Site Sponsors
Page 2 of 2 FirstFirst 1 2
Results 51 to 58 of 58

Thread: Compressed VS Uncompressed DNGs - A quick test

  1. #51
    Senior Member MaxKißler's Avatar
    Join Date
    Nov 2010
    Location
    Berlin
    Posts
    387
    Post Thanks / Like

    Re: Compressed VS Uncompressed DNGs - A quick test

    Quote Originally Posted by Shashin View Post
    Why? Don't you like higher bit depths? All the tones are in the same place in the visual gamut with an 8-bit or 16-bit image, you just don't have as many with 8-bit. This is what effectively happens with different gamut sizes with a given image gamut--ProPhotoRGB can results in fewer discrete colors. That could be important if you process your images. That is just how color is encoded in gamuts.
    ...
    I'm sorry I couldn't follow you. Could you please explain this further. What did this have to do with bit depth? Were you referencing the compressed raw files which, I guess (looking at file sizes), are being downscaled to 8 bits per channel?

    Thanks

  2. #52
    Super Duper
    Senior Member
    Shashin's Avatar
    Join Date
    May 2011
    Location
    Florida, USA
    Posts
    4,501
    Post Thanks / Like
    Images
    141

    Re: Compressed VS Uncompressed DNGs - A quick test

    Quote Originally Posted by MirekE View Post
    Yes, I think I am missing this. Would you mind giving an example?
    Lets use 8-bit because the numbers are smaller. 8-bit has 256 from 0 to 255 for each channel. Those RGB values represent all the possible coordinates in a color space. Regardless of the size of the gamut, there cannot be any more colors in a color space than 256 X 256 X 256.

    Since the number of coordinate in a space is fixed, for it to cover a larger gamut, there must be a larger distance between coordinates. Colors that exist between those coordinates have to be binned up or down.

    Now, if I have an image with a small gamut, the number discrete colors that can be created depend on how many coordinates in the color space fit. The larger the color space, the fewer the coordinates. If you have an image of a brick wall, ProPhotoRGB will bin those colors in the wall into fewer bins than AdobeRGB.

    Now, you can actually see the effect on a histogram. Open an image in Photoshop in AdobeRGB and then covert the profile to ProPhotoRGB. Undo and redo while watching the histogram. You will see a change in the histogram area and a shift to the left for ProPhotoRGB. And as you know, when things shift to the left in a histogram, it is not ideal as there is a compression in the information, even though the appearance may not change. Now, the significance will depend on the image and the processing of the image. But the difference is there.

  3. #53
    Administrator Bob's Avatar
    Join Date
    Oct 2007
    Location
    Prescott, Arizona
    Posts
    4,492
    Post Thanks / Like
    Images
    367

    Re: Compressed VS Uncompressed DNGs - A quick test

    Quote Originally Posted by 250swb View Post
    Yes we all know about calibrated monitors etc. but if you can't see what a colour space does for you, why use it? It is adding additional information at the printing stage that you can't see on your monitor, so it's a crap shoot, you glory in it when it goes well, you go back to stage one (something closer to what you see on the monitor) when it does things you don't expect. ProPhoto is a great idea but if it isn't seen on a monitor how can it convey what you want in post processing?

    Steve
    Not really that at all.
    Digital sensors have the potential for recording images quite outside of any "usual" gamut such a proPhoto. When those images are rendered into something like a 16 bit tiff, they are converted to the user selected colorspace according to the rendering intent. Depending on your raw converter, sometimes the intent is followed, sometimes not quite as well, but in any case, I now have an image that has a fairly small amount of compression in the color domain.
    While I am editing, I usually soft proof and show areas that are outside the destination gamut and use my judgement about how I would like to manage this.
    I sometimes assign (not convert) other profiles, such as one of those offered by Joseph Holmes Joseph Holmes - RGB Working Space Profiles.
    During editing the image may be manipulated in a number of ways that move color and saturation around until I get what I want.
    Usually I end up with something that falls within the colorspace of my output device, sometimes I don't, but the gamut warning is useful in localizing those areas. I might select my rendering intent for output depending on the device and the image to produce the results I want.
    Is it a crap shoot, well every time you push the shutter button it is to some degree, but I want as much control as possible.
    Premature conversion of tristimulus values during the tiff generation to a small color space simply reduces the information that the file contains. I prefer to remove as little as possible until the end.
    -bob
    Likes 1 Member(s) liked this post

  4. #54
    Administrator Bob's Avatar
    Join Date
    Oct 2007
    Location
    Prescott, Arizona
    Posts
    4,492
    Post Thanks / Like
    Images
    367

    Re: Compressed VS Uncompressed DNGs - A quick test

    To illustrate the point I am trying to make, I took one raw file and generated two 16 bit tiffs, one in sRGB, and one ProPhoto.
    I used a tool to compare the two images and to mark each pixel where there is a difference. Black means a difference, clear means no difference.

    As you can see, although the differences appear subtle when viewed on-screen, there are enough differences between the 16 bit tiffs to cause concern.
    -bob
    Likes 2 Member(s) liked this post

  5. #55
    Super Duper
    Senior Member

    Join Date
    Dec 2007
    Location
    Germany
    Posts
    3,623
    Post Thanks / Like

    Re: Compressed VS Uncompressed DNGs - A quick test

    IMO space of cards or disk space or processing speed are nowadays no series reasons to accept any possible compromise in IQ when you spend thousands for lenses and cameras, why save some few dollars on drives and cards?
    Thats why I just witch to no compression.

  6. #56
    Senior Member MaxKißler's Avatar
    Join Date
    Nov 2010
    Location
    Berlin
    Posts
    387
    Post Thanks / Like

    Re: Compressed VS Uncompressed DNGs - A quick test

    Quote Originally Posted by Shashin View Post
    Lets use 8-bit because the numbers are smaller. 8-bit has 256 from 0 to 255 for each channel. Those RGB values represent all the possible coordinates in a color space. Regardless of the size of the gamut, there cannot be any more colors in a color space than 256 X 256 X 256.

    Since the number of coordinate in a space is fixed, for it to cover a larger gamut, there must be a larger distance between coordinates. Colors that exist between those coordinates have to be binned up or down.

    Now, if I have an image with a small gamut, the number discrete colors that can be created depend on how many coordinates in the color space fit. The larger the color space, the fewer the coordinates. If you have an image of a brick wall, ProPhotoRGB will bin those colors in the wall into fewer bins than AdobeRGB. ...
    That's a good point you're making and is defintely worth considering in one's color management workflow.

    However I'm always doing as much processing as possible at raw stage, export 16bit tiffs in ProphotoRGB and only do minor adjustments, retouching if necessary and color management in PS. So far I've never had any problems with ProPhotoRGB- works for me.

    Regarding your example with the brick wall: Doesn't it also depend on the subject (in this case the brick wall)? If the bricks are rather desaturated it makes sense but as soon as the sun is shining for example and the bricks are well lit and saturated, in AdobeRGB there will/might be colors that are out of gamut.


    Regards
    Last edited by MaxKißler; 16th December 2013 at 10:31.

  7. #57
    Member
    Join Date
    Mar 2012
    Posts
    49
    Post Thanks / Like

    Re: Compressed VS Uncompressed DNGs - A quick test

    Quote Originally Posted by Shashin View Post
    Lets use 8-bit because the numbers are smaller. 8-bit has 256 from 0 to 255 for each channel. Those RGB values represent all the possible coordinates in a color space. Regardless of the size of the gamut, there cannot be any more colors in a color space than 256 X 256 X 256.

    Since the number of coordinate in a space is fixed, for it to cover a larger gamut, there must be a larger distance between coordinates. Colors that exist between those coordinates have to be binned up or down.

    Now, if I have an image with a small gamut, the number discrete colors that can be created depend on how many coordinates in the color space fit. The larger the color space, the fewer the coordinates. If you have an image of a brick wall, ProPhotoRGB will bin those colors in the wall into fewer bins than AdobeRGB.

    Now, you can actually see the effect on a histogram. Open an image in Photoshop in AdobeRGB and then covert the profile to ProPhotoRGB. Undo and redo while watching the histogram. You will see a change in the histogram area and a shift to the left for ProPhotoRGB. And as you know, when things shift to the left in a histogram, it is not ideal as there is a compression in the information, even though the appearance may not change. Now, the significance will depend on the image and the processing of the image. But the difference is there.
    Thank you. Yes, I get that. That's why I was talking about the posterization. Less values in extreme gradients could cause visible steps and it apparently does happen in 8 bit. ProPhoto is more than 2x larger than AdobeRGB, so the chances are higher.

  8. #58
    Member
    Join Date
    Apr 2009
    Location
    Vancouver, BC
    Posts
    221
    Post Thanks / Like
    Images
    1

    Re: Compressed VS Uncompressed DNGs - A quick test

    With the M8 and M9 I tested for the difference in image content between compressed and uncompressed, and could never detect a difference.

    As these cameras are rather slow in clearing their buffers and take about twice as long to do so shooting uncompressed, I decided to shoot compressed for everything except static scenes. Even when shooting uncompressed and in single shot mode I still often have to wait for the cameras. That's one of the main reasons I got the M.

    With the M there is of course no point in shooting uncompressed, and the Monochrom will not allow in camera compression, so these are not part of the discussion.

    Henning

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •