Flat picture controls and raw

Discussion in 'Nikon' started by fluppeteer, Jan 7, 2015.

  1. Hi all. Happy new year.

    Here's an odd thing. I was just reading the D810 brochure, and it mentions flat picture mode. That makes perfect sense for video shooting (because you might want to reduce the dynamic range of what you're shooting so you can re-grade it), but they say:
    Flat is ideal when shooting stills in raw (NEF) or when capturing video.​
    Huh? Does "flat" do anything interesting to the raw mode? Does it affect the exposure settings in some way, or is something weirder going on? I wasn't expecting it to have any significant effect, especially since it appears to be similar to what you could achieve with a custom picture control anyway. Is this just the brochure being confused, or is there a trick I could learn here? A magic ETTR mode would be quite welcome, if that happens to be where they were going with this. (I had hopes for highlight metering doing that, but I got the impression it doesn't.)

    My D810 is still in being-lusted-after mode, so I can't just try it and find out... Can anyone educate me, please?
     
  2. What I have understood, the flat picture control can be applied afterwards too. So it does not necessarily do anything to the "raw" but interpretes the file content differently using different parameter values compared to usual picture controls like standard, vivid or custom.
    I have tested the flat setting as a starting point to some pictures even with D700, which does not support that setting in the camera. So I am not able to use flat picture control as a camera preview with D700.
    And according to my memory, flat picture control has not gotten any curves applied to the "raw". Actually what we are getting out of a camera as raw, is still more or less processed. So images via a flat picture control setting, will be least processed when you see the image.
    To me the hilight preserving exposure mode is a different thing.
     
  3. This isn't even an attempt to answer your question Andrew, just an observation.
    It seems to me that camera maker's haven't quite grasped the fact that they've also taken over the job of the film manufacturer as well. Kodak, Ilford and the rest were absolutely brilliant at supplying curves and data about their sensitive materials, and it was pretty easy and transparent to see what transfer characteristic you were going to get from a given subject brightness through to a density or print tone - provided you could get your head around all the maths.
    Now, all that transparency (no pun intended) has disappeared and we're left to find out empirically what the heck's going on inside the camera. Or by someone actually reverse engineering the machinations of a given DSLR's image processing. In fact you can't even rely on standard colour-space tone curves being followed in the JPEG output. Nikon's sRGB, for example, doing nothing like following a gamma of 2.2, even well above the crazy shadow area linear region.
    Maybe if we bombarded DSLR maker's with demands for deeper technical info things might change - it's not as if a tone curve can be subject to a patent or be a trade secret. But somehow I doubt that there's enough of a general want for such information; more's the pity.
    Edit: I definitely support the want for an auto ETTR metering mode though. Jeez, it can't be that difficult to implement. And much more predictable than flaky Matrix Colour 3D blah-de-blah-blah... which still manages to overexpose if faced with anything like a dark background or moderately contrasty subject.
     
  4. Andrew;
    You tempted me to look into this issue a bit. I have never shot in this mode because the dynamic range from this camera in the default setting is outstanding. Derrell young's excellent book says that the intention of this mode is to increase dynamic range without using HDR. I will play a bit with this and report any interesting results.
    PS. Stop lusting and buy it. It is killer.
     
  5. I think the only reason they say that is because the histogram will give you a somewhat more accurate representation of what will be clipped in the shadows and highlights.
     
  6. It's a flattened tone curve that is -- I guess -- somewhat closer to linear than "neutral"? This will yield a greater dynamic range in a full 16-bit capture over the "standard" and "portrait" profiles in that the highlights are not boosted and the shadows are not attenuated to create a "pleasing" contrast.
    For many projects, it is best to start with the closest thing to "linear" that you can get, especially if you're doing ETTR, and roll your own tone curves. Too many people spend too much time trying to "undo" the consequences of applying a default "standard" tone curve, which bunches up tones and gives false areas of oversaturation.
    The "flat" control, in comparison with the "neutral" control appears to maintain color saturation in the shadows as natural, which would enable them to withstand more processing in the end, especially if that processing involves boosting the shadows.
     
  7. With my d800 I generally shoot with the camera set to portrait with low contrast and saturation settings. Understood I can change all this in pp but its about having what I consider a good starting point for pp straight out of camera. It gives me the ability to gradually increase the contrast curve and saturation to my liking without having to undo what the camera has done to produce a "finished" photo. If I was shooting jpeg for immediate use without pp I would use higher contrast and saturation settings (perhaps standard). I think this aproach may affect exposure settings to some extent but I have not experimented much. With regards d810 flat mode I had considered this an extension of this approach, producing a flat file without the need to customise the profile, or indeed the option to produce an even flatter file. For me a low contrast, low saturation but bright image is the best starting point for pp. When I say bright, I increase the brightness in the picture control, expose ettr but retain important highlight info in each colour channel.
     
  8. set up the camera on a tripod and always shoot the same scene
    take shots in all different modes, compare histograms
     
  9. I think Olivier hits the nail: the histogram should be a bit more reliable as the histogram is based on the "embedded JPEG preview" in the NEF file.
     
  10. Thanks, everyone (and sorry about the silence, I posted before going to bed). My impression was that "flat" was deliberately designed to map a large dynamic range to the image (like a low-contrast film), with a view to being able to apply adjustments in postprocessing. Which is useful in video, where there's no "raw" mode, and useful if you can only process JPEGs and not raw files (for some reason), but if it doesn't affect the raw file content as such. I could believe it affects metering (I generally leave my camera on the defaults and auto white balance as a first guess, but fix everything in post), and I guess on-camera histogram is equivalent. I was just curious that this was a good enough justification for Nikon to claim that "flat" was especially useful to raw shooters - but maybe it is.

    Thanks, all.
     
  11. I suspect that "flat" is rather like raw movie film:
    http://petapixel.com/2015/01/06/movies-scenes-look-like-straight-camera-versus-theaters/
    It gives you a lot of latitude to modify the original without bumping up against limits.
     
  12. I notice that "flat" is an option in Capture NX-D, but grayed out for my D3200 raw files. If this is just a picture control, any idea why it's not made available? Is the D3200 NEF file already too processed for the flat control?
     
  13. When opening the CNX-D, there is an option to select "Camera compatible" or "Newest" Picture Controls. When the camera compatible is selected and when the camera does not support the Newest Picture Controls, then part of the options are grayed out.
    Newest are available though for developing phase at your computer.
     
  14. Matthew,
    You have to change "Camera Compatible" to "Latest Picture Control" and then they all show.
     
  15. Thanks to those above: CNX-D does indeed produce "flat" when reconfigured. It required a reboot, but worked then.
    Not sure what I'll actually do with it, but at least now I can check it out.
     
  16. Nick: For video, yes. If you wanted to edit a JPEG, yes. In both cases, restoring the original dynamic range will produce more posterisation than a "correct" exposure, but you're less likely to clip. I just don't expect it to have any effect on the raw file, which already has the full dynamic range captured - except for the suggestion that it might change the metering (presumably to preserve highlights). Unless, of course, someone is counting "small raw" as actually being "raw" for the purposes of this discussion!

    Interesting NX-D behaviour, not that I've ever gone near Nikon's software. :)
     
  17. I don't see that changing the picture control settings will affect the raw file as such but since changes affect the jpeg they also affect what we see on the lcd in terms of the picture's appearance and the histogram. If we use the histogram to influence our choice of exposure, the picture control settings will surely be a factor?
     
  18. Keith: Indeed. I guess it also affects the automatic metering decisions, although I'm curious to know exactly how. I'd love this to have an ETTR effect, but I'm not optimistic that it's that radical. I'll experiment when (or maybe if) I eventually get my hands on a D810. Nikon just threw me with their wording.
     
  19. Yes, the picture control setting does of course affect the histogram and it seems to me that there is a big difference between shooting RAW and JPEG in that the RAW exposure can if required be so large that the sensor starts to saturate or at least be on the edge of its linear region. However the RHS of the histogram only shows the point where the JPG image hits the endstops; it tells us nothing about sensor saturation etc. Presumably a low contrast JPEG (is that what 'flat' does? Is it lower contrast than 'neutral'?) tells us a little bit more about what the sensor is doing than does a high contrast one.
     
  20. I notice that "flat" is an option in Capture NX-D, but grayed out for my D3200 raw files. If this is just a picture control, any idea why it's not made available? Is the D3200 NEF file already too processed for the flat control?​
    Pretty sure that I was able to choose it for my D7000 raw. You need to override the default camera picture controls. Pretty sure I could choose it for my D40 too and that doesn't even have in camera picture controls!
     
  21. I don't see that changing the picture control settings will affect the raw file as such but since changes affect the jpeg they also affect what we see on the lcd in terms of the picture's appearance and the histogram. If we use the histogram to influence our choice of exposure, the picture control settings will surely be a factor?​
    I think this is what is going on. Flat has lower contrast, so the histogram shows a wider dynamic range window. I guess this makes the meter more useful for raw, as the invisible "highlights headroom" shrinks. If you turn of autoWB, you get closer to the UniWB approach. And the shadows get slightly lifted too. But preview generally looks crappier.
    I haven't checked out this new flat PC. One could edit the PC and set lower contrast before though, but the issue was that the dynamic range of the processing was narrower than the one of the sensor already on the D90. So one could get clipping on both edges of the histogram in certain limit cases where the raw data was in fact not blown by Rawnalyze. VNX would blow the same in a single conversion (but one could do multiple with different exposure slider positions and HDR blend), but LR and DxO OP were able to process this nicely.
    I was hoping the new HWM would do Uni-WB-like auto-ETTR for raw too, but I hear from several reputable sources that this is unfortunately not the case (that it sometimes exposes lower than desired).
    DPR claims that ADL no longer adjusts raw data exposure when used in conjunction with HWM. I'd guess this should make better looking previews while throwing the histogram less off than before (with my D90 on MM, ADL helps save the highlights sometimes if I don't have time to adjust EC for harsh light; on the other hand, one gets grossly underexposed raw data when tuning EC to ETTR using the camera's histogram).
    There is however a rumour about an upcoming Nikon firmware update program supposedly bringing raw histograms. This is a sign of a new wind; if they see the need for raw histograms, they might see one for auto-exposure based on them, although there are a few practical issues (what to do with the sun in the frame; small specular highlights and the relatively low resolution of the exposure sensor). And eventually they could also consider how to render better previews in harsh light based on such exposure.
     

Share This Page

1111