Announcement

Collapse
No announcement yet.

Documentary workflow with Canon 6D, Magic Lantern and Denoiser II (or Neat Video)

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

    Documentary workflow with Canon 6D, Magic Lantern and Denoiser II (or Neat Video)

    Hi,

    I'm trying to establish best workflow for shooting a short documentary with Canon 6D. I'm going to shoot H.264 (obviously) with ML and VisionTech profile.

    I've read a lot on the best DSLR workflow, but I'm still not clear on some of the issues. The most common recommended workflow looks (as far as I understood) like this:

    1) Open your footage in After Effects, change color depth to 16-bit/channel
    2) Denoise
    3) Dither
    4) Export to ProRes or DNxHD
    5) Edit
    6) Apply LUTs and color correction (I might be getting this step wrong... and I'm not clear on the exact LUT process)
    7) Export video to H.264

    It all seems fine when there are just a few files to process, unfortunatelly - it's usually not the case. So how do you go about processing multiple files? I figured out two possible ways, but neither is really convinient:

    Workflow 1:
    a) Open footage in After effects, change color depth to 16-bit/channel
    b) create a separate sequence for each clip,
    c) apply denoiser to each clip (Denoiser II seems more convinient, as there's no need to create a profile),
    d) dither
    e) save the project and re-open it in media encoder (each sequence separately)
    f) export all the files to ProRes 422(LT)
    g) Edit
    h) Apply LUTs and color correction (I might be getting this step wrong... and I'm not clear on the exact LUT process)
    i) Export video to H.264

    Workflow 2:
    a) Convert all footage to ProRes (in media encoder)
    b) Edit in Premiere
    c) Apply Denoiser (again Denoiser II seems a better option, as there are some issues with Neat Video in Premiere CC) and Dither in Premiere, render each clip on a timeline
    d) Apply LUTs and color correction (I might be getting this step wrong... and I'm not clear on the exact LUT process)
    e) Export video to H.264

    Second workflow seems easier, but I might be missing some important issues. Maybe there's another, better way to handle the process? I have to admit I feel that Neat Video gives me much better results then Denoiser II, but seems harder to integrate into the workflow (and is significantly pricier - it's 50USD for Denoiser II Academic [works in Premiere and AE] vs 163 USD for Neat Video [Premier and AE, 97 USD for Premiere/AE only]. Any thoughts on that?

    Thanks a lot for all your input, I feel a bit lost at the moment .

    #2
    If it were me, I would test both workflows to see which one was a better fit.

    Consider the time needed, as well as the storage space needed, IE Expense, to do all that conversion of X hours of doc footage.

    Comment


      #3
      That seems unnecessarily ridiculous.

      My worfklow = edit.

      That's it. You shoot h264, drop it in your edit and then you edit.

      If you feel that you need some special sauces, grades and noise reductions afterward...do them afterward, once the edit locked.

      Comment


        #4
        Garbage in, Garbage out. Any workflow is only going to make your footage look marginally better at best. If you really feel that you need it, get a better camera, time is money and your workflow takes a lot of time.

        The only thing I would recommend with any camera (which you don't have on your list) is applying a subtle amount of sharpening to your entire sequence before exporting.
        Cameras: 2x - Sony FS7, 2x - Sony A6500, Canon 5D IV, DJI Mavic Pro, Canon 5D II, Canon 60D, Canon G16, Canon Rebel XT, GoPro Hero 7, Gopro Hero 6 (RIP), 6x - GoPro Hero 3+ Black Edition, Canon XL2, iPhone 4, iPhone 6, Ricoh KR-10, Fed-2, Fujica Half Frame, Canon ZR-100, Sony DCR-TRV 310.

        Comment


          #5
          You can produce pretty amazing results with a Canon DSLR. The way to get the absolute highest image quality from them is as follows in my years of tests:

          -Expose highly without clipping, underexposong is the key to getting crappy Canon footage. Use ML waveform or Histogram.

          -Don't shoot with a third party picture style, they all have a downfall like a weird rendition of a certain coloir, increased noise at certain channels, the only one that's 100% consistent is Technicolour's Cinestyle (as it was developed with Canon) and of course the trusted Neutral with contrast -4.

          -If you find the footage to be too soft, add a single notch in the picture style menu it really does make a difference, just try it.

          -Moire and aliasing are your only enemy in Canon DSLRs so avoiding wide detailed landscape shots and concentrating on longer close up coverage makes the quality appear to audience so much better.

          -Take that footage into your editor natively, straight from the camera, edit cut and colour grade the footage to your liking (and work in 16/32bit space)

          -After you reach the final result, add a layer of neat video, go to advanced tab and tweak the NR settings to your exact preference, adding NR before grading is much worse than after, as doing it after grading eliminates the colour pushing artefacts and gives an overal higher control of how NR will look.

          -If the image is too soft, add a sharpening filter here after NR, a small one.

          -If you have banding or find the footage too clean, dither by adding a small layer of grain.

          -Export at ProRes or DNxHD.

          Don't complicate it too much. Just edit natively, grade, NR, sharpen, dither or not, export to a high quality format (anything not lpwer than the native H.264 34mbps 4:2:0 8bit)

          Comment


            #6
            Hi Ebrahim,

            Originally posted by Ebrahim Saadawi View Post

            -Don't shoot with a third party picture style, they all have a downfall like a weird rendition of a certain coloir, increased noise at certain channels, the only one that's 100% consistent is Technicolour's Cinestyle (as it was developed with Canon) and of course the trusted Neutral with contrast -4.


            Does it include also VisionColr and VisionTech?

            Comment


              #7
              Visiontech was in the tests among the rest and I do remember it being pretty neutral and conservative, no weird stuff compared to all the other custom made profiles

              Comment


                #8
                Thanks.

                Comment


                  #9
                  Originally posted by maarek View Post
                  That seems unnecessarily ridiculous.

                  My worfklow = edit.

                  That's it. You shoot h264, drop it in your edit and then you edit.

                  If you feel that you need some special sauces, grades and noise reductions afterward...do them afterward, once the edit locked.
                  +1

                  Also I would add it is better to do noise reduction as your final step.

                  I work in ML raw but can say that Neat Video in Resolve can really slow down the program especially if you're being aggressive so it is best to do color first.

                  The AE workflow (#1) seems like a lot of work and I'm not crazy about using AE as a batch renderer.
                  Kahn Artist Filmworks
                  FaceBook

                  Comment

                  Working...
                  X