On Wed, 24 Mar 1999 09:47:38 +1000 [EMAIL PROTECTED] wrote:
>Memory usage.
>
>As for the memory requirements, 256x256x256 is actually 24 bit colour not 16
>bit. However, the camera is not trying to store 16+ million bits of
>information. Each sensor in the meter, 1005 elements, reads the light falling
>on it and presents, at 256 levels, 8 bits of information (256 in binary is 1111
>1111 or 8 bits) so this is close to 8 kbytes for the colour information.
No, you have 8 kbits, or 1 kbyte, as I said.
>I don't specifically know that the meter uses 256 levels for each sensor. I
>chose that number as it is used in most PC paint programs to represent colour.
>I was attempting to describe the principle of operation not the technical
>specifications. Should it use only 64 levels of colour on each sensor, the
>memory requirement drops to only 2k. Even 64 levels per sensor would be
>adequate to determine the PREDOMINANT colour of the scene (perhaps even 32
>levels - 1k memory).
>
>Now for the 30,000 pictures. The way I understand the literature that Nikon
>have distributed about their matrix meters is that 30,000 actual pictures were
>used to develop the database in the camera, however, each of those 30,000
>pictures had a single exposure so we are only storing 30,000 exposure indexes.
>This would not be a great memory burden.
Well it makes no sense to have a database of 30,000 exposure indexes since there
are much fewer possible ahutter/aperture combinations than this possible between
EV -1 and 21. There is also little point having a colour meter if the color
data from the meter is not compared to other colour data, so I think that the
'database' must include color information
>But there are more pieces of information gathered by the meter
>than just the colour values. The brightness and contrast of the scene are also
>obtained.
Under the RGB scheme you describe, the contrast and brightness information
information comes from the RGB sensors. The only way it knows this is because
0, 0, 0 is less bright than 255, 255, 255, and the two of them give a lot of
contrast. So contrast and brightness are based on the color values.
>Your grey card in the sun is going to be less bright than a white
>card in the sun. In the shade the white card will still be brighter.
If, according to a normal meter, a grey card in the sun gives EV 16 and a white
card in the shade gives EV 16, how is the white card in the shade brighter??
When this
>is combined with the colour information the program in the camera comes up with
>the correct exposure.
>
>Take an example of data from the meter 10,25,50,14,22 being
>red,green,blue,brightness,contrast, although distance is also used I believe
>that this is primarily for flash work. Now imagine you have a huge box with
>other boxes inside, take out the #10 red box. Open this and take out the #25
>green box. Open this for the #50 blue box, open this for the #14 brightness
>box
>and then the #22 contrast box. Inside this is a piece of paper with the
>exposure index on it. Convert this to speed and aperture and away you go.
Yes, but the exposure isn't just one box, it's a matrix of many hundreds of
boxes. Unless, of course, you simply wan't to average all the color
information, which doesn't seem to be making the most of the situation. For
example, if you average all the information the camera will think a picture with
bands of red (200, 0, 0), blue (0, 200, 0) and green (0, 0, 200) is the same as
a picture of solid grey (200, 200, 200).
