Skip to main content

Two feature request for skin tones

Comments

14 comments

  • Marcin Mrzygłocki
    Top Commenter

    I'm not sure if bundling requests is the easiest way to put them forward, because developers might decide to apply one but not the other - what the status could be then? "Implemented" or "Not currently planned"?

    Apart from that, you have my vote.

    1
  • SFA
    Top Commenter

    As I recall the Red centric skin tone adjustments were developed, across the industry, because digital technology tended towards excessive red toning in some situations. The results were more consistently noticeable in some skin tones than others although the red effects would exist across all images in certain situations - depending on how the White Balance in the camera was set up.

    There is nothing, so far as I am aware, that would prevent specific presets, styles (and therefore style brushes) to be created for any number of additional finely tuned skin tone of any colour, even for one-off shoots.

    However, I think that the original red adjustment was intended to correct consistent problems introduced by technology rather than offer some final cosmetic enhancement  - something that is left to the user to complete even after the basic red balance adjustment.

    1
  • Underground Lightning

    Marcin Mrzygłocki  Sorry for bundling both suggestions in one request and thank you for the vote.  This is my first time submitting a C1 request, and I will keep your pointer in mind for future submissions.

     

    SFA I was not aware of the digital sensor lineage of red-skin correction, so I unknowingly assumed it to be a cosmetic feature.  Thanks for sharing this.

     

    I manually correct yellow or gray skin tones, and find the vectorscope skin line immensely useful when doing so in video editors.  Perhaps a Smart Style tool that grabs selected (or AI selected) face hues and nudges them towards vectorscope skin-line might be a more generalized tool to speed up workflows.  Would such a smart style be useful in your opinion?

     

     

    2
  • SFA
    Top Commenter

    Styles are useful for anything that one uses regularly long term or for a single shoot use if a lot of images are involved.

    For 2 images with the same needs a style would be useful but copy and paste may be enough.

    Some of the original Styles that are or were provided by C1 (the processing engine has change once or twice and styles may have been adapted to the new engine or dropped as no longer of value)  are related to subtle change in colour hue related to skin tone, However the "Skin Tone" tool in the colour Editor, while intended to provide the subtle changes that skins often require, is not actually limited to skin tones at all and can be used for any colour changes.

    I know little about the workings of  Vectorscopes but the idea of AI seems reasonable. However, a set of presets or styles may well offer the same opportunities for editing still images if the need for absolute consistency exists. I'm not totally convinced that the requirement for stills is the same as the requirement for video where I can imaging that highly variable rendition of skin tones for a series of frames might produce an extremely bad look on screen.

    The challenge with AI selection is that the "rules", once refined across billions of samples, will end up with everything looking the same. Not just for a selection of images, but for all images because everyone is using AI.

    However, in theory your suggestion is exactly what the capabilities of AI should be able to deliver and I doubt the average viewer of the video or still images would notice or even care much that, within a certain number of common "designs" for humans (or whatever) the results all looked very similar. 

    We already see the results of "best design" making a lot of products look the same, airplanes and cars being some of the most visible partly because of aerodynamic efficiencies dictating certain values as optimal and physics offering a very limited number of different ways to deliver what is required.

    Something very similar could happen with how people look in digitally enhanced images. It might be even more likely if the images were fully AI creations.

    Whether that is a good thing or a bad thing is a matter of opinion. I think it is interesting but not something to be welcomed without proper consideration.

    Artistically speaking one might suggest there is no real insistence on accurate skin tones. T+rather that they are influenced by the surrounding light (when outside of a studio setting where the light is, potentially, more controllable) and so should be aligned with whatever ambient light is delivering OR whatever settings have been used to modify the look of the ambient light.

    AI can do that as well, we are told. But that adds another level of context to the discussion here.

    1
  • BeO
    Top Commenter

    Skintone line

    Why don't you just figure out where your best suited skintone line is and save it as a presets for the Skin Tone Color Editor?

    The point and the slice gives you the same information you request as a line, doesn't it?

    The real benefit of a vectorscope is that it shows you your image data as points or clusters of points eg. as in darktable:

    The above shows you the points from a complete image, but you can also draw a rectangle (a special color picker mode in darktable) in your image which not only shows you color values min, max and average, but also shows you the points from that rectangle in the vectorscope (and histogram).

    Hence you can pick an area from a face, then see where the colors lie in the vectorscope and shift the hue towards where you think it belongs. A skintone line is not available though in darktable either.

    The usefulness of such a skintone line is not a common agreement in the video world, at least what I have read, but I cannot really judge this statement.

    Anyways, without the image data, the reference line is not very useful I think.

    SFA, maybe you remember a couple of days ago I pasted the above vectorscope in another thread, and since then I think about a feature request which uses the Advanced Color Editor as the tool to depict the image points, essentially merging the functionalities of the Color Editor with those from a vectorscope.

    You could not only see your image data but directly manipulate them here.

    [Image: Combined Color Editor and vectorscope]

     

     

    1
  • SFA
    Top Commenter

    BeO,

    Yes I remember and I thought of the Davinci vectorscopes I recall quickly looking at a few months ago before deciding I would wait until my video project was underway to try to work out how they might be useful.

    ( I suspect they will not, given the project is the digitization of some ancient 8mm cine film. However, I am always delighted to be surprised by the discovery of something especially useful yet unexpected.)

    My inclination would be to work with the image via "View Selected colour range". However, I fully appreciate that sometimes that can lead to subtle changes in unexpected places that the vectorscope display concept might more readily present.

    0
  • BeO
    Top Commenter

    Hi SFA,

    The checkbox "View Selected colour range" shows you the source colors and reflects the adjustments in the Color Editor within in the image. Very useful indeed, I don't want to miss that.

    What it not does, it does not give you any reference. The only reference you have is the neutral border color in the viewer (very useful too) and the other colors in other areas in your image. But these are all relative references to either neutral or colors, and the assessment can only be done in your brain, with all the shortcomings of color perception like adaption, eye and brain strain etc.

    A vectorscope shows you the image colors always in the same location of the color wheel, so identical colors in all your images at the 5 o'clock position are always at that 5 o'clock. 'Underground lightning' (the requester here) asks for a reference line for skin tones. Your brain can easily recognize if the color clouds are at the same spatial reference position where you want them to be, or the skin near your reference line. That is the idea.

    Assume a landscape photographer who always (or for the images from that day and location) wants to have his sunlit foliage depicted by very similar green-yellows. Or a portrait photographer who wants Lucy's skin always in the same hue range. Etc. Which tool in C1 can give you that?

     

    Another use case is if you want to edit your image towards a color harmony, e.g. analoguous, complementary, triad etc.

    https://www.institute-of-photography.com/colour-harmony-in-photography/
    http://www.naturephotographers.net/articles1112/ab1112-1.html
    https://iso.500px.com/take-photos-with-color-wheel/

    In darktable you can overlay a color harmony scheme to help with that (the white segments)

    If I would like to use the analogous complementary color sheme for this image I could try to move the greens and magentas into or towards the respective segments.

    A vectorscope in darktable can show the whole images pixels, or you can draw a rectangle to select areas.

    Color casts

    Here I selected an area (from a different image) which is supposed to be white. Do I have a color cast? Or is it actually a red-pinkish white and therefore correct, or to my liking?:

     

    I selected a big region of the midday blue sky. Is this the midday blue I want or do I need to adjust my WB, or at least the sky.

     

    For your video digitization project I don't know if a vectorscope helps you, I think it depends on what you want to achieve. If you want to set a WB once and then let the film do what it did at that time, documenting the film properties  and being authentic about the lighting conditions, and film degradation over time, then it is maybe of limited help. If you want to check and eventually slightly correct colors then it will probably help, that's what it is made for.

    1
  • BeO
    Top Commenter

    BeO wrote

    Which tool in C1 can give you that?

    Granted, you can pick a pixel in the Advance Color Editor with the picker and it shows you that one pixel in the color wheel, but this is junior stuff compared to a vectorscope, for the use cases in mind.

     

    0
  • SFA
    Top Commenter

    Very interesting and helpful. Thanks BeO.

    My cine to video conversion, as you say, may not benefit much from any colour management. The whole project was my Father's and starts about 70 years ago, covering about 20 years. Various film stock used, I suspect. All spliced together onto 4 large reels. And some commentary on tape.

    I went through the exercise of creating a digital version almost 20 years ago after a professional Audio Visual company converted the cine to digital, without the sound files. The video fioes were, I suspect, created by projecting the cine film onto a white surface and video recording it. In some parts that worked adequately, in others not so well. I suspect I can get a better result  - but who knows. The starting point will be cleaning the film. 

    I have used an old, short film for some tests and they looked promising BUT it became obvious that cleaning would be a necessity and, of course, the source of the reel is different to the source for the main project so as yet I still don't know by how much, if any, the results can be improved.

    I thought I had lost my original editing work to a disk failure but eventually discovered a copy -  just after I bought the film digitizer machine. So my project is running almost 2 years late, mainly because the level of priority decreased. The problem is the existing file is huge so I would like to make it smaller and of improved quality. Getting the soundtrack commentary to align well with the video may be the biggest challenge - as it was back in the day with a Cine Projector and a reel-to-reel tape machine.

    Nuances of colour matching may not be a top priority for long!

    I'll read the links you kindly provided before getting too deeply into the video editing.

    Many thanks.

    0
  • BeO
    Top Commenter

    I see, you have quite a few other challenges for your winter project...:-)

    0
  • SFA
    Top Commenter

    Indeed.

    The first one was being unable to find a full-size spare reel for the film.

    I should have one somewhere. I remember seeing it. But not for some time and a few movements of "stuff".

    Eventually, I bought another one. But then what we called "Summer" appeared.

    I need the spare reel to be able effectively clean the film before attempting to scan it. Both for, hopefully, improved results and to prevent any dust and dirt building up on the scanner aperture and all the stuff that transports the film. 

    Meanwhile, going completely off topic, my wife would like to get a print or digital version of a 35mm negative she took 30+ years ago. No problem, but my scanner has been boxed away for some years and I doubt it has drivers for Win10. My lens adapter, would work fitted to a digital camera but it really needs a mounted file - like slides. I could cut up the film strips but would prefer not to do so.

    The negatives may not have great remaining colour rendition. I'm not sure yet.

    So the quick and dirty experiment is to use a Tablet to get the brightest and lightest screen displayed, lay the film strip on it and take some macro shots with my phone camera. Next step is to move them to C1 and see what happens.

    0
  • BeO
    Top Commenter

    To bring this on topic again, the old negative probably shows a human subject with skin, right? :-)

    Your phone needs to capture raw, I think, because a negative to positive conversion from a jpg is probably going to look awful. If she is up for a joke, make a small print from the negative as-is :-)

    Btw, I assume you have other negatives too and your scanner is likely not going to work anymore, so consider upvoting https://support.captureone.com/hc/en-us/community/posts/360009560077-Adding-Analog-Negative-FILM-Conversion-Features :-)

    0
  • SFA
    Top Commenter

    Well, the film strips have one image with people but the other shots, the ones of interest in this case, are just pictures of a brick wall.

    Before I saw the results of my "not quite a scan" I had the impression that there was some sort of memorial plaque on the wall but that is not the case. Basically, the main subject is a single brick with someone's initials carved in it. Apparently, this is (or was 20 +  years ago) in the wall of a house built by my wife's Grandfather some decades before. She wants to share the image with a cousin who shares the Grandfather and interest in the family history.

    So, using a tablet as a light box and the Macro lens of my phone together with C1 to invert the colours, I now have some idea of the contents.

    The phone set to Macro mode, as far as I recall, only produces something like 2mpx jpeg files. The lighting needs some diffusion to prevent the somewhat "granular" effect of the screen panel of the tablet. But considering this was a very quick experiment that I did not expect to deliver a great result, it's not bad even at a first attempt. At least I can see the contents of the negatives. It didn't take long to do it.

    I did use the existing method of inverting colours and adjusting in C1. Clearly the requirements I have are quite basic since, for the test, I'm not looking for top quality. So what is available works quite well.

    I think the challenge for the Inversion option request may have some less than obvious issues to deal with. It hase been around a very long time.

    Back in the distant past I remember seeing a  post on a blog that mentioned (from memory) the possibility of inversion for colour negatives but shortly later disappeared.

    At around the same time the dedicated CH version of C1, created in conjunction with a third-party partner and mainly or totally intended for use as a turnkey solution for Museums and archivists using Phase One backs, seemed to include a logical UI tool that permitted Negative inversion without the need to remember to make everything work in reverse.

    Since that tool, assuming that is exists and does what I think it does, has not been moved to the mainstream product in the past decade (maybe more) I wonder of it's development was mostly the work of the partner and if the partner holds the rights to it.

    Realistically, in all of the scanning I have previously undertaken, negatives and slides at 35mm and 120 sizes, colour and B&W, the quality achievable from old film is limited by the level of grain in the film and the longevity of the chemicals used for the light sensitive surface. In most cases both are quite poor compared to digital competence . Move up to large format and 8x10 and things change somewhat, but that must be a very limited user base size. Either way the question of accuracy and precision available from the source data of a scanned image mat not really give a RAW converter application much, if any, advantage that users would benefit from in a way that would attract new C1 users.

    Even if we accept that there is a resurgence in interest in film cameras right now I doubt the demand for negative inversion capability is critical functionality for the masses right now.

    Should it be the case that others hold the rights to the concept as might be offered through the C1 UI , developing an alternative version and using a different UI for it might not make sense.

    Of course this is pure speculation on my part. 

    But the bottom line is that I doubt I have any negatives I could scan that could benefit significantly from a negative to positive conversion at a level that would justify anything more a quick colour flip. I think other C1 users may have a very different level of quality functionality that might be more challenging to deliver at the level they might feel to be necessary.

    So if it's available and can be added easily, then great, I'll take it.

    If it needs to be developed and maintained as a NEW function, then it might not be so interesting to me because I don't think I have any negatives that would be able to take proper advantage of C1-level quality development editing.

     

    0
  • Francesco Provino

    I'm definitely interested in the vector scope feature and I think BeO nailed it.

    0

Please sign in to leave a comment.