New 4K & 5K iMacs support 10-bit screen color for improved image accuracy

124»

Comments

  • Reply 61 of 74
    indyfxindyfx Posts: 321member
    Quote:
    Originally Posted by polymnia View Post

     



    I made some samples based on your gradient image showing various amounts of dither to mitigate the banding. These techniques would be standard operating procedure if I were tasked with building a gradient like this.

     



    It's worth noting that the above inline image is NOT the original as saved out of Photoshop in both my & Marvin's posts. You need to go into the image gallery tool and click on the 'Original' link to get the actual, unprocessed original file.




    I think you are missing the point of a 10 bit display,(or perhaps are just confused)-it doesn't change the image (file)-, it allows you to see higher bit depth images as  they actually are.

    Let me explain, I have 2 extended depth images, one of them has some banding inherent in the image and the other does not, on an 8 bit display you see both as banded  (and dithering gradients is an 8 bit workaround and typically should not be performed on high bit depth (or raw images))

    That's the problem on an 8 bit monitor you can't see if the file actually has color banding or if it is just you monitor adding it -because it is unable to display all of the levels that the image contains.-

     

    this is similar to the matte screen display discussion, if a matte screen monitor adds luma bloom to the image -you see- then you can't really tell if the image does have "X" amount of luma bloom or if it's just the monitor showing it (even though it isn't there).

     

    Make sense?

  • Reply 62 of 74
    polymniapolymnia Posts: 1,080member
    Quote:

    Originally Posted by IndyFX View Post

     



    I think you are missing the point of a 10 bit display,(or perhaps are just confused)-it doesn't change the image (file)-, it allows you to see higher bit depth images as  they actually are.

    Let me explain, I have 2 extended depth images, one of them has some banding inherent in the image and the other does not, on an 8 bit display you see both as banded  (and dithering gradients is an 8 bit workaround and typically should not be performed on high bit depth (or raw images))

    That's the problem on an 8 bit monitor you can't see if the file actually has color banding or if it is just you monitor adding it -because it is unable to display all of the levels that the image contains.-

     

    this is similar to the matte screen display discussion, if a matte screen monitor adds luma bloom to the image -you see- then you can't really tell if the image does have "X" amount of luma bloom or if it's just the monitor showing it (even though it isn't there).

     

    Make sense?




    Yes, makes total sense.

     

    And in this case the file has is not extended depth and is banding because there aren't enough color levels. This image would band on an 8 bit or a 10 bit display without some additional work. So I don't see how it makes the point you are referring to.

     

    Being the type of image it is, 8 bit workarounds are completely appropriate. Applied as an edit leaving the original intact, of course.

     

    To really drive home the point you suggest, the gradient should have been built in 16 bit per channel mode in Photoshop. Marvin could have then shown it to us, assuring everyone that even though we see banding on our 8 bit displays, there is more data in the file that would only become apparent if we viewed through a 10 bit display pipeline.

     

    The thing is, I always start my work at the highest res mode, but my deliverables are 8 bit per channel assets most of the time. Web formats must be delivered in 8 bit (or less). Even print is more reliable in 8 bit mode if you don't want to hear back from less-than-cutting-edge vendors who have a hard time processing high bit depth image files.

     

    My point being, 8 bit is where the rubber hits the road right now. It's great to have the high bit depth originals for editing and authoring. But for most uses in my business (print & web design) the file will eventually be shoehorned into an 8 bit file.

     

    Maybe I missed the point of the demonstration, but you may be missing my point that even having all the high end equipment and bit depth workflow to diagnose banding, simulate end user's views, etc., we will still have to use traditional measures in many cases at final output time.

     

    If that green gradient were provided to me in a high bit depth non-banded format to incorporate into a web asset, I'd still have to do something about the banding, because my deliverable is an 8 bit file.

     

    I guess I'm less worried about banding 'being there' in my original and more worried about it existing in my deliverable.

  • Reply 63 of 74
    indyfxindyfx Posts: 321member
    Quote:
    Originally Posted by polymnia View Post

     



    Yes, makes total sense.

     

    And in this case the file has is not extended depth and is banding because there aren't enough color levels. This image would band on an 8 bit or a 10 bit display without some additional work. So I don't see how it makes the point you are referring to.

     

    Being the type of image it is, 8 bit workarounds are completely appropriate. Applied as an edit leaving the original intact, of course.

     

    To really drive home the point you suggest, the gradient should have been built in 16 bit per channel mode in Photoshop. Marvin could have then shown it to us, assuring everyone that even though we see banding on our 8 bit displays, there is more data in the file that would only become apparent if we viewed through a 10 bit display pipeline.

     

    The thing is, I always start my work at the highest res mode, but my deliverables are 8 bit per channel assets most of the time. Web formats must be delivered in 8 bit (or less). Even print is more reliable in 8 bit mode if you don't want to hear back from less-than-cutting-edge vendors who have a hard time processing high bit depth image files.

     

    My point being, 8 bit is where the rubber hits the road right now. It's great to have the high bit depth originals for editing and authoring. But for most uses in my business (print & web design) the file will eventually be shoehorned into an 8 bit file.

     

    Maybe I missed the point of the demonstration, but you may be missing my point that even having all the high end equipment and bit depth workflow to diagnose banding, simulate end user's views, etc., we will still have to use traditional measures in many cases at final output time.

     

    If that green gradient were provided to me in a high bit depth non-banded format to incorporate into a web asset, I'd still have to do something about the banding, because my deliverable is an 8 bit file.

     

    I guess I'm less worried about banding 'being there' in my original and more worried about it existing in my deliverable.




    See there you go, it is not useful to every consumer/hobbiest (who typically work in 8 bits).  However, many (most?) professional photographers are now shooting and delivering (for printing / further compositing) in RAW or raw converted to extend depth CMYK to printers (magazine)

    (BTW 4color offset is one of the reasons you shouldn't dither images (particulary not raw images), it can create Moirés later in the process) 

    Pro Film (video) is also nearly always at higher bit depth and as I said even at the hobbyist /prosumer level there are many cameras and add ons (software) that allow raw or high bit depth images to be recorded a huge asset to low/no budget "indy" filmmakers.

  • Reply 64 of 74

    Cool thread, interesting comments. Thanks guys. 

  • Reply 65 of 74
    polymniapolymnia Posts: 1,080member
    indyfx wrote: »

    See there you go, it is not useful to every consumer/hobbiest (who typically work in 8 bits).  However, many (most?) professional photographers are now shooting and delivering (for printing / further compositing) in RAW or raw converted to extend depth CMYK to printers (magazine)
    (BTW 4color offset is one of the reasons you shouldn't dither images (particulary not raw images), it can create Moirés later in the process) 
    Pro Film (video) is also nearly always at higher bit depth and as I said even at the hobbyist /prosumer level there are many cameras and add ons (software) that allow raw or high bit depth images to be recorded a huge asset to low/no budget "indy" filmmakers.

    I don't find many photographers providing me with RAW images. Most consider that proprietary intellectual property. They certainly shoot RAW and edit them as high bit depth artwork. But when they are delivered to me they are generally saved down to 24 bit TIFFs. I can sometimes convince them to send me a specific image in RAW format if I need something that I cannot extract from the rendered TIFFs.

    I think you misunderstand my use of dither. Remember, I was working on a synthetic gradient. When preparing a synthetic gradient for offset printing it is common practice to introduce noise. Offset is quite susceptible to banding because there are even more transformations applied to artwork to accommodate the conversion to CMYK, dot gain, etc. in fact, back in the day I remember having to use the Scitex Blends plugin for Quark that replaced Quark native blends in the RIP with added-noise gradients to avoid banding. If the technique is good enough for Scitex, it's good enough for me. I specifically said you wouldn't need to use these techniques on a captured original as the artwork comes with noise built in.

    I realize you want to put my thoughts in the non-pro box and dismiss what I've got to say. But it's simply not true. I make my living in Photoshop editing images. Everything from RAW conversions to building GIF animations. I'm pretty well versed in the digital arts.

    I take all your points about being able to view high bit depths. It is a big step forward to get a better view of high-bit depth images. Hell, I'm excited about it myself. I am just adding after all the 'hooray for high bit depth viewing!' business, it is quite possible (I'd suggest quite likely) your artwork will need to work in 8 bit as well. Even if you are a pro.
  • Reply 66 of 74
    hmmhmm Posts: 3,405member
    Quote:
    Originally Posted by IndyFX View Post

     



    See there you go, it is not useful to every consumer/hobbiest (who typically work in 8 bits).  However, many (most?) professional photographers are now shooting and delivering (for printing / further compositing) in RAW or raw converted to extend depth CMYK to printers (magazine)


    I'm confused by extended depth here. 16 bits per channel is common, as photoshop can save as a background process now. It really sucked back when it took several minutes to save a 16 bpc file of significant resolution with 50+ layers. The layers add up though if you have to do a lot of color matching and as you mentioned compositing.

     

    Even today I don't really see a reason for them to go past Adobe RGB for most current applications (will happen more in the future). CMYK mainly falls outside it in the maroons. Adobe provides prophoto rgb as a destination space, but it has its own issues. I'm not familiar with any really stable workflows that output straight to a linearized/hdr encoding from Lightroom or Capture One, and they're still probably the most widely used raw processors (and I somehow still remember Leaf Capture). I think your points are probably much more common in video and vfx than they would be capturing still images even today.

     

    Quote:

    (BTW 4color offset is one of the reasons you shouldn't dither images (particulary not raw images), it can create Moirés later in the process) 



     

    I've never seen that due to dithering. Dithering is basically a noise pattern. If I needed to get rid of moire or banding, I would probably use noise in the first place. There are obviously other ways, and it's probably ideal to apply that to the destination gamut. I am highly skeptical on this, given that noise is used as a kind of anti-pattern. It introduces the impression of randomization. That would be a very bad dithering implementation, although I haven't seen this, which makes it difficult to add a meaningful comment.

     

    Quote:


    Pro Film (video) is also nearly always at higher bit depth and as I said even at the hobbyist /prosumer level there are many cameras and add ons (software) that allow raw or high bit depth images to be recorded a huge asset to low/no budget "indy" filmmakers.


     

    That's more typical in video. Linear workflows are much better supported, and video started to rely on LUTs long ago as opposed to ICC profiles and all of their broken features. That kind of workflow requires more bits due to its relative sparsity and the use of floating point encodings. You can't really use floating point encodings with 8 bits, as that would leave only 4-5 mantissa bits.

  • Reply 67 of 74
    indyfxindyfx Posts: 321member
    Quote:

    Originally Posted by hmm View Post

     

    I'm confused by extended depth here. 16 bits per channel is common, as photoshop can save as a background process now. It really sucked back when it took several minutes to save a 16 bpc file of significant resolution with 50+ layers. The layers add up though if you have to do a lot of color matching and as you mentioned compositing.

     

    Even today I don't really see a reason for them to go past Adobe RGB for most current applications (will happen more in the future). CMYK mainly falls outside it in the maroons. Adobe provides prophoto rgb as a destination space, but it has its own issues. I'm not familiar with any really stable workflows that output straight to a linearized/hdr encoding from Lightroom or Capture One, and they're still probably the most widely used raw processors (and I somehow still remember Leaf Capture). I think your points are probably much more common in video and vfx than they would be capturing still images even today.

     

     

    I've never seen that due to dithering. Dithering is basically a noise pattern. If I needed to get rid of moire or banding, I would probably use noise in the first place. There are obviously other ways, and it's probably ideal to apply that to the destination gamut. I am highly skeptical on this, given that noise is used as a kind of anti-pattern. It introduces the impression of randomization. That would be a very bad dithering implementation, although I haven't seen this, which makes it difficult to add a meaningful comment.

     

     

    That's more typical in video. Linear workflows are much better supported, and video started to rely on LUTs long ago as opposed to ICC profiles and all of their broken features. That kind of workflow requires more bits due to its relative sparsity and the use of floating point encodings. You can't really use floating point encodings with 8 bits, as that would leave only 4-5 mantissa bits.


     

    Both of you are just spouting nonsense at this point, Im done. 10 bit displays real and important feature for both pro's and for consumers.

    Sorry you two seem to have so much butthurt over the fact that the iMacs have 10 bit displays (5K 10 bit displays no less) but you could fix this in a jiffy. Junk your PC's and buy a mac, then it wouldn't be so weird that you constantly hang out posting on a mac news site.

  • Reply 68 of 74
    polymniapolymnia Posts: 1,080member
    indyfx wrote: »
    Both of you are just spouting nonsense at this point, Im done. 10 bit displays real and important feature for both pro's and for consumers.
    Sorry you two seem to have so much butthurt over the fact that the iMacs have 10 bit displays (5K 10 bit displays no less) but you could fix this in a jiffy. Junk your PC's and buy a mac, then it wouldn't be so weird that you constantly hang out posting on a mac news site.

    Am I one of the 'you two' you refer to?

    I've been using macs since 1986. And I'm very pleased the iMac (and other macs connected to appropriate displays) can carry a 10 bit per channel stream. I'm also interested in how this fits into a larger workflow.

    I'm glad to hear you are done because I don't think you understand what I'm saying, and many things you have said don't square with what I have been doing for years to deliver graphic assets to my clients and employers. You don't seem interested in trying to understand what I'm saying. You seem far more interested correcting points I haven't made and contradicting image processing techniques I know to be valid. Believe it or not, I'm trying to understand you and have a constructive discussion. But I can't do all the work myself.

    Have a nice day :)
  • Reply 69 of 74
    hmmhmm Posts: 3,405member
    Quote:
    Originally Posted by IndyFX View Post

     

     

    Both of you are just spouting nonsense at this point, Im done. 10 bit displays real and important feature for both pro's and for consumers.

    Sorry you two seem to have so much butthurt over the fact that the iMacs have 10 bit displays (5K 10 bit displays no less) but you could fix this in a jiffy. Junk your PC's and buy a mac, then it wouldn't be so weird that you constantly hang out posting on a mac news site.




    I'm starting to wonder if you're an internet troll. Some of what you said was accurate, but nothing in my post warranted that response. I've also used Macs exclusively since OSX 10.3. I used one before that running some version of OS 9, but I didn't own that one. I used them for work as well. I've had to test displays and RIPs and all kinds of other stuff, which is why I asked you these kinds of questions. Some of your posts make very little sense, which I thought was an issue of wording. Your response really says more about you than it does about me though.

     

    Edit : I just re-read my last post. First of all extended depth isn't a real term. I wanted to know whether you meant linear floating point encodings, which allow for values outside a normalized range or just a wider gamut. As I mentioned these things are also applied differently to video, where color management is handled primarily by LUTs. It's important to realize these are two different things, and you didn't really describe one or the other. One is used for high dynamic range encodings such as HDR and EXR. The other just means your normalized range (signed or unsigned) is fit to a broader gamut. You will probably clip fewer values, but you are better off dealing with things that might clip in the raw processing stage. The only ones that might fall outside of certain RGB spaces and still fit in cmyk under normal circumstances are dark purples/maroons. i'm ignoring anything using a touchplate or any kind of spot color here, because no one accounts for that as a part of raw processing unless they output straight to cmyk.

  • Reply 70 of 74
    indyfxindyfx Posts: 321member
    Quote:

    Originally Posted by hmm View Post

     



    I'm starting to wonder if you're an internet troll. Some of what you said was accurate, but nothing in my post warranted that response. I've also used Macs exclusively since OSX 10.3. I used one before that running some version of OS 9, but I didn't own that one. I used them for work as well. I've had to test displays and RIPs and all kinds of other stuff, which is why I asked you these kinds of questions. Some of your posts make very little sense, which I thought was an issue of wording. Your response really says more about you than it does about me though.


    You wrote some very bizarre things (some just wrong some unrelated and some just weird tangents) that indicate you don't really understand what you are espousing and that it's far more likely you just looked up stuff on the internet and regurgitated it.

    Just saying you have had  XX mac's since XXXX is standard operating procedure BTW

     

    The actual point is that the new iMac's 10 bit display capability is real, visible and a tangible benefit to most all users: consumers, prosumers and pro's. 

  • Reply 71 of 74
    polymniapolymnia Posts: 1,080member
    Quote:
    Originally Posted by IndyFX View Post

     

    You wrote some very bizarre things (some just wrong some unrelated and some just weird tangents) that indicate you don't really understand what you are espousing and that it's far more likely you just looked up stuff on the internet and regurgitated it.

    Just saying you have had  XX mac's since XXXX is standard operating procedure BTW

     

    The actual point is that the new iMac's 10 bit display capability is real, visible and a tangible benefit to most all users: consumers, prosumers and pro's. 




    That is about the most obvious point you could make, certainly one that doesn't require assassination of anyone's professionalism or intellect to make. I don't think anyone is arguing that point. And it would be a really boring discussion if that's where we ended it.

     

    I don't feel the need to give you (and make everyone else read) a full CV to prove my bona fides, but you can follow the link in my sig if you'd like to learn more about me. It's pretty safe to say I'm not an imposter. I don't really have any reason to believe (or not believe) you have the knowledge you claim to have, but I am giving you the benefit of the doubt and not assassinating your intellect. I've been bringing up considerations related to bit depth, deliverables and edits. I've commented on and contributed example artwork demonstrating the behavior of low bit depth artwork. Which, being the opposite of high bit depth artwork, is useful to demonstrate the advantages (and possible false sense of security) a high bit depth workflow provides. I don't feel I've gotten off topic.

     

    I think the point of discussion is to explore ideas. I'm sorry if mine have made you angry.

  • Reply 72 of 74
    polymniapolymnia Posts: 1,080member
    Quote:



    Originally Posted by IndyFX View Post

     

    You wrote some very bizarre things (some just wrong some unrelated and some just weird tangents) that indicate you don't really understand what you are espousing and that it's far more likely you just looked up stuff on the internet and regurgitated it.

    Just saying you have had  XX mac's since XXXX is standard operating procedure BTW

     

    The actual point is that the new iMac's 10 bit display capability is real, visible and a tangible benefit to most all users: consumers, prosumers and pro's. 


     

    Quote:
    Originally Posted by polymnia View Post

     



    That is about the most obvious point you could make, certainly one that doesn't require assassination of anyone's professionalism or intellect to make. I don't think anyone is arguing that point. And it would be a really boring discussion if that's where we ended it.

     

    I don't feel the need to give you (and make everyone else read) a full CV to prove my bona fides, but you can follow the link in my sig if you'd like to learn more about me. It's pretty safe to say I'm not an imposter. I don't really have any reason to believe (or not believe) you have the knowledge you claim to have, but I am giving you the benefit of the doubt and not assassinating your intellect. I've been bringing up considerations related to bit depth, deliverables and edits. I've commented on and contributed example artwork demonstrating the behavior of low bit depth artwork. Which, being the opposite of high bit depth artwork, is useful to demonstrate the advantages (and possible false sense of security) a high bit depth workflow provides. I don't feel I've gotten off topic.

     

    I think the point of discussion is to explore ideas. I'm sorry if mine have made you angry.

     


     

     

    Actually, I'm feeling a bit less sorry. I thought your handle looked familiar and looked at some of your other posts around here. You have quite a fondness for negativety. I found this exchange we had some time ago, one of my more unpleasant experiences here at AI:

     

    http://forums.appleinsider.com/t/188032/apple-installs-banners-flags-at-bill-graham-civic-auditorium-ahead-of-sept-9-event#post_2771156

     

    And just before that he unloaded on another AI commenter with this gem:

     

    http://forums.appleinsider.com/t/188028/runtastic-debuts-moment-smartwatch-combining-analog-face-fitness-tracking#post_2771041

     

    Welcome to my block list :)

  • Reply 73 of 74
    hmmhmm Posts: 3,405member
    Quote:

    Originally Posted by IndyFX View Post

     

    You wrote some very bizarre things (some just wrong some unrelated and some just weird tangents) that indicate you don't really understand what you are espousing and that it's far more likely you just looked up stuff on the internet and regurgitated it.

    Just saying you have had  XX mac's since XXXX is standard operating procedure BTW

     

    The actual point is that the new iMac's 10 bit display capability is real, visible and a tangible benefit to most all users: consumers, prosumers and pro's. 


    I'm not getting into a debate here over who uses what. I gave you that comment in response to a silly accusation on your part, so try to consider it in context. As for the rest, it's mixed up because you posted about a number of different things without using common names. I surmised from the "FX" in your name that this probably influences your needs in terms of image processing, so there are some implicit assumptions.

     

    Regarding the imacs, I don't believe I said there wouldn't be any tangible benefit. I said you need to examine a working unit when it comes out rather than rely solely on specs. Specs can be misleading. It's certainly an update. You should pay more attention to where it's at after it comes out. Some of the early 10 bit displays were actually rated rather poorly at the time, but that was five years ago.

     

    If you want to clarify what you actually meant by extended depth, I'm happy to respond on the relevant concept. You could be talking about linear encodings (the mention to video suggests it) where your normalized range is 0 to 1, but you can store values outside that range. You could be simply talking about higher bit depths in general the way 16 bpc works in photoshop, meaning the same range with coordinates spaced tighter than they are at 8. You could have tied some concept related to gamut boundaries in there somewhere. It's not a well defined term, so I'm guessing. Perhaps you should try to be a better communicator :).

Sign In or Register to comment.