exposure value

radiance-general
1

Hi Greg,

thanks for your explanation about the weights for computing luminance. That has helped.

Now, our next problem deals with the exposure value:

If you use rview on the octree, trace returns among other values a luminance value. As far as we understand the octree should be free of any image processing stuff, like exposure values. So, we regard this luminance value as a reflection of the reading of a photometer.

If we then use ximage on the picture, generated from the very same octree, we read the very same luminance value as under rview, even though getinfo returns an exposure value (Note, we have not set any exposure values in rpict and pfilt (apart from the default values).
According to the luminance equation and other notes, one should actually divide each radiance value by the exposure value, or divide the luminance reading by the exposure value.

So, which luminance reading is the 'right' one in ximage? The one returned from the image or the one that one obtains by dividing the returned one by the exposure value?

And how do I avoid any exposure setting or correction when producing a picture (radiance picture that is)?

THanks for all your help in the hectic pre-christmas time.......

alexa and marina
=dumb and dumber :slight_smile:

···

Message: 1
Cc: <radiance-general@radiance-online.org>,
   Alexa Ruppertsberg <A.I.Ruppertsberg@Bradford.ac.uk>
From: Greg Ward <gward@lmi.net>
Date: Tue, 16 Dec 2003 08:35:38 -0800
To: M BLOJ <M.Bloj@Bradford.ac.uk>

The weights correspond to the conversion matrix between Radiance's RGB color space and CIE XYZ (the Y channel, specifically). This is not very well correlated to the V-lambda curve, because the conversion depends so strongly on the exact choice of red, green, and blue primaries. In the case of Radiance and most standard RGB systems, the red primary is much closer to the peak of V-lambda than is the blue primary. Thus, the contribution of the blue channel to luminance is much reduced.

I hope this explains things for you.
-Greg

2

Ximage does the division by the exposure value for you when you use the 'l' command. If you use the 'c' command, it reports the RGB from the file without exposure correction. The "right" value in terms of absolute numbers is the one reported by the 'l' command. The one reported by 'c' is what is actually at that point in the image, which has been premultiplied by the value(s) reported by getinfo.

To save an image without changing it's exposure in rview, set 'e =1' prior to writing it out. Rpict never adjusts the exposure of its output -- you have to run pfilt yourself for that.

Hope this helps!
-Greg

···

From: "Alexa I. Ruppertsberg" <a.i.ruppertsberg@Bradford.ac.uk>
Date: December 17, 2003 8:04:56 AM PST

Hi Greg,

thanks for your explanation about the weights for computing luminance. That has helped.

Now, our next problem deals with the exposure value:

If you use rview on the octree, trace returns among other values a luminance value. As far as we understand the octree should be free of any image processing stuff, like exposure values. So, we regard this luminance value as a reflection of the reading of a photometer.

If we then use ximage on the picture, generated from the very same octree, we read the very same luminance value as under rview, even though getinfo returns an exposure value (Note, we have not set any exposure values in rpict and pfilt (apart from the default values).
According to the luminance equation and other notes, one should actually divide each radiance value by the exposure value, or divide the luminance reading by the exposure value.

So, which luminance reading is the 'right' one in ximage? The one returned from the image or the one that one obtains by dividing the returned one by the exposure value?

And how do I avoid any exposure setting or correction when producing a picture (radiance picture that is)?

THanks for all your help in the hectic pre-christmas time.......

alexa and marina
=dumb and dumber :slight_smile: