Personal View site logo
BlackMagic Pocket Cinema Camera 4K
  • 497 Replies sorted by
  • I'm sure there are cases where perfect re-creation of the colors that were perceived at the time a thing was shot are useful, but short of walking around with a densitometer and sampling everything, that is pretty unrealistic. Plus colors in the real world are made up of more than 256 levels of red, green, and blue and yet a lot of us manage to get by with 8-bit codecs.

    I'm not really sure what the problem is if (in grading), midtones are introduced/restored that weren't present in the original lossy image.

  • Also there is no objective way to determine the colors that are actually perceived by each one of us. Information in the file corresponds to light emitted by the display. At any point in the process from lens to our eyes via all the technology involved, information and subsequently perceived color may be altered with a multitude of means. But generally if you don't get the shot, it costs time, money or both to recreate what is missing. I am really fed up with 8 bit 420, it seems broken, 4k of course becomes 422 at hd. But it seems muddy, maybe some kind of similar processing is going on with the Bmpcc, prorez hq doesn't seem as sharp and punchy as it can be.

  • @libertas

    Also there is no objective way to determine the colors that are actually perceived by each one of us.

    This is incorrect. Color science actually started from such experiments and advanced extremely.

  • 4k downsampled to 1080p seems "muddy" to you? Maybe you're using the wrong software to downscale - with a good lens and decent bitrate, downscaled 4k tends to be razor sharp for me. "Punchy" is subjective, but by adjusting curves/contrast/saturation, you should be able to make just about any image at any resolution "punchy" enough to suit your tastes (possibly with the trade-off of banding and/or blown highlights/crushed shadows).

  • @markr041

    "Loading a REC202 LUT does not magically increase the REC709 color space."

    Yeah, no kidding... in Film Mode aka LOG I don't believe it's clamping your available color Gamut to REC709 outputs...

    Probably the standard video output setting is clamped to REC709 Gamut...

    If there is information anywhere about the Film Gamma/Gamut being clamped to REC709, let us know - otherwise I think it defaults to WideGamut without clamping... but I could be wrong.

    I totally know about Color Gamuts and Gamma being two different things... so, don't worry about me, I'm not your average LUTard . Just saying if you wanted to clamp your unmitigated signal into a BT2020 Clamped Gamut you could probably do that easily enough.

    RAW sensor data should definitely not be clamped unless you clamp it or set your project gamut in resolve

  • From my experience with Blackmagic cameras “Film” gamma doesn’t clip anything compared to RAW footage. The only difference few years ago was that Prores footage defaulted to 422 color space, while RAW was 444, but I could never see any practical advantage in it (I’m not that much into SFX).

    In fact when you develop RAW in DaVinci Resolve you have same 2 settings (Film and Video) so after extensive testing I started to shoot with Prores “Film” gamma.

    As for HDR and Rec2020: it’s just a way to map very bright values on your final image. It doesn’t have anything to do with a recording format, in fact you can make HDR cartoon or HDR game and not use any camera at all. Of course the wider the DR and gamut of your camera the more logical it is to use Rec2020 for final output, so BMPCC makes perfect sense to use it. In the end it what’s your client asks for, and if you are gonna output in HD it’s 99% gonna be Rec709.

  • @bannedindv I downloaded an original video clip from the camera. The metadata said it was REC709 color. I do not know whether it was shot in "film" mode, but it sure was flat and lacked color (which is good).

    If we had an original "film" mode video clip we can see right away what the color gamut is from the metadata. I have not seen anywhere Blackmagic indicating it has a special wide color gamut, but I could have missed it. Nor do I see a camera menu item for color gamut type (but I could have missed it). There has been no mention of HDR in any of the PR about the camera (and I do not mean HLG).

    I am just asking the question about whether there is wide gamut color or not from the camera. If so, I am sure DaVinci Resolve Studio will have the right transform to REC2020 color (the correct math). If not, then it is not possible to create HDR video taking advantage of its wider true color gamut spec using this new camera. Right now, Resolve Studio has the transforms for Sony SGamut to REC2020 and Panasonic V-gamut to REC2020, so I know with Sony and Panasonic cameras you can produce HDR videos with more true colors than REC709 (whether or not it provides the full REC2020 gamut and ignoring bit depth).

    Of course you can take advantage of the extended dynamic range shooting RAW or shooting in film mode - let us not confuse the extra dynamic range and different luminance mapping of HDR with the issue of number of colors (or number of color gradations - bits). Without colors beyond the REC709 gamut recognized by the camera sensor or recorded, one is missing the other advantage of HDR. And thus it is NOT HDR.

  • @markr041

    Every damn single camera on the market already records beyond Rec709. Rec709 is just 5 stops of latitude, I can’t even imagine someone will shoot anything with such a shiity DR in our days.

  • @Lohmatij Oy! Here we go again (talk about damn).

    I am talking about REC709 color gamut. Not DR range. I do not understand how anyone can read my above posts and still confuse DR with color gamut. I get that one code word for both DR and color gamut (REC709) is confusing, but still...And yes, gamut and gamma sound and look similar too, so I could see why there is confusion, but let's get back to the issue:

    Sony offers Sgamut, Panasonic offers V-gamut, Canon has C-gamut. These are all color gamuts that contain colors beyond the REC709 color gamut (get it, color gamut, not DR). So, what is the equivalent extended color gamut for Blackmagic cameras? Is there one? what is it? where's the spec? Every original BM 4K camera .mov clip I have downloaded is REC709 color except the one RAW clip.

    Is the extended gamut called "BMD Film"?

    Btw, some guy is posting "HDR" videos made from the original downloaded BM clips. The original clips are all REC709 color, so these are bogus HDR videos in terms of color.

  • Blackmagic Film Log has its own gamut, it's not Rec709. (You can easily test this with the Color Space Transformation tool in Resolve.)

    The difference between Raw and ProRes on all their cameras btw. is that Raw is 12bit and ProRes is 10bit.

  • @markr041

    I’d recommend you to download DaVinci Resolve Manual and read it throughoutly. You’ll get answers and better understanding for most of your questions.

    For proper gamut transform I’d recommend using ACES or “DaVinci Color science”, it’s not gonna clip your DR or color gamut (LUT’s will). You can use it for BMPCC as well.

  • @Vitaliy_Kiselev This is incorrect. Color science actually started from such experiments and advanced extremely.

    Do you mean psychological experiments that try to determine the nature of perception or experiments that measure physical characteristics of light? I would like to know more, if there is some soviet type experimentation that is not well known.

    As per gamut and DR I think these diagrams can clear things up a bit. Note that the DR of what our eyes can see is adaptive, meaning that night vision and seeing in strong sunlight are two different functions.



  • @libertas

    Color is very subjective, there is no way to tell how different people perceive color, you can only make tests of how different people distinguish different colors. Studies did show that around 8% of men can’t perceive some colors, so they have a color blindness of some kind.

    The only thing you can measure is how color on screen corresponds to color in real life. The problem here that there is no technology (real or futuristic or imaginable), which can fully simulate all light characteristics of a real subject, so you always have to make a simplified model for the sake of comparison. Even XYZ color space (the base of all color models we use in our days) can’t describe all colors accurately: for example there is a big chunk of purple spectrum which just can’t be described in math.

    There is also no proper way to compare eye “parameters” to monitors and cameras. Everything you do is subjective: something can look good enough now and really bad after few decades. Eye is a really unique instrument, I don’t think there will be a day anytime soon, when we will be able to completely “simulate” it. Just a simple example: in Soviet Union there were experiments which tried to find out what is the minimum amount of light the eye an register. It was found out that after a proper adaptation (staying in a dark room for long time) a person could register a single photon (!!!) of light.

  • Oh my god. What a babe.

    And the footage looks damn good too. But maybe I’m blinded by the model’s beauty.

    She’s thrown off my gamut response.

  • somewhat unrelated question. I've been studying camera movement lately. We all the slow motion shots mostly handheld? thanks.

  • @Lohmatij "for example there is a big chunk of purple spectrum which just can’t be described in math."

    This statement seems ridiculous. What we perceive as colors are portions of the visible spectrum of wavelengths. Purples are the wavelengths at the short end of the visible spectrum (wavelength 450–400 nm, frequency 670–750 THz) and consequently can be easily "described in math."

    If you want to say that no existing color space can accurately render every possible shade of purple, on the other hand, that may very well be true. I'm not familiar enough with color spaces to say.

  • The last shot in the Bubblegum video is of the operator holding what looks like a Ronin S, FWIW.

  • @eatstoomuchjam

    I wasn’t correct when I mentioned purple “spectrum”, the proper way would be to say “a group of purple colors.” I was talking about XYZ color model which can’t describe those colors accurately, and because RGB color model and all existing RGB color spaces are based on it, we can say that they will have trouble with rendering of those colors too.

    Keep in mind that while some colors, like violet, can be described as a a single wavelength , some colors has to be described as a combination of different wavelengths. Purple doesn’t have any corresponding wavelength, it’s a combination of pure blue and pure red, 2 colors on the opposite side of spectrum. The other simple example is color “white”, which consists of equal combination of all visible wavelengths.

  • I find the lee filter swatches very useful. There is a great app for android and iphone.

    The x and y are the coordinates in the CIE 1931 colorspace, which is the theoretical reference of every color we see and are able to reproduce, and then some presumably.

    @Lohmatij I agree with the absolute subjective dimension of colour, there is no way to determine if we two perceive a given wavelength in the same way. Any comparison is done through language. Wittgenstein's Remarks on color is a fascinating though pretty demanding read.

    Hope I am not hijacking this thread.

  • @bannedindv specs provided by the shooter of the Bubblegum video:

    • RAW 3:1 4K DCI
    • 24fps Project / 60fps Off-Speed
    • ISO 400, 1250, 3200, 5000
    • Samsung T5 SSD through USB-C

    DJI Ronin-S

    • Metabones EF to MFT T Speedbooster XL 0.64x
    • Shot with Contax Zeiss MM primes: 35/f2.8, 50/f1.4, 85/f1.4, 135/f2.8 + Hoya Pro NDs
    • 95% Ronin-S / 5% handheld.
  • The previous BMPCC wasn’t known for its low noise. It produced a fantastic picture, but it wasn’t exactly a low light performer. Fast forward to the BMPCC4K and things have changed, a lot. The new camera has dual native ISOs of 400 and 3200. However the camera will switch automatically to the upper gain circuit at ISO1250. Using ISO settings above or below those two native settings will have an effect on dynamic range, but only minimal.

  • Considering that even the top of the line 30 000$ high end color grading display don't even have the capability to properly display Rec2020 nor reproduce peak brightness necessary for HDR, I don't even know why the heck so many people actually want it in camera. Right now your only option unless you own two 4K laser Christie Projector is to generate a profile LUT to adjust your maximum screen capability to display correctly what it'll look like as a HDR output but you'll never be able to play with full gamut. And yes Vitaliy_Kiselev is right, you do color space transformation even from Rec709 to Rec2020, it all depends of your screen capabilities. The last tendency I've seen that messes up a lot of final output is people rushing to color Grade in P3 on their monitor (Eizo & BenQ to cite the most affordable). It's a huge mistake considering that it first don't display 100% of the gammut but has a gamma response that differs from the actual screen signal. It's far better to grade in RGB Rec709 and do a color space transform to DCI-XYZ, the result will be much more faithful if the display is well calibrated to Rec709.

    But if it's a concern, shoot Raw and do your own color management. Footages looks fantastic btw, I'm waiting though to see some report on bug/crash and what that cheap price has to bring forward as a compromise. But just for the fact you might get Resolve studio, it's a very good thing.

  • @GeoffreyKenner Have you ever produced an HDR video? Have you ever seen one on a 4K HDR TV, even a $299 one? Or even on an HDR-capable smart phone? If you have not, then I understand why you don't get it. If you have, then you must be blind. More colors are better than less colors, 1000 nits is better than 250 nits. And 12 stops of DR is better than 5-6. The difference between SDR and HDR is stunning seen on any HDR-capable device. Most of this is the DR advantage rather than the color, but still, to argue against reproducing more colors is silly. Is this - if it's not perfect, it's not worth doing?

    The Shogun Inferno can display 1500 nits, so you can even monitor in HDR in the field, albeit with less than the full color gamut.

    I agree that not many displays can display the full REC2020 gamut (one m, btw). So therefore, only shoot in REC709? What exactly are you recommending? And no, going from REC709 to REC2020 produces false colors. You cannot recreate colors in the world that are not recorded, by any math transform. Look at the 2D gamut graphs, REC709 misses more colors that we see than does REC2020 or S Gamut or V-gamut.

  • @Vitaliy_Kiselev can you please split off all posts about colour "science" into an extra topic and cleare this one up to be only about the new BMPCC4k?