Bug? Sensor pattern artefacts in overexposed images

Commentaires

41 commentaires

  • flashbangwallop
    I have seen the same yellow artefacts with over exposed images too, shooting X-T2 and using the full pro version of CO11. However I do have other, possibly graphic card, funnies using the Black & White tool, so I assumed it was graphics card again. I'm using Windows 7 and a GeForce 8600GT card, all about 6 years old.
    Martin
    0
    Actions pour les commentaires Permalien
  • SFA
    As an experiment go into the Preferences In the "General" tab and set the values to Hardware Acceleration to "None".

    Then try the images again to see what happens.

    It may be worth restarting Capture One before retrying.

    Also, with or without the re-start, what happens if you reduce the Exposure setting. Does it change the effect or does it remain and simply become darker?

    Once you know the answers to those suggestion I would suggest creating a Support Case to let the Capture One Support team have a look at the problem studying the log files and also using the sample image to see what they can discover.

    HTH.


    Grant
    0
    Actions pour les commentaires Permalien
  • P1337
    [quote="flashbangwallop" wrote:
    I have seen the same yellow artefacts with over exposed images too, shooting X-T2 and using the full pro version of CO11. However I do have other, possibly graphic card, funnies using the Black & White tool, so I assumed it was graphics card again. I'm using Windows 7 and a GeForce 8600GT card, all about 6 years old.
    Martin


    This triggered me first as I remembered to have also a GeForce 8600GT, but I then I realized to have a GeForce GTX 660Ti since 2 years now. And Windows 10. But interesting, that I am not alone!! Do you had the issue with CO 11.3 or an older version?

    [quote="SFA" wrote:
    As an experiment go into the Preferences In the "General" tab and set the values to Hardware Acceleration to "None". [...]It may be worth restarting Capture One before retrying.

    The support just told me the same, no success.

    [quote="SFA" wrote:
    Also, with or without the re-start, what happens if you reduce the Exposure setting. Does it change the effect or does it remain and simply become darker?

    It just gets darker.

    [quote="SFA" wrote:
    Once you know the answers to those suggestion I would suggest creating a Support Case to let the Capture One Support team have a look at the problem studying the log files and also using the sample image to see what they can discover.

    A few minutes ago they asked for that, I'll send it in.
    0
    Actions pour les commentaires Permalien
  • flashbangwallop
    Re suggestions from SFA (thank you), I also find that hardware acceleration does not make a difference and that the patch gets darker when reducing the exposure.
    Martin
    0
    Actions pour les commentaires Permalien
  • P1337
    [quote="flashbangwallop" wrote:
    hardware acceleration does not make a difference and that the patch gets darker when reducing the exposure.

    Thanks for confirming this. I submitted the log and sysinfo files to the support yesterday, but got no reply yet. Maybe it helps the support, if you also would submit your data to find the root of the problem? For reference you can mention my case number: 301270

    You can open a support case here:


    How to find the log files and create a system information:
    0
    Actions pour les commentaires Permalien
  • SFA
    [quote="flashbangwallop" wrote:
    Re suggestions from SFA (thank you), I also find that hardware acceleration does not make a difference and that the patch gets darker when reducing the exposure.
    Martin


    Definitely a Support Case issue in that case.

    The only other quick check of "what happens if.." I can think of would be to clip the top end exposure (rather than adjust it) by using the Levels tool.

    The objective would be to move that high exposure data out of the process completely to see what happens.

    I sometimes see a pink colouration in slightly overexposed areas of some images from a couple of my older Canon cameras and that appears to be related to peaks in the red channel at the top end of the exposure in individual files. Two versions of the same shot taken in continuous mode a fraction of a second apart can have one showing the effect and the other perfectly OK. Clipping the red channel by a tiny amount on the problem shot usually resolves it (as does using the problem free version!).

    However the extreme pixellation seen in the originally posted example is not present in my Canon images so the cause may be very likely be different and the suggestion may therefore not work at all.

    HTH.

    Grant
    0
    Actions pour les commentaires Permalien
  • P1337
    [quote="SFA" wrote:
    The only other quick check of "what happens if.." I can think of would be to clip the top end exposure (rather than adjust it) by using the Levels tool.

    The objective would be to move that high exposure data out of the process completely to see what happens.

    Ok, I tried that too, but the pixelated area persists. They have normal color and brightness values, just like any yellow and red pixel. So these pixel patterns seem not to be just a wrong rendering in the C1 image viewer, but a complete misinterpretation of the RAW file.
    0
    Actions pour les commentaires Permalien
  • flashbangwallop
    Grant, I also find that levels has no effect. Tech support case from me has been instigated.
    Martin
    0
    Actions pour les commentaires Permalien
  • P1337
    After the suggestion of updating the graphic cards driver and no success, the support forwarded the case to R&D. Maybe there will be a fix in future releases.

    Guess we have to hope then... 😕
    0
    Actions pour les commentaires Permalien
  • daveh_72
    I'm glad I found this thread, as I'm having exactly the same issue and it's been driving me mad 😊

    OK so the real answer is not to blow the channels, but sometimes a tiny bit of "blowout" isn't a big issue for the picture.

    But Capture One is insistent on applying this strange pattern to the blown highlights and then exporting it with the image.

    Always seems to be on yellows (or possibly it's down to the red channel?)

    I did notice that in some images I took on Saturday, the green channel had blown in a couple of tiny areas and it had put green "sensor pattern" blotches on the preview. BUT these did not export with the image like the yellow ones (in the same image) do.

    It's somewhat frustrating as it's not really correctable with any photoshop work either :-/
    0
    Actions pour les commentaires Permalien
  • SFA
    As a guess here I would think that if there is anything you can do (other than avoid the reported modes that seem to induce the problem) the most likely best option would be at the RAW conversion step rather than some sort of Post processing.

    As I don't have a file with which to experiment with I can't offer any suggestions based on experience but if there is a way to tweak the settings for the values used in the RAW conversion in a way that is satisfactory then that would be the most likely immediately available "best approach" to the problem at this time.


    HTH.


    Grant
    0
    Actions pour les commentaires Permalien
  • P1337
    [quote="SFA" wrote:
    As a guess here I would think that if there is anything you can do (other than avoid the reported modes that seem to induce the problem) the most likely best option would be at the RAW conversion step rather than some sort of Post processing.

    Which reported modes? What do you mean by doing anything about the RAW conversion? This is a bug in the RAW-conversion of CO itself. We cannot manipulate the RAW-conversion of CO, or do you know a way to do that?


    @daveh_72: What camera are you using? X-T2?
    0
    Actions pour les commentaires Permalien
  • daveh_72
    yeah it's in the image no matter what I try at the import stage and I can't get rid of it.

    It's not there in the camera's previews or full JPEGs of course!

    Doesn't matter what profile I try either, or what processing engine.

    It almost looks as if the clipping warning is being sent out with the processed image? But that's not based on any expertise, just guesswork on my part! 😊

    X-T20 here, so same sensor as the X-T2
    0
    Actions pour les commentaires Permalien
  • SFA
    [quote="NNN634449929237185266" wrote:
    [quote="SFA" wrote:
    As a guess here I would think that if there is anything you can do (other than avoid the reported modes that seem to induce the problem) the most likely best option would be at the RAW conversion step rather than some sort of Post processing.

    Which reported modes? What do you mean by doing anything about the RAW conversion? This is a bug in the RAW-conversion of CO itself. We cannot manipulate the RAW-conversion of CO, or do you know a way to do that?


    @daveh_72: What camera are you using? X-T2?


    Apologies re the modes comment - I was conflating this thread subject with the one running in the Mac section for a different problem.

    viewtopic.php?f=71&t=29072

    You may well be right about there being no adjustments that can resolve the issue right now but the conversion has a number of stages and it may be possible to exclude problematic data by certain means late in the process since the problem is at one of the extremes of data values. I don't know for certain but it might be worth a look as a temporary approach to getting images processed.

    If anyone has an example file they can share for me to download I would be happy to spend a few minutes investigating.

    (Unless of course the C1 support crew have already suggest a way around the problem as a temporary solution.)


    Grant
    0
    Actions pour les commentaires Permalien
  • daveh_72
    Hi Grant.

    Happy to share a file (although of course I'm hoping it gets fixed in an update soon!). What's the easiest way to do so? I'm not a big forum user.... 😊

    cheers

    Dave
    0
    Actions pour les commentaires Permalien
  • CCWH
    Hi,
    I have downgraded to Capture One 11.1 to fix this bug.
    With 11.2 this artefact was not fixed..
    0
    Actions pour les commentaires Permalien
  • SFA
    [quote="daveh_72" wrote:
    Hi Grant.

    Happy to share a file (although of course I'm hoping it gets fixed in an update soon!). What's the easiest way to do so? I'm not a big forum user.... 😊

    cheers

    Dave


    Hi Dave,

    Sorry, I thought I had replied to this earlier but I guess not. Typed it but maybe the send activity was interrupted before it was completed successfully.

    The only real option is to use a file sharing service such as Dropbox or OneDrive.

    Email can be limited by file sizes and there is no direct facility on the forum which is probably wise given the speed with which the largest of the modern RAW files could eat up storage capacity!


    Grant
    0
    Actions pour les commentaires Permalien
  • CCWH
    [quote="CCWH" wrote:
    Hi,
    I have downgraded to Capture One 11.1 to fix this bug.
    With 11.2 this artefact was not fixed..


    I have 2 PC and the results were different:
    - PC1: C1Pro11.3 installed. No "new" Fuji RAW processed. When i install CPro 11.2 no artefact on my RAW files
    - PC2: C1Pro11.3 installed. Some news Fuji RAW files processed with artefacts. With version 11.2 artefact are not removed. I need to install 11.1.1 to remove these artefacts. I have rebuiled preview pictures but it doesn't fix the problem.

    For my test i've used the same RAW files.
    0
    Actions pour les commentaires Permalien
  • P1337
    [quote="CCWH" wrote:
    I have 2 PC and the results were different


    Thanks for your check, that is the same observation also I made with 11.3. Artefacts on my PC, but no artefacts on my old laptop. Interesting that this problem didn't occur in 11.1! But I think I won't buy a very expensive piece of software just to only use the old version. 😊

    [quote="SFA" wrote:
    If anyone has an example file they can share for me to download I would be happy to spend a few minutes investigating.

    (Unless of course the C1 support crew have already suggest a way around the problem as a temporary solution.)


    The support didn't come up with any solution yet. Thank you very much for your willing to help! I have here two images to share: The window (already known) and an image of 2 bikes, where greenish/red artefacts are appearing on the right bike.

    0
    Actions pour les commentaires Permalien
  • SFA
    Hi Dave,

    I have looked at these images using C1 11.2 and 11.3.1 and do not see the problems previously reported during edit or after output to 2 different jpg sizes and compressions.

    The original example of the door and window is indeed extremely blown but using a Linear Curve and selecting for WB on the blown section seems to give a well balanced result considering the starting point as a test shot.

    The 2 bicycles recover well with some small blown out areas but they come up white with no artefacts other than a little colouration on some edges. That looks like diffraction effects but in fact responds to purple fringing adjustment despite not being purple or close to purple - but the cause is likely very similar.

    I can see it at 200% zoom but it is not a concern below that.

    All the evidence seems to suggest it is a local problem on a particular machine rather than a generic bug. However that maybe makes it more difficult to pin down exactly why the problem has appeared since there are so many possible variables.

    It probably would not do any harm to clear out the system and local Cache and also discard any old Temporary files.

    Perhaps the Support Team have already suggested that?

    My Spec, fr what the information may be worth;

    Dell Precision Mobile Workstation from about 2012.
    i7 CPU
    24GB RAM
    SSDs for primary and second drive
    Quadro K1000M GPU (Which shows very little use with your RAF files when running 11.3.1. The K1000M is in any case not at all powerful as a GPU.)
    Windows 7 Pro.


    I hope this helps in some way.


    Grant
    0
    Actions pour les commentaires Permalien
  • P1337
    Just to clarify: I'm Phil, not Dave 😜 (I don't know why I've got this cryptic number as user name...)

    Thanks for your test! You are right, this further clarifies that this issue is only present on some machines (luckily not on yours) and there seems to be nothing we can do about it yet, except reporting it to raise the developers awareness and hope for a fix.

    In case anyone has the same problem and wants to compare the hardware, here are my specs as well:

    AMD Phenom II X4 965, 3,4 GHz
    Nvidia Geforce GTX 660 Ti, driver 416.16 (older driver tested before)
    12 GB RAM
    SSD and HDD
    Win 10 Pro 1803 (and a lot older version a few days before)
    0
    Actions pour les commentaires Permalien
  • SFA
    [quote="NNN634449929237185266" wrote:
    Just to clarify: I'm Phil, not Dave 😜 (I don't know why I've got this cryptic number as user name...)

    Thanks for your test! You are right, this further clarifies that this issue is only present on some machines (luckily not on yours) and there seems to be nothing we can do about it yet, except reporting it to raise the developers awareness and hope for a fix.

    In case anyone has the same problem and wants to compare the hardware, here are my specs as well:

    AMD Phenom II X4 965, 3,4 GHz
    Nvidia Geforce GTX 660 Ti, driver 416.16 (older driver tested before)
    12 GB RAM
    SSD and HDD
    Win 10 Pro 1803 (and a lot older version a few days before)


    Sorry Phil!

    I did make an attempt to work out who I was replying to but must have settled on the wrong post when scrolling up and down the list.

    You are right, the cryptic user names can be tricky!

    They are issued to ensure, as far as possible, that new users don't all end up with names like "New User" or "First Name" or something. You can change it to a name of your choice by logging in and heading somewhere in the account management options. Not needed to visit there for a long time but I guess you will find where I mean without me needing to describe it!

    I think this is one where finding a pattern in all the technical stuff in the Log files (if there is one) may be one way forward. Also the stuff being reported by Windows.

    One of the nicer things about staying with WIn 7 is that not much changes these days!

    As for random things - an example of stuff that happens.

    My Android phone is about 4 years old I suppose. I have never updated the OS since new. It just worked and I don't place a lot of demand on it so why bother?

    After about 2+ years the "Torch" program stared doing odd things. It would switch on fine but sometimes not switch off no matter what I tried except if I activated the camera. That turned it off. Eventually it never turned off unless I turned the camera on. No big deal really but I was curious about what caused it.

    Eventually I got very low on internal memory and it was clearly not happy though still working pretty well most of the time. So I removed a load of music, which freed up 50% of the internal memory, and it become happier, not exactly immediately but over a day or three. And the Torch app now works as expected every time.

    So since the Torch problem started at about 60% memory used I can only assume it was not lack of memory per se that caused the problem but some facet of how the memory was being used. Removing the music files and clearing a few hundred Mb of cache presumably allowed the OS to tidy up its memory usage and so resolve whatever the problem was. Somehow.

    Windows, in my opinion and maybe also based on a rare experience or two of otherwise unexplained problems that seemed to self resolve, may throw up odd issues with file and file fragment management from time to time. As might "disk" management routines.

    Devilishly tricky to figure out - which is why the favourite approach from a corporate IT dept. perspective is to simply "re-image" any machine reported to them after some cursory checks. Sometimes not even the cursory checks!

    But it that suggestion does go some way towards an explanation of why a problem can exist on one machine and not another when they are both reasonably similar in terms of the general configuration employed and may have worked equally consistently at some previous point in time.

    I'd be very tempted to compare the sections of log files generated on each machine when running the same process and getting different results. There is no certainty that the particular problem has been predicted and is being monitored and of course it might not be something that can be identified readily at the point of process - after all it's not like the process is crashing - but it's about all there is to work with.

    Otherwise memory might be an issue. 12Gb should be OK but if you are running other applications at the same time I guess it's just possible it might lead to additional file swapping and that would perhaps increase the risk something unpredictable happening or having happened and still influencing results at a much later stage. That is purely speculation of course ....

    Hope you get it resolved completely very soon.


    Grant
    0
    Actions pour les commentaires Permalien
  • Gustavo Ferlizi
    Interesting, there seems to be no data at all in there.

    Can't reproduce it on 11.3.0 / Thinkpad X230 - Windows 10 up to date.

    Will be quite hard to track it down, if possible.
    0
    Actions pour les commentaires Permalien
  • CCWH
    [quote="SFA" wrote:
    Hi Dave,

    I have looked at these images using C1 11.2 and 11.3.1

    [...]

    All the evidence seems to suggest it is a local problem on a particular machine rather than a generic bug. However that maybe makes it more difficult to pin down exactly why the problem has appeared since there are so many possible variables.

    [...]

    It probably would not do any harm to clear out the system and local Cache and also discard any old Temporary files.


    Grant


    One of my PC use an AMD cpu and the other one used Intel cpu.
    I have the artifacts with AMD cpu.
    0
    Actions pour les commentaires Permalien
  • daveh_72
    Apologies for not coming back sooner. Decorating project is keeping me busy! 😊

    Another AMD processor here

    AMD Phenom II X4 960T
    Radeon HD6770 Graphics
    8GB Ram
    Windows 7

    I've not tried going back to an earlier version yet. Might give that a go and see what happens. There is nothing I can do to get rid of the artifacts in the current version. They occur during the preview generation, as far as I can see, but then "stick" to the image.

    Forget what I said in an earlier post about some green artifacts showing but not exporting. I'd accidentally turned the focus check tool on. Doh 😊
    0
    Actions pour les commentaires Permalien
  • P1337
    [quote="daveh_72" wrote:
    Another AMD processor here

    AMD Phenom II X4 960T


    Wow, nearly the same as mine. Maybe we found something here...
    0
    Actions pour les commentaires Permalien
  • SFA
    Be sure to let Phase know via your Support Cases!
    0
    Actions pour les commentaires Permalien
  • nhs
    Yeah I'm not alone, same issue many have here.
    Below my system setting:

    - Capture One PRO 11.3.1
    - Fuji X-T3 lossless compressed RAW
    - Windows 7 SP1 64-bit
    - AMD Phenom II x6 1055t
    - AMD Redeon HD ???? (believe HD6770 Graphics)
    - 16 Gb RAM
    - 512 Gb SDD

    Hopefully they fix it soon before my 30 trial days expire 😁

    Here a screenshot of mine
    https://cdn.fuji-x-forum.de/uploads/monthly_2018_11/c1_issue3.jpg.3722050fb1acb1c7327828693647614a.jpg
    0
    Actions pour les commentaires Permalien
  • P1337
    Thanks, i reported it.

    [quote="CCWH" wrote:
    One of my PC use an AMD cpu and the other one used Intel cpu.
    I have the artifacts with AMD cpu.

    CCWH, which AMD processor do you use?
    0
    Actions pour les commentaires Permalien
  • Dave R
    [quote="phil_1337" wrote:
    [quote="CCWH" wrote:
    I have 2 PC and the results were different


    Thanks for your check, that is the same observation also I made with 11.3. Artefacts on my PC, but no artefacts on my old laptop. Interesting that this problem didn't occur in 11.1! But I think I won't buy a very expensive piece of software just to only use the old version. 😊

    [quote="SFA" wrote:
    If anyone has an example file they can share for me to download I would be happy to spend a few minutes investigating.

    (Unless of course the C1 support crew have already suggest a way around the problem as a temporary solution.)


    The support didn't come up with any solution yet. Thank you very much for your willing to help! I have here two images to share: The window (already known) and an image of 2 bikes, where greenish/red artefacts are appearing on the right bike.



    Had a look at the two pictures on my Win10 Laptop
    (Win 10 Pro v1803
    Intel i7-8750H
    32Gig RAM
    NVidia GTX1060 with 6GB GDDR5 memory)

    Capture One 11.3

    I can't see any yellow artifact at all in the Window picture and all that I could see wrong on the cyclist picture was some fringing which I got rid of with a 67 setting in the Purple fringing Defringe slider.

    Dave
    0
    Actions pour les commentaires Permalien

Vous devez vous connecter pour laisser un commentaire.