Hello,
I'm new to digikam and RAW processing. I converted a RAW (CR2) file to DNG and it went from about 8MB to 17MB. I edited and then saved it as PNG. The PNG size is 40MB. What am I doing wrong? Thanks, Conway _______________________________________________ Digikam-users mailing list [hidden email] https://mail.kde.org/mailman/listinfo/digikam-users |
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1 Hi, Am 22.01.2012 21:50, schrieb Conway Upshur: > Hello, I'm new to digikam and RAW processing. I converted a RAW > (CR2) file to DNG and it went from about 8MB to 17MB. I edited and > then saved it as PNG. The PNG size is 40MB. What am I doing wrong? > Thanks, Conway we just had this discussion (for me the second time, and I still don't understand it fully, yet I try): it has to do with colour depth and other things, which are added to the files to make sure all data is kept intact. Basically, there is more info about the image or rather about each pixel stored in the files. I'm sure someone else can explain it more detailed. Cheers Martin -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) iEYEARECAAYFAk8cgTgACgkQUmmuY48ByEhDWwCeMydZAUokXpme+Xwbj/4YMzBt glcAoLVYoy8DrMuoA90YDm7jQ9iERpmX =tOzA -----END PGP SIGNATURE----- _______________________________________________ Digikam-users mailing list [hidden email] https://mail.kde.org/mailman/listinfo/digikam-users drmartinus.vcf (383 bytes) Download Attachment |
Thanks, I'll keep studying.
2012/1/22 Dr. Martin Senftleben <[hidden email]> -----BEGIN PGP SIGNED MESSAGE----- _______________________________________________ Digikam-users mailing list [hidden email] https://mail.kde.org/mailman/listinfo/digikam-users |
In reply to this post by Conway Upshur
On Sunday 22 January 2012 12:50:40 Conway Upshur wrote:
> Hello, > I'm new to digikam and RAW processing. > I converted a RAW (CR2) file to DNG and it went from about 8MB to 17MB. > I edited and then saved it as PNG. The PNG size is 40MB. > What am I doing wrong? Thanks, Conway I can't give the exact details, but here are some indications: - CR2 -> DNG: DNG has the possibility to embed the original file (thus adding a few MB), further, teh compression of the RAW data might be different, and certain metadat could be stored in a different format (text, instead of binary codes, esp. for makerdata fields). - PNG: forgetting compression for now, RAW (CR2) files store 12 or 14 bits per pixel. But those pixels have no inherent colour information (look up 'Bayer matrix'). The PNG stores an image that has as many pixels as the RAW file, but with colour information, at either 8 or 16 bits/colour, and possibly transparency (also 8 or 16 bits). That gives you at least 24 bits/pixel (8 bits/colour, no alpha channel) and at most 64 bits/pixel. And you might have added tags, caption, title, etc. Given that you go from 8 MB for the raw to 40 MB for the png, I suspect you use 16-bit colour depth Remco _______________________________________________ Digikam-users mailing list [hidden email] https://mail.kde.org/mailman/listinfo/digikam-users |
In reply to this post by Conway Upshur
Am 22.01.2012 21:50, schrieb Conway Upshur:
> Hello, > I'm new to digikam and RAW processing. > I converted a RAW (CR2) file to DNG and it went from about 8MB to 17MB. > I edited and then saved it as PNG. The PNG size is 40MB. > What am I doing wrong? Thanks, Conway Raw data does not contain any colour information. it is a black/white/grey picture where the so called bayer-pattern is layed over. The bayer-pattern takes four points from the raw greyscaled photo and sets one to red, one to blue and two to green and then you have coloured photos (this process is called demosaicing). With raw data you have to compress the grey values only (but with 12 up to 16 bit). See http://en.wikipedia.org/wiki/Bayer_pattern and http://en.wikipedia.org/wiki/Demosaicing for more information. For DNG conversion: This depends. You can embed the original raw data together with the converted DNG data, so the result will have at least double size. PNG on the other had is a RGB colour format (with alpha channel). If you encode it with 16bit per colour you increase the data again. If this can have 40MB depends on the size and the data. Uniform coloured areas can be compressed better, high iso noise is worse. Regards Martin _______________________________________________ Digikam-users mailing list [hidden email] https://mail.kde.org/mailman/listinfo/digikam-users |
Free forum by Nabble | Edit this page |