Announcement

Collapse
No announcement yet.

Oops! BRAW isnt Raw Its a YCbCr codec

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Oops! BRAW isnt Raw Its a YCbCr codec

    I did an in depth look at the green screen footage shot by Nate Porter and came to a surprising discovery. BRAW is not RAW. It has undergone a color space conversion to YCbCr.

    I do a lot of green screen compositing, so Im highly attuned to the various artifacts that different color spaces and codecs present to the keyer. One quick way to see problems is to examine the red channel. Below are the red channels from some of Nates clips. Ive enlarged them 2x to make it easier for you to see.

    The telltale signature of a YCbCr color space is a white ringing or halo along the border between flesh tones and the green screen. It appears on other colors as well, reds in particular. As you can see in the pictures, Prores 422 and 444 have this ringing. And so does BRAW! However, Cinema DNG is perfectly clean, which is what youd expect from true RAW.

    So, what do we make of all this? I'll leave that for you folks to say. As for me, when I shoot green or blue screen, I will definitely be using Cinema DNG!

    Prores 444.jpgProres 422.jpgCDNG.jpgBRAW.jpg

  • #2
    i also noticed that all BRAW samples have some kind of strange color halo at color patches edges.
    Also less noise, less moire but same time less sharpness. Probably not a problem when you downscale 4K to HD, but for Pocket or Micro with native HD resolution and OLPF filter BRAW may be too soft with too compressed chroma.


    All my custom made accessories for BMMCC/BMMSC now available here https://lavky.com/radioproektor/

    Comment


    • #3
      Well, they already said the part of the demosaicing was done in-camera. It's good to know fow VFX plates but the BMRaw quality is good enough for me and my fiction pieces. Good finding!
      lex Montoya on IMDB
      https://www.imdb.com/name/nm0600257/

      Comment


      • #4
        Interesting, wonder if BM will comment on this. For me all the positives outweigh this but it's good to know.

        Comment


        • #5
          Compression always loses something. But perfection is relative to purpose.

          Comment


          • #6
            Nice find. Of all of the forms of "lossy" compression, Y’CbCr is probably the least egregious, but it is good to know that for pristine 4K compositing cDNG remains the best option.
            Pocketluts: Purpose-built LUTs for the Blackmagic Design Pocket Cinema Camera 4K
            Pocketluts Store

            Comment


            • #7
              Good to know, but as I mostly just shoot ProRes, if it is as good or better with less data, that is good enough for me.

              Comment


              • #8
                Interesting find! What is it about Y’CrCb that causes this ringing even when encoded at 444?

                Comment


                • #9
                  Interesting. Still a very nice codec, but this means there's stuff for which you'll still want to use cDNG.
                  My YouTube channel

                  Comment


                  • #10
                    Originally posted by Alex.Mitchell View Post
                    Interesting find! What is it about Y’CrCb that causes this ringing even when encoded at 444?
                    Good question. I don't know the specifics, but there's something in the mathematical calculations of deriving Y;CrBr from RGB that causes it. I frequently see that people think 444 is identical to RGB. It is not. And this is a great example right here.

                    Comment


                    • #11
                      I'm looking at some other test footage. I don't know who shot it but it has a color chart, a mannequin bust with a red shirt, a laptop, and a TV playing chroma noise and I can't see the same ringing. Porter's footage appears to be lit brighter so maybe that has to do it and this footage doesn't have any areas with nearly as much contrast between red and green.

                      Also what are the advantages, compression-wise or otherwise, of converting it to YCbCr? Edit: Actually now that I think about it I guess it would make two of the channels lower contrast and thus easier to compress.
                      Last edited by Myownfriend; 09-20-2018, 09:45 AM.

                      Comment


                      • #12
                        Originally posted by Myownfriend View Post
                        I'm looking at some other test footage. I don't know who shot it but it has a color chart, a mannequin bust with a red shirt, a laptop, and a TV playing chroma noise and I can't see the same ringing. Porter's footage appears to be lit brighter so maybe that has to do it and this footage doesn't have any areas with nearly as much contrast between red and green.

                        Also what are the advantages, compression-wise or otherwise, of converting it to YCbCr? Edit: Actually now that I think about it I guess it would make two of the channels lower contrast and thus easier to compress.
                        I think the main advantage is that you can easily reduce the resolution or compress the Cb and Cr channel without a big visual loss. That's the idea behind 422 and 420. I think most compressed codecs are YCbCr for this reason. It's interesting that CDNG 3:1 does still seem to be RGB. I guess CDNG only allows RGB and that switching to YCbCr with BRAW may have made some of these big improvements in compression possible for BMD.

                        Maybe BRAW only switches to YCbCr after a specific compression ratio, say up to 5:1 it's RGB and then 8:1 and more is YCbCr.
                        Last edited by weui; 09-20-2018, 12:24 PM.

                        Comment


                        • #13
                          Interesting. So would this recommendation hold for just green/blue screen FX work, or any situation where you'd need to key (like wanting to key an actor so you could place text "behind" them and the plate, for example)?

                          Comment


                          • #14
                            Would be interesting to test:
                            braw 3:1
                            braw 5:1
                            braw 8:1
                            braw 12:1

                            Comment


                            • #15
                              Originally posted by S_Berger View Post
                              Would be interesting to test:
                              braw 3:1
                              braw 5:1
                              braw 8:1
                              braw 12:1
                              Same result for all ratios. I only showed a few to keep the size of the post manageable.
                              You can download the original clips and see for yourself.

                              https://drive.google.com/drive/folde...jt2gDYI8H-n7PY
                              Last edited by Ralph B; 09-20-2018, 02:07 PM.

                              Comment

                              Working...
                              X