Zum Hauptinhalt gehen

⚠️ Please note that this topic or post has been archived. The information contained here may no longer be accurate or up-to-date. ⚠️

Pink hue to conversions from Canon S100

Kommentare

15 Kommentare

  • SFA
    Tim,

    I would start by checking the options for White Balance adjustment.

    Find somewhere in the image to represent something close to mid-grey (parts of the bottom wall may offer something) or go for a full white or full black area - or as near as you can get. Then see what sort of different, if any, appears.

    On a positive note the foliage in the converted file is much better than the jpg, as one would expect.

    BTW although I don't have an S100 I do have an S90 and I don't recall seeing this sort of shift although one would expect differences between the images since they are being processed differently.

    That said I'm not at all sure that I would always be convinced that the camera internal processing has picked the absolutely correct WB, indeed there may be no such thing since different parts of an image will be subject to different colour temperature interpretation. Of course this can have an effect on all parts of the subject at different times of the day or in different ambient light situations.

    The S90 has an option to set a manual white balance or to adjust the default settings for of the defaults. Even AWB.

    I assume the S100 has the same. It may be worth quickly checking that nothing has slipped in there to influence the settings provided by the camera for the jpg.

    If it is a genuine overall colour tint rather then a WB issue then the colour balance tool should provide a solution for you.

    You could also compare the different options for the "Curve" in the Base Characteristics tool to see what sort of a difference that would make and whether a different curve provides a better starting point.

    HTH.



    Grant
    0
  • Tim Bennett
    Grant
    Many thanks for your prompt reply.
    I have been playing around with C1 prior to my original post and picking a gray (or black )area whilst going some way to resolving the problem does also lead to a fairly washed out effect.I can then build up the saturation etc of individual colours to the desired effect but it's fairly long winded
    WRG to White Balance setting on the camera,I'm pretty sure this has not been altered and in any event would it effect the RAW file if it had?
    I hear what you say about the foliage but overall the look of the jpg as produced by the camera seems to me to be more "realistic" than the C1 conversion.
    Cheers
    Tim
    0
  • SFA
    [quote="TimB64" wrote:
    Grant
    Many thanks for your prompt reply.
    I have been playing around with C1 prior to my original post and picking a gray (or black )area whilst going some way to resolving the problem does also lead to a fairly washed out effect.I can then build up the saturation etc of individual colours to the desired effect but it's fairly long winded
    WRG to White Balance setting on the camera,I'm pretty sure this has not been altered and in any event would it effect the RAW file if it had?
    I hear what you say about the foliage but overall the look of the jpg as produced by the camera seems to me to be more "realistic" than the C1 conversion.
    Cheers
    Tim


    Hi Tim,

    Any in camera WB adjustment would only affect the jpg permanently, although that would include the jpg preview file embedded in the RAW. Once you start editing the RAW file in C1 you would be seeing C1's interpretation rather than the camera's jpg preview/thumbnail/whatever-name-you-want-to-call-it-by of an unmolested RAW file.

    So my thinking was that the original jpg may not be entirely neutral to event the Canon engineer's usualy interpretation of a scene (which tends towards vivid, especially orange/reds because that is what people seem to prefer or so we are told). Nothing wrong with that, everything is down to an interpretation here and one of the benefits of RAW is that it is easier to personalise the interpretation and retain quality, as you already realise.

    Obviously I don't know the subject of the photo nor the nature and colouring effect of the light at the time so anything I mention is entirely speculative. However, ignoring the fully lit struture for a moment, to my eyes the jpg image you posted (and yes this is VERY subjective and there are more than a few issues involved with assessing colour rendition of a compressed image after transmission through a series of internet technologies!) for the out of camera results looks a touch yellow to me whilst the jpg from the C1 edit does indeed look a touch pink. However if you look at the wall at the bottom of the image, through my browser and whatever settings windows 7 has managed to preserve reasonably accurately for the screen to display, the greys in the C1 edit seem more natural than the greys in the jpg image. So the difference may ab a very fine line indeed.

    Of course, as mentioned, I have no idea what the subject should look like from direct visual observation.

    It sound like the WB setting is a good starting point, maybe with a slight hue adjustment - hue can be very sensitive in my experience. I'm not sure why WB change should introduce a significant desaturation effect but I would guess that adding a little more saturation in the exposure tool ought to be a first tweak and then look at the individual colour channels.

    You could probably use the Color Readout feature to work out which colour adjustment(s) may be required by comparing the same points (approximately) on the jpg version and the C1 RAW conversion. Whilst the absolute values of the chosen points may vary the relative differences could give in indication of what to adjust to get the colour rendition you prefer.

    How you apply the adjustment is another matter - Colour Balance, Colour editor or a Curve adjustment on a particular colour channel would seem the obvious options to consider once WB is closer to your liking but there may be more to it than that - there are so many ways to tweak things it's not that easy offer a suggestion with any degree of certainty and in any case I tend to try to avoid detailed colour adjustments mostly so there must be others here who can offer much more specific advice and may have a less obvious but more effective approach for your needs.

    It's a tricky one. My guess is that the best balance for your purposes is somewhere between the two interpretations. Finding that point may be an interesting exercise. If you want to make the RAW conversion the same as the jpg (or at least a lot closer to it given the there will likely be more data in play for the RAW file than the already compressed camera jpg) then I think you need to get that feel for what the differences are using the Color Readout option as a starting point before seeking the best approach for that specific image that will adjust the C1 interpretation to the out of camera jpg result. Maybe one more option would be to try to get the jpg to look like the C1 result and then invert that adjustment for the RAW file? I'm not sure that would work but it might provide some useful insights.

    HTH.


    Grant
    0
  • Tim Bennett
    Grant,
    thanks again-there's a lot to digest and try there!
    On the purely subjective point anyone familiar with the subject-Windsor Castle -will confirm that the "true" colour of the stone is a mellow yellow and looking at the large area around the vertical arrow slit at the bottom of the pic gives the largest clear area.
    Anyone else have any views?
    Tim
    0
  • SFA
    [quote="TimB64" wrote:
    Grant,
    thanks again-there's a lot to digest and try there!
    On the purely subjective point anyone familiar with the subject-Windsor Castle -will confirm that the "true" colour of the stone is a mellow yellow and looking at the large area around the vertical arrow slit at the bottom of the pic gives the largest clear area.
    Anyone else have any views?
    Tim


    Tim,

    Hmm, Windsor Castle. I suppose I should have recognised it although looking back it may be 20 years since the last time I passed through Windsor - and you have an unusual angle too, not a typical Windsor shot.

    I ran some test shots on the S90 today - a not very strong sunset was about the best I could do. I'm still in 'what have I got here mode - the shots are totally different to your subject matter and composition but ...

    Yep, the OOC jpg suggests quite a yellow sunset whereas the RAW interpretation has more orangeness about it.

    The comparison idea is sort of valid except that the selected colour read out points will likely not be consistent.

    The S90 does not have any official auto support of the lens correction - the shots I took happen to be at the wide end so correction would be quite significant for the jpg and so it proves. C1 manual lens correction gives a perfectly acceptable result for my purpose BUT the selected colour readout points shared by the images do not align exactly between the RAW with manual adjustment and the jpg with adjustment applied in camera. That makes the colour analysis trickier.

    I tried applying the S100 lens profile to the S90 and that worked well enough for correction BUT produced a smaller overall image size and therefore once again the alignment of the colour assessment points was a little off making direct and accurate comparisons more difficult.

    I'm still looking at it but first impression are that performing the checks may not be as straightforward as one would like. However they may be accurate enough, given the right image, to form an opinion and define a corrective strategy.

    I'm have not yet looked into that part of the process. More later, hopefully. However a quick Google for images of the Castle produces a lot of results with a variety of colours! Which is interesting although not entirely unexpected.


    Grant
    0
  • Tim Bennett
    Re the composition,the site I used to upload the pic has cut it in half 🙄
    0
  • Keith Reeder
    It hasn't, Tim - this forum truncates larger images, but if you post the URL of the posted image into the address field of the browser (or drag the image to the address field), the whole picture is there.

    We've had quite a lot of discussion on here - in the past and again recently - about Capture One's profiles being biased towards red/pink/orange (depending on the camera and what's in the picture): I suspect that what you're reporting here is no more or less than "The Capture One Look".

    And for what it's worth, I prefer the Capture One rendition here - these aren't colour-accurate profiles, and the bias towards warmth that Capture One gives is a selling-point for many.

    My Canon 7D's profile is overly orange, and too saturated for my tastes - but it's easy enough to pick an alternative profile from the drop-down (I use the 5D Mk II profile) which is closer to what you want to see.

    Once you've found a profile that suits, it's a simple matter to navigate to the profiles folder (Program Files > Phase One > Capture One > Color Profiles > DSLR in Windows), back up the original "offending" profile, and copy/rename the preferred one, which will then be called as your camera's default.
    0
  • Tim Bennett
    Many thanks for that Keith.
    I hadn't realised there was such a distinct Capture One style and that it could vary from camera to camera.I suppose I was comparing it to results from my main camera,a Pentax K-R ,which I am perfect happy with.I will play with different profiles for the S100 and see how it goes.
    Tim
    0
  • SFA
    Tim,

    I played around a little more with my S90 files.

    The few I took yesterday were mostly just a location assessment. The light was going and I was going to silhouette type shots so not really comparative to your castle shot.

    However the C1 processed RAW files are clearly warmer than the OOC jpgs (created by letting the camera do what it wanted to do.)

    On the other hand most of the shots have a lot of under exposed foreground (disposable to the image idea I was considering) that is rendered monochrome (almost) in the jpg but still carries a more correct green colour in the converted RAW.

    I suspect that neither rendition is entirely accurate (as per Keith's commentary previously).

    I have yet to find a single 'fix' that would make the RAW conversion match the jpg fairly closely. Nor a more complex multi-tool fix that gets everything right, though I can get parts of the system to better match. (That's an exercise as an experiment since I don't necessarily see the jpg as being a perfect colour source either.)

    I'll have to go back to some older stuff shot in better light before playing further. Yesterday's shots most likely are too compromised by the conditions to be of much use for the colour assessment.



    Grant
    0
  • Keith Reeder
    Personally I don't even try to match my cameras' default output - I don't consider them to be inherently "correct", and will let my eye dictate the end result every time.

    Saves a lot of frustration!

    😉
    0
  • Permanently deleted user
    Hi there

    Oh dear! Back to the bane of a photographer's life - how do you get the colours of different outputs of the same image to look similar and 'right'. Hopefully the following contribution will be of use.

    Firstly, we have to accept that each RAW converter will embed in the TIF output its own 'take' on the image that the RAW file represents. This means noise, sharpness and tonal values including colour will contain an echo of the RAW converter's algorithms. JPEG outputs from a camera will also have processing idiosyncrasies built in, whether the in-camera settings are set on their default values or tweaked by the user. So it is hardly surprising that RAW conversions and JPEG outputs of the same image will look different. [The only exception I have found is when using Canon DPP software that allows me to set the RAW processing parameter values exactly the same as the in-camera settings.]

    Moreover, such mismatches will still occur if the computer screen colour calibration is slightly off - because the screen settings are a constant whereas it is the image files that vary. So how can we decide which output is more 'correct'? I have stumbled across a procedure that seems to do the trick, and it starts with a black and white conversion of the colour image which retains the colour setting options. So, select whichever colour image you prefer and follow the following recipe.

    Print this B&W image out on neutral white photo paper as a proof (A6 size is quite adequate) and inspect it under a daylight bulb. Compare the tones with what is on the screen. If they are very different you need to adjust the screen colour settings. With my screen I can adjust RGB channels separately, and I can use the screen driver software to tweak the colour temperature. Note these settings for future reference, and save them. In effect we have slightly adjusted the colour calibration of the screen, and we may need to reset this in the future.

    It is likely that the colour tone/cast of the screen image will still look a little different from the printout. You can correct this in software (e.g. CS5, using 'Hue/Saturation') using minimal adjustments to get a good match (often just tweaking the master hue will suffice), and note this setting. In effect we have tweaked the B&W image in a known and repeatable manner to replicate the colour tone of the printout.

    Now we go back to the colour image and apply the same CS5 tweak, and print it out (using the same paper and printer settings as for the B&W image).

    Voila! The printout should be a good facsimile of what is on the screen, and both images should look realistic and 'correct'.

    If you wish, repeat the above for the second colour image and compare the outputs.

    Good luck!

    Peter
    0
  • SFA
    Keith,

    I agree about treating things as what they are but in this case I think Tim has made a clear and interesting observation and of course the challenge, having concluded the jpg is a closer match to what an eye/brain combination might think it sees, is to get the RAW conversion somewhere close.

    Peter,

    That's an interesting approach and sounds very pragmatic. Not sure if Tim is going for a print or a screen rendition. It might be easier to work with the pink RAW and go directly to a more correct print than to go through the intermediate screen correction step! 😉

    Tim,

    Just looking at a few of my S90 files and pondering what I can see in terms of the RAW interpretation vs the jpg.

    The Histograms of the converted but unedited RAW are reasonably similar to the jpg in terms of levels and curves appearance. Not identical but mostly close enough that any changes would likely have little evident effect on a coloour by colour basis except for very specific points in the range.

    However to take one example, for WB the jpg show a temp of 5000 and a hue adjustment of 0.

    The straight, unedited, RAW conversion shows a temp of 4753 and a hue adjustment of -10.3. Compared to the (sRGB) jpg the blues (sky with clouds) look reasonably close but a sunset is notably more orange in places rather than yellowish as per the jpg.

    If I change the RAW edit settings for WB to around 4450 (+/-50 or whatever nuance you like) and the hue to -12.6 (+/- .5 or so) AND change the Base Characteristics to make the Curve "Linear Response" the colours in the troublesome range start to look very similar indeed. Close enough, in my sample file, to make it all but identical to the eye in the troublesome colour range.

    It might be worth trying similar adjustment value differences on some of your files Tim, although the actual adjustments required for your screen set up and equipment may well differ from mine. I would be interested to know if that offers you anything.

    I doubt whether these will prove to be definitive adjustment figures but it would be interesting to find out whether the idea has any general merit.

    HTH.


    Grant
    0
  • Keith Reeder
    [quote="SFA" wrote:
    Keith,

    I agree about treating things as what they are but in this case I think Tim has made a clear and interesting observation and of course the challenge, having concluded the jpg is a closer match to what an eye/brain combination might think it sees, is to get the RAW conversion somewhere close.

    Not sure about that, Grant.

    If the default jpeg could be considered true(ish)-to-life, then yes, it's a good starting point: but why does it actually matter? The photographer has seen the scene, and knows how the colours should be, so unless the in-camera jpeg is a necessary point of reference (and I've never found that need) then I'd argue that it isn't relevant.

    Furthermore, most cameras have multiple-choice "picture styles" which can be applied to jpegs, and they can change the output considerably. Which one's the right one for an accurate starting point?

    I know that my 7D has (among others), "Standard", "Faithful" and "Neutral" styles: none are characterised as accurate, and in fact Canon assumes that image using "Faithful" and "Neutral" will be subject to post processing (which logically includes white/colour balance where necessary) so which is the best reference rendition?

    Won't "best" change anyway, depending on the subject and the light hitting it?

    Naaah, to my mind it's very debatable that trying to match in-camera rendering is a useful way to spend time: Tim obviously knows what colour the castle should be, and that's the issue - discussion about matching the in-camera jpeg is incidental and a distraction.

    As I've suggested, I "get" the issue, as I've long been dissatisfied with the default Canon 7D profile, but a good thing about Capture One is that we can try every other profile that the software comes with in order to find one that does the job...
    0
  • Tim Bennett
    Thanks to all who have had a input to my OP.

    The very reason for posting was that the jpg out of the camera far closer matched my perception/memory of the scene than the unadulterated RAW file,it was not really a question of trying to slavishly recreate it.

    Grant- prior to posting on this forum the best "solution" I could come up with is as mentioned in your last reply I.e.playing with temp and hue adjustments and then fine tuning individual colours but as you've no doubt found that's a pretty hit and miss/long winded affair.

    It may be that I'll reserve using RAW files from the S100 to cases where there are critical exposure difficulties etc which seems a shame.

    Cheers all,

    Tim
    0
  • SFA
    [quote="TimB64" wrote:
    Thanks to all who have had a input to my OP.

    The very reason for posting was that the jpg out of the camera far closer matched my perception/memory of the scene than the unadulterated RAW file,it was not really a question of trying to slavishly recreate it.

    Grant- prior to posting on this forum the best "solution" I could come up with is as mentioned in your last reply I.e.playing with temp and hue adjustments and then fine tuning individual colours but as you've no doubt found that's a pretty hit and miss/long winded affair.

    It may be that I'll reserve using RAW files from the S100 to cases where there are critical exposure difficulties etc which seems a shame.

    Cheers all,

    Tim



    Tim,

    I think if you can find a combination that seems to work consistently well enough (tricky for outdoor stuff with constantly varying light unless you are very careful or pedantic about taking colour sampling shots even with a compact) it would be less of an issue to get closer on a consistent basis. One would just save the 'close' result as a preset and make it a default for the camera.

    That begs the question of how close the result has to be to be acceptable to the creator.

    I have been wondering just how much lens processing goes on in the camera and whether it includes colour adjustments along with geometry.

    If there is colour patch correction in place than the chances of getting things right using manual sliders is is surely rather slight.


    Grant
    0

Post ist für Kommentare geschlossen.