- VS & Multi Pro: Huge RAW files and exposure mystery
- Posted by Jouko Vierumäki on July 11th, 2003
I noticed "Increased bits of precision on Scan Multi Pro" in the release
notes of VS 7.6.51... Now, the RAW file size has increased by 50%. Does this
mean there are 24 bits of data per channel (instead of 16)? If this is the
case, where do the new bits come from?
Another thing, now "Input|Exposure clipping (%)" can be adjusted even if
exposure is manually set. Which is the one overriding the other in this
case? With the same exposure clipping value I got the same exposures values
as before - but the film base color values are ~8% higher?
Jouko Vierumäki
http://galleria.vierumaki.com/
- Posted by Ed Hamrick on July 11th, 2003
"Jouko Vierumäki" <jouko@vierumaki.com> wrote:
In previous versions of VueScan, only 12 bit data was being written
to the raw scan file for the Scan Multi Pro.
In previous versions of VueScan, the Exposure clipping (%) option was
being ignored and zero was being used.
The reason it's visible even when exposure is manually set is that
you can change "Input|Exposure clipping (%)" and then do the
"Scanner|Exposure" menu command to recompute "Input|RGB exposure".
Regards,
Ed Hamrick
- Posted by Jouko Vierumäki on July 11th, 2003
"Ed Hamrick" <usenet@hamrick.com> wrote in message
news:belvpb$d7$1@ngspool-d02.news.aol.com...
Oh... 
... But, my images are 7008x4672 pixels in size. As data, they
should be
7008 * 4672 * 4 color channels (RGBI) * 16 bits/channel / 8
bits/byte
which figures about 255MB. And this is what the RAW files have been - now
the RAW files are 50% larger?
Jouko
- Posted by Ed Hamrick on July 11th, 2003
"Jouko Vierumäki" <jouko@vierumaki.com> wrote:
Try turning off "Output|TIFF compression".
Regards,
Ed Hamrick
- Posted by Jouko Vierumäki on July 12th, 2003
"Ed Hamrick" <usenet@hamrick.com> wrote in message
news:bemqhj$f5r$1@ngspool-d02.news.aol.com...
OK. Are you hinting I shouldn't even try to understand what's going on? I
mean, normally compression works the other way around ;-)
Jouko
- Posted by Bart van der Wolf on July 12th, 2003
"Jouko Vierumäki" <jouko@vierumaki.com> wrote in message
news:Q4RPa.5556$sj6.97678@tornado.fastwebnet.it...
I didn't miss your smiley, but to make sure you do understand ;-)
Lossless compression usually depends on replacing frequently occurring bit
patterns, by shorter patterns. For reconstruction, a kind of dictionary is
needed to look up the full meaning of these abbreviations.
If the 16-bit patterns no longer have the least significant bits always set
to zero because there are more actual data bits, there will probably be less
frequent repetitions of patterns. So the potential gain for replacing those
bit patterns is reduced, but the dictionary still has to be built, and there
are more entries in the dictionary. Thus less compression and larger
overhead may actually grow the total number of bytes you tried to compress,
resulting in a larger file.
Bart
- Posted by Bob Shomler on July 12th, 2003
Yes, but 50 percent larger?
Bob Shomler
Bart van der Wolf wrote:
