was successfully added to your cart.

After thorough testing of the BMPCC (Black Magic Pocket Cinema Camera) I started to notice some specific behaviour of the camera, and just to prove some of my assumptions I decided to run a few additional tests to verify them.

Questions that I want to answer in this article:

  1. What are Zebras?
  2. What is the behaviour of Zebras on BMPCC?
  3. What is the best way to use Zebras on BMPCC to achieve best results?

Zebras

Zebras, or Zebra Patterning is a video camera feature to aid in correct exposure. Areas of the image that are over a certain threshold are covered with a striped pattern.

Zebras over a clipped area
Zebras over a clipped area

Blackmagic Pocket Cinema Camera gives us the option of using a range of the threshold settings from 75% to 100% (with 5% step in between).

ISO vs Zebras

The question I had in mind was – is using Zebras at 100% safe in every scenario? Does it protect highlights in RAW, ProRes, at any ISO setting? At any mode (film, video)?

Let’s have a look at series of tests.

Firstly let’s shoot at native ISO 800, and grab 2 clips:

  1. Exposed to the right (ETTR), no clipping
  2. Open the iris one step, to clip the highlights

This way we know where the line is. Then we are going to repeat the exact same test with different ISO settings.

ISO 800 (native)

ISO 800, not clipping at f/3.2
ISO 800, not clipping at f/3.2
ISO 800, clipping at f/2.8
ISO 800, clipping at f/2.8

ISO 200

ISO 200, not clipping at f/3.2
ISO 200, not clipping at f/3.2
ISO 200, clipping at f/2.8
ISO 200, clipping at f/2.8

ISO 1600

ISO 1600, not clipping at f/3.2
ISO 1600, not clipping at f/3.2
ISO 1600, clipping at f/2.8
ISO 1600, clipping at f/2.8

Observations

From the tests above it’s clear that Zebras remained unchanged. At f/3.2 there was no clipping, and after opening a lens a bit more to f/2.8, white area of the card started to clip. We got the exact same result on every ISO setting which means that ISO has no effect on Zebras and they are most likely calculated at the native ISO 800.

It’s worth noting that the behaviour of the histogram is different. The pictures above show that the histogram is calculated based on the displayed picture (after applying the LUT and ISO gain).

This is the main reason why we shouldn’t judge the exposure using the histogram on the Pocket Camera.

Clipped image at ISO 800 and ISO 1600

Now let’s have a closer look at what happens with hot areas when we increase the ISO to 1600.

Below is the waveform of clipped image at ISO 800 and ISO 1600:

ISO 800 vs ISO 1600 (exposure: -1)
ISO 800 vs ISO 1600 (exposure: -1)

The image on the right was corrected by -1 in exposure to match the left one.

They are identical which means that increasing the ISO to 1600 doesn’t clip the image more. Why is that? Keep in mind that BMPCC records RAW data in 12 bits. This means that the values recorded by the sensor can be amplified and safely stored in 12 bits of data without losing any information.

ProRes (Film and Video mode)

I did the same test with prores codec in both modes (film and video) and while for the film mode results were exactly the same as for RAW, in the video mode I got different results – zebras appeared way before, compared to the previous scenarios. Every exposure change made by ISO setting shifts the histogram as well as updates the Zebras. This means that the Video LUT was probably applied before Zebras were calculated.

Conclusions

  1. RAW – shift in histogram, no zebras change
  2. ProRes Film mode – shift in histogram, no zebras change
  3. ProRes Video mode – shift in histogram, zebras change

Both RAW and ProRes Film mode behave the same and changing ISO values has no effect on Zebras. Setting zebras to 100% and exposing to the right (ETTR – setting the exposure as high as possible without blowing out the highlights) is valid at any ISO setting.

Below is the diagram I created based off of results I got and it might not be 100% correct, but according to the behaviour of the camera it seems that it works more or less like this:

BMPCC signal flow
BMPCC signal flow

In the diagram we can see that Zebras go directly from the sensor and might be affected only by Record LUT which is used in one specific case – when the codec is set to ProRes and the mode is set to Video.

Meta indicates that when in RAW mode, the ISO value is stored as a parameter, image data is exactly the same for any ISO value.

  • 100% Zebras = sensor clipping (regardless of any settings)
  • Histogram – not reliable tool in terms of exposure judgement

Golden Rule of exposing with the BMPCC

Shoot RAW, set Zebras to 100% and use ETTR (expose to the right).

The above formula has been confirmed by many professional DPs and delivers stunning results. When you provide enough light to the sensor, the image will be as clean as it can be and the quality will be stunning.

More

Want to read more about this or similar topics? Let me know and subscribe to the newsletter to stay up to date! No spam, I promise!

  • Salvatore Castellana

    Thank you for this very informative! So for Raw Shooting I got it, then about prores film mode I just need to set the Iso but that will not affect the highlights, so for no clipping I need to adjust the exposure, right?

    • For ProRes Film mode the everything is the same as for RAW, except the ISO setting is “burned in” to the video file, and we cannot go back to native 800.

      So if you are in ProRes film mode just check zebras at 100% for clipping, and you are good to go. The ISO setting will just make your final image brighter or darker, but will not affect the clipping (if it’s not showing 100% zebras at ISO 200, it won’t clip even at ISO 1600).

  • Tom Rivest

    I must be missing something I see the histograms suggesting clipping in all the shots were clipping was occurring. What did I miss?

    • This is because histogram is based on the displayed image, not the image that is being recorded. In the screenshots I’m recording in “film” mode, and display is set to “video” mode. So the histogram is showing that the “video” mode is clipped, while sensor records perfectly fine image in film mode.

      This is why I mentioned that histogram cannot be trusted for the exposure judgement. Hope it’s clearer now.

  • joe schmoe

    I would set the bmpcc zebras at 90%-95%, instead of 100%

    There has been discussion on some of the boards about the bmpcc zebras “averaging” the three color channels. So, with the zebras set at 100%, the “average” of the three color channels might not show clipping, while one or two of the channels could actually be clipped — with no indication from the zebras.

    • Hi Joe, thanks for your comment. This is very interesting. It has never happened to me, and I was shooting loads of footage with zebras set to 100%, maybe this happens in a specific situation — I was trying to clip pure reds and blues, but didn’t manage to. Will have a closer look and will try to reproduce the problem myself.

  • Wouter

    Hi, first off: great blog you’re writing! Thanks!

    Secondly: I’m wondering if, when you set the zebra’s to 100% and ETTR, with RAW there are still one or more extra stops to recover highlights right? How much would this be? Haven’t had the time to test this, but maybe you have?

    • Wouter

      Ah, I’m sorry, I now see this: “100% Zebras = sensor clipping (regardless of any settings)”. I really thought there was still some wiggle room there…

      • Yes, I wouldn’t rely on highlight recovery. There might be “some” room, but in vast majority cases when the zebras are visible at 100% the details and color information are gone. And thanks for the nice words – glad to hear that it was useful for you!

  • Criss

    Great article, it improve the image a lot! Now I use this method of exposing!

  • Brad Confer

    I’m moving from a Lumix G6 to the BMPCC and I’m still learning to maximize the image possible for this camera. Fabulous explanation. Only one article on the topic and I get it. When I set up the camera I had an “Idunnknow” moment and kept the zebras wherever they were. I think 75% out of the box. I couldn’t figure out why everything was clipping in the monitor! That’s the well duh part. The part that I think is awesome is your diagram regarding iso, sorry…asa. That makes so much sense. But it brings up a question. It appears the camera is shooting at 800 regardless and simply boosting or reducing gain after the fact. So, this sounds to me like there is no reason at all to ever use 200 or 1600. You should get the same results only with greater control boosting or reducing exposure later in your favorite editing program. Is this true?

    • Hi Brad, I’m glad this article was helpful for you!
      It’s certainly confusing at first and it took me some time to figure it out 🙂
      To answer your question – you are absolutely right as long as you shoot in RAW. The only reason you might change the ISO to 200 is when you want to have a better preview on your screen.
      In ProRes the situation changes because the gain is applied before the frame is compressed into the prores codec. This means that shooting at ASA 200 and increasing the exposure in post will most likely increase the noise in the shadows.

      • Guillo Marin

        This right here said volumes. In other words using your asa as a tool to help see better on your monitor while still exposing correctly. My question: considering you are shooting film mode RAW and are using 200ASA to have a better preview would you switch back to 800 before rolling i.e. after you’ve confirmed your lighting is nice and the shot is ready to roll? Or does it essentially not matter as you can revert it back to 800 in post and then grade the image? Is the end result the same? A post on using false colors would be amazing! Thank you so much for your help this has been awesome and enlightening!

        • Personally I shoot at 800ASA all the time – I got used to the image I get on the screen, but if I had to switch to different ASA it wouldn’t matter that much. The RAW data is exactly the same and the Zebras behavior is unchanged as well.
          The false color is a bit tricky when it comes to blackmagic cameras (except for the URSA) because we have to rely on the monitors’ features so I would need to customize to false color ranges according to the image that the camera outputs. Having a LUT would help in this case as it would expand the image to let’s say rec 709 and we could work in the proper contrast and brightness ranges rather than BMD film.

          Hope this helps! Thanks for the comments Guillo!

  • Nicholas Recagno

    That is a fantastic article which provided me with the confirmation for the settings I already use which only really differ in the fact that I set the Display to Film Log knowing that as long as I push the signal as far to the right without seeing Zebras I will get the largest dynamic range. However the log display makes impossible to assess colour temperature which I had always fond

    • Hi Nicholas, thanks for your comment. Keep in mind that the exact same rules apply when your display is set to video mode. You can rely on the zebras the same but you have a bit more contrasty and colorful image to judge the WB. I am also looking forward to using external monitoring with a LUT applied in real time.
      Thanks again and have a great weekend! 🙂

      • Nicholas Recagno

        After I wrote the comment, it niggled me. You are absolutely right the Zebra does not change, don’t know why I was convinced otherwise. The problem is still there though, evaluating white balance in the display in video mode is hit and miss compared to using the LUT 🙂

        • Exactly, especially with the external monitor as the HDMI signal is always in film mode (there is no way to set the HDMI signal to video mode).

          • Guillo Marin

            Interesting I did not realize this as I have both cameras a pocket and a 2.5k production BMCC and usually use the big brother for my major work with monitor. Can you discuss how you would go about using a LUT with a monitor in conjunction with a monitor for exposure (Zebras)?

          • More or less what I mentioned in the comment above. I use the internal zebras to know where the clipping point is which helps me expose to the right and push as much light to the sensor as possible. I could trust the False Color if it was applied in the camera internally. Using False Color in the monitor I would need to work on the image first to “normalize” it to the same scale that the monitor works (let’s say rec 709).

Read previous post:
Caching secrets in DaVinci Resolve

Caching in DaVinci resolve was mentioned in my recent article about its performance, and today I decided to write a...

Close