At http://depositfiles.com/files/qy1a48oal you can download a series of
tests with vuescan (same files as mentioned in "Cannot laod DNG-file
from Vuescan"). And now I try to find out with these files, what is the
best format for darktable?

If someone has problems with the download, let me know, and I will give
you an url to my homepage, with limited bandwidth.

The goal is to use a file with the best quality (lossless) and the best
compression, so the smallest filesize is used.

With my sample scans I got:

series2:
color:
uncompressed tiff:
36M series2_scan-130414-0001.tif
36M series2_scanraw0001.tif

exiftool -Compression series2_scan-130414-0001.tif
Compression                     : Uncompressed

exiftool -Compression series2_scanraw0001.tif 
Compression                     : Uncompressed


identify -verbose series2_scan-130414-0001.tif
Image: series2_scan-130414-0001.tif
  Format: TIFF (Tagged Image File Format)
  Class: DirectClass
  Geometry: 3037x2066+0+0
  Resolution: 600x600
  Print size: 5.06167x3.44333
  Units: PixelsPerInch
  Type: TrueColor
  Base type: TrueColor
  Endianess: MSB
  Colorspace: RGB
  Depth: 16-bit
  Channel depth:
    red: 16-bit
    green: 16-bit
    blue: 16-bit
...

So we have a colorscan with 48bit.

If I convert these files to png (lossless, compression options -quality
90) I get

30M series2_scan-130414-0001.tif.png
18M series2_scanraw0001.tif.png

So you see the raw-file of vuescan was compressed a lot more than the
normal file.

I am not sure, but as I understand vuescan, you can use a
"vuescan-rawfile" to modify the image without scanning again. So it
should be possible to create series2_scan-130414-0001.tif from
series2_scanraw0001.tif and series2_scanraw0001.tif should be the file
which should be used with darktable, since it is a lot smaller. But I
have to say, I couldn't produce a similar image to
series2_scan-130414-0001.tif created from series2_scanraw0001.tif as
source with darktable (but I am a dt newbie)

The 2nd best way, would be to use a tiff-file, which was compressed by
vuescan. (series3)

41M series3_scan-130414-0001.tif
36M series3_scanraw0001.tif

exiftool -Compression series3_scan-130414-0001.tif
Compression                     : LZW

exiftool -Compression series3_scanraw0001.tif
Compression                     : Uncompressed

Hey that's really strange, the LZW compressed file is bigger than the
uncompressed file and bigger as the raw file. Ok that's a different
scan, but with the same settings as before. Only the compression was
changed. Please see screenshots of the settings.

identify -verbose series3_scan-130414-0001.tif
Image: series3_scan-130414-0001.tif
  Format: TIFF (Tagged Image File Format)
  Class: DirectClass
  Geometry: 3037x2066+0+0
  Resolution: 600x600
  Print size: 5.06167x3.44333
  Units: PixelsPerInch
  Type: TrueColor
  Base type: TrueColor
  Endianess: MSB
  Colorspace: RGB
  Depth: 16-bit
  Channel depth:
    red: 16-bit
    green: 16-bit
    blue: 16-bit





It looks different with b&w scans.

series5
black and white, uncompressed

12M series5_scan-130414-0001.tif
12M series5_scanraw0001.tif

exiftool -Compression series5_scan-130414-0001.tif
Compression                     : Uncompressed

exiftool -Compression series5_scanraw0001.tif
Compression                     : Uncompressed

identify -verbose series5_scan-130414-0001.tif
Image: series5_scan-130414-0001.tif
  Format: TIFF (Tagged Image File Format)
  Class: DirectClass
  Geometry: 3037x2066+0+0
  Resolution: 600x600
  Print size: 5.06167x3.44333
  Units: PixelsPerInch
  Type: Grayscale
  Base type: Grayscale
  Endianess: MSB
  Colorspace: RGB
  Depth: 16-bit
  Channel depth:
    gray: 16-bit
  Channel statistics:
    Gray:
      min: 8027 (0.122484)
      max: 65535 (1)
      mean: 36400.8 (0.555441)
      standard deviation: 15888.2 (0.242439)
      kurtosis: -1.19674
      skewness: -0.0278737


So we have a scan with 16bit gray.

If I convert these b&w files to png (lossless, compression options
-quality 90) I get

4,3M Apr series5_scan-130414-0001.tif.png
4,5M Apr series5_scanraw0001.tif.png

So there is a lot more harddrive space saved than with a color scan,
but while the tif-files are nearly identical, the "raw-png-file" is
less compressed than the "normal" one.

Series 6 shows the compressed tif-version:
6,2M series6_scan-130414-0001.tif
12M series6_scanraw0001.tif

exiftool -Compression series6_scan-130414-0001.tif
Compression                     : LZW

exiftool -Compression series6_scanraw0001.tif
Compression                     : Uncompressed

Obviously all raw-vuescan-scans are uncompressed and the
output-compression-settings are not used for "raw".


So IMHO it would be best to use png-files for darktable to improve the
scan, but what is not clear, if the color-png-file should be used, that
was created by "raw-vuescan-scan", while with b&w it is better not to
use the "raw-vuescan-scan", but the "normal" scan.

I would be happy the hear any suggestions from you. I would like to
scan a lot of paper photos, not slides or negatives at the moment. And
there is a huge difference in filesize, depending which image-format is
used.

-- 
Αl


------------------------------------------------------------------------------
Precog is a next-generation analytics platform capable of advanced
analytics on semi-structured data. The platform includes APIs for building
apps and a phenomenal toolset for data science. Developers can use
our toolset for easy data analysis & visualization. Get a free account!
http://www2.precog.com/precogplatform/slashdotnewsletter
_______________________________________________
Darktable-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/darktable-users

Reply via email to