best regards,

Giovanni

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

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

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

5

Hi Giovanni!

I am not performing any filtering but maybe there is something happening
behind the scenes...
I simply set up a job generating 9 images using rad and then a simple
pcompos line to tile them together

  pcompos -a 3 test00*.pic> composite.pic

is this offering more insights?

rad will typically call pfilt

1) to filter down resolution (at any quality setting not equal to LOW) to achieve antialiasing as well as
2) to apply a useful exposure (at any quality setting).

If you want to assemble the filtered images, you should set a common exposure value to all of them. See the manpage of rad, where the EXPOSURE variable is explained as well as how to find a suitable value.

You can also simply keep the unfiltered images as generated by rpict. You could then assemble the raw images and filter the complete, assembled mosaic. Tell rad to keep the unfiltered images by specifying a name pattern using the RAWFILE variable. Again, the details are explained in the manpage of rad.

If you want to read numerical values from filtered images, usually the -o option are of help. This gives you access to the original pixel values by reversing the exposure mapping performed by e.g. pfilt. See e.g. the manpage of pvalue.

Cheers, Lars.

6

Hi Giovanni,

As Lars has pointed out, rad typically ends an image with a call to pfilt, which WILL adjust the exposure either implicitly per image (likely the source of your variations) or explicitly by
specifying what pfilt should do in your rif file.

-Jack

···

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

On 4/18/2012 6:19 AM, Lars O. Grobe wrote:

Hi Giovanni!

I am not performing any filtering but maybe there is something happening
behind the scenes...
I simply set up a job generating 9 images using rad and then a simple
pcompos line to tile them together

    pcompos -a 3 test00*.pic> composite.pic

is this offering more insights?

rad will typically call pfilt

1) to filter down resolution (at any quality setting not equal to LOW) to achieve antialiasing as well as
2) to apply a useful exposure (at any quality setting).

If you want to assemble the filtered images, you should set a common exposure value to all of them. See the manpage of rad, where the EXPOSURE variable is explained as well as how to find a suitable value.

You can also simply keep the unfiltered images as generated by rpict. You could then assemble the raw images and filter the complete, assembled mosaic. Tell rad to keep the unfiltered images by specifying a name pattern using the RAWFILE variable. Again, the details are explained in the manpage of rad.

If you want to read numerical values from filtered images, usually the -o option are of help. This gives you access to the original pixel values by reversing the exposure mapping performed by e.g. pfilt. See e.g. the manpage of pvalue.

Cheers, Lars.

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

7

Thanks a lot both Lars and Jack,

I thought it was something rather simple I was missing!

Using the unfiltered raw images just did the trick;

Thanks,

G

···

-----Original Message-----
From: Jack de Valpine [mailto:[email protected]]
Sent: 18 April 2012 14:12
To: Radiance general discussion
Subject: Re: [Radiance-general] Composing multiple radiance images

Hi Giovanni,

As Lars has pointed out, rad typically ends an image with a call to
pfilt, which WILL adjust the exposure either implicitly per image
(likely the source of your variations) or explicitly by
specifying what pfilt should do in your rif file.

-Jack

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

On 4/18/2012 6:19 AM, Lars O. Grobe wrote:

Hi Giovanni!

I am not performing any filtering but maybe there is something

happening

behind the scenes...
I simply set up a job generating 9 images using rad and then a simple
pcompos line to tile them together

    pcompos -a 3 test00*.pic> composite.pic

is this offering more insights?

rad will typically call pfilt

1) to filter down resolution (at any quality setting not equal to LOW)

to achieve antialiasing as well as
2) to apply a useful exposure (at any quality setting).

If you want to assemble the filtered images, you should set a common
exposure value to all of them. See the manpage of rad, where the
EXPOSURE variable is explained as well as how to find a suitable

value.

You can also simply keep the unfiltered images as generated by rpict.
You could then assemble the raw images and filter the complete,
assembled mosaic. Tell rad to keep the unfiltered images by specifying

a name pattern using the RAWFILE variable. Again, the details are
explained in the manpage of rad.

If you want to read numerical values from filtered images, usually the

-o option are of help. This gives you access to the original pixel
values by reversing the exposure mapping performed by e.g. pfilt. See
e.g. the manpage of pvalue.

Cheers, Lars.

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

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