Blackmagic Pocket Cinema Camera 6K Lab Test – Dynamic Range, Latitude, Rolling Shutter & More

September 2nd, 2019
Blackmagic Pocket Cinema Camera 6K Lab Test - Dynamic Range, Latitude, Rolling Shutter & More

After part 1 and part 2 of our coverage of the new Blackmagic Pocket Cinema Camera 6K we finally can bring the cinema5D Lab results to you. Interested to see how it fares? Will the lab reveal differences to the Pocket 4K? Well – let’s get started …

Being a long time owner & user of the original Blackmagic Pocket Cinema Camera for 6 years now, and having tested the BMPCC 4K already in our lab a while ago, I was now really curious to get my hands on the new BMPCC 6K.

The spec’s on Blackmagic Design’s website would list no difference to the Pocket 4K in terms of dynamic range – both cameras are listed with 13 stops dynamic range.

So we decided to do our new latitude test in addition to the standard dynamic range and rolling shutter test with the BMPCC 6K and BMPCC 4K in order to see if we can find differences. And yes, as you will see below there are differences.

But let’s start first with the rolling shutter measurements.

BMPCC 6K Rolling Shutter Test

We are using a new test method based on a high frequency, quartz & micro controller based strobe light at 300Hz. As the sensor readout happens from the top to the bottom, we are getting a sequence of black and white bars from the strobe.

The strobe test reveals a rolling shutter of 19.8 [ms] – now we know why it cannot do full sensor readout in 6K faster than 50 frames per second (60fps at 1:2.4 aspect ratio) – there is simply not enough time for a higher frame rate in full sensor mode.

This rolling shutter value is significantly higher than the BMPCC 4K which has a rolling shutter of 16 [ms].

BMPCC 6K rolling shutter result: 19.8 [ms]

Dynamic Range of the BMPCC 6K

Like always, we use our DSC Labs Xyla 21 Stepchart in conjunction with the IMATEST software to derive our results (see how we test here).

The BMPCC 6K uses a dual ISO sensor with 400 and 3200 as base values, hence both will be shown below.

Dynamic Range at full sensor readout, ISO400, 6K BRAW 25p

Using the full sensor readout at the maximum resolution of 6K (6144×3456) with BRAW 3:1 constant bitrate settings, the waveform plot of the stepchart visually reveals about 12 stops of dynamic range.

BMPCC 6K waveform plot of the stepchart. About 12 stops can be identified above the noise floor.

Developing the raw files via DaVinci Resolve and exporting high quality tiff’s to IMATEST a dynamic range of 11.8 stops at a signal to noise ratio of 2 is calculated (12.9 stops for SNR = 1).

BMPCC 6K IMATEST result of 6K BRAW ISO400. 11.8 stops are calculated for a signal to noise ratio of 2.

This is 0.2 stops better than the result we got for the BMPCC 4K.

Dynamic Range at full sensor readout, UHD ProRes HQ ISO400

Quite interestingly, using the ProRes HQ full sensor readout at Ultra HD, the dynamic range increases slightly from 11.8 stops to 11.9 stops at a signal to noise ratio of 2 (13 stops for SNR = 1).

As mentioned by Blackmagic Design, the full sensor 6K resolution is scaled to Ultra HD, thereby reducing the noise hence increasing slightly the dynamic range – quite nice!

Dynamic Range at full sensor readout, ISO3200, 6K BRAW 25p

At ISO3200 the dynamic range drops quite significantly to 10.0 stops (SNR = 2, medium value in the upper right hand table below) according to the IMATEST calculations, see the result below. 

BMPCC 6K IMATEST result of 6K BRAW ISO3200. 10 stops are calculated for a signal to noise ratio of 2.

This is less than the BMPCC 4K which shows a corresponding value of 10.5 stops at the higher ISO setting of the dual gain circuit.

But as mentioned many times before, while transmissive tests using stepcharts are the simplest/most accurate/comparable tests of dynamic range – they also give us very little information regarding how the camera responds to colors and detail across that range.

That is why we are introducing an additional lab test, the latitude test.

The cinema5D latitude test

First results of this new test have been shown already in the Panasonic S1 VLOG article of my colleague Johnnie, but let me briefly explain again what it is.

Latitude basically shows the capability of a camera to retain color and detail while over- or underexposing the image and normalizing it thereafter.

In our latitude test, we have a standard scene where either Johnnie or Nino are sitting next to a Color Checker.

Standard studio test scene showing Nino’s face at 60% luma value, the white sheet of paper at 65% luma.

The face is exposed at max. 60% luma value, the white sheet of paper at 65%. From there on in a succession of tests we always double the shutter value from scene to scene, hence underexposing the scene by 1 stop from the previous one until we have reached 5 stops of underexposure.

BMPCC 6K 5 stops underexposed

Hence, we brutally underexpose the studio scene – and then normalize it back to 0 stops to reveal the color, detail  and noise. This is really a torture test and the images of most cameras with less sophisticated codecs break down already at 2-3 stops underexposure.

Both the Pocket 4K (using 3840×2160 resolution) and 6K (6144×3456 resolution) were setup with BRAW 3:1 constant bitrate, ISO400 25p. The shots from both cams were graded in DaVinci Resolve using a Blackmagic film to video LUT, and then I tried to match both cams as closely as possible by some smallish tweaks.

BMPCC 6K 5 stops underexposed – pushed back to 0 via the camera raw tab in DaVinci Resolve

And above you can already see the result – the BMPCC 6K camera fares very very well in this test. It starts to show horizontal stripes (which cannot be removed by noise reduction as you will see further down) and the noise becomes excessive, but color information is retained very well.

See below the same scene, shot with the BMPCC 4K, also 5 stops underexposed and brought back to 0:

BMPCC 4K 5 stops underexposed, brought back to 0 using the camera raw tab in DaVinci Resolve

Clearly, color information deteriorates with the BMPCC 4K and the image becomes pink. Let’s remember that both cams started like this:

BMPCC 4K vs 6K at the zero underexposure studio scene setting.

Here are both cams again side by side at 5 stops underexposure, brought back to zero:

BMPCC 4K vs. BMPCC 6K at 5 stops underexposure, brought back to 0

Hence, in this test it becomes clear that the BMPCC 6K retains the color surprisingly well, whereas the BMPCC 4K image starts to become pink already at 3 stops under, and larger patches of pink noise appears – which is difficult to remove as you will see below.

What I found interesting though, while the waveforms at zero underexposure look completely identical on both cams (after grading), the more we underexpose and bring it back by adjusting the exposure (by the exact amount of stops) in the camera raw tab in Resolve, the brighter the BMPCC 4K image appears. The BMPCC 6K image stays the same across the exposure range.

BMPCC 4K vs. BMPCC 6K side by side at 3 stops underexposure, brought back to 0 using the camera raw tab in DaVinci Resolve – the limit of underexposure is reached with the BMPCC 4K whereas the Pocket 6K is still fine

Now, as we shot in BRAW with both cams, the big question is whether using noise reduction in DaVinci Resolve can save some of the underexposed shots (temporal NR: 3 frames, threshold:30, spatial threshold: 10).

Yes, it can, but only on to a certain point – as soon as the horizontal stripes start to appear in the image, noise reduction cannot get rid of those and I found that the pink chroma noise is not removed any longer at 4 stops underexposure with the BMPCC 4K.

On the BMPCC 6K, noise reduction helps to get an almost stripe and noise free image up until 4 stops under – impressive! The 2x higher resolution of the BMPCC 6K along with noise reduction certainly helps to achieve this result.

And this brings us also to the clear, final result, using noise reduction:  BMPCC 4K is very clean until 3 stops under, BMPCC 6K until 4 stops under!

(Note: this result is clear during playback of the video files – the static, scaled images in this article tend to look much better than the moving images).

BMPCC 4K vs BMPCC 6K at 4 stops underexposure, brought back via the camera raw tab in Resolve, temporal and spatial noise reduction applied

Conclusion

The lab results obtained with the new BMPCC 6K are impressive to say the least. The combination of a very good dynamic range with a superb codec (Blackmagic RAW) leads to superb images.

At a time where brand new much more expensive “cinema” camera’s are released with H.265 / H.264 as the only codec options, Blackmagic Design once more shows us all what is possible at this price point. 

The only little drawback with the Pocket 6K that I see from the lab test is the rolling shutter value which is a bit on the higher side – given the lack of internal stabilization handheld shots are risky business with this camera.

Have you made your first experiences with the new BMPCC 6K? What do you think? Let us know in the comments below.

48
Leave a reply

guest
1
1
- 2
Filter:
all
Sort by:
latest
Rolf
Rolf
Guest
July 22nd, 2020

Can we assume that there around 12.3stops when shooting in Prores 4K (Since it downsamples from 6k to 4k)

Anthony
Anthony
Guest
June 11th, 2020

If it doesn’t have 13 stops, why waste money on it?

Interceptor121
Guest
May 13th, 2020

This is a good test because RAW is recorded linearly by the camera. You should perform the same test on the GH5s using a standard Like709 profile to see the difference and not V-Log. It should produce similar result with a skew for the jpeg engine. Are the value reported on the BMPCC4K done with the same procedure?

Joshua Kidd
Guest
April 21st, 2020

Went back and watched this after watching a few of the other tests. It may be worth noting that if you drop down to 5.7k from the full 6k, the rolling shutter improves significantly…around 3ms

 Brandon Moye
Brandon Moye
Member
September 12th, 2019

As of Sept 12th the 6K Pre-order is at $1996usd on B&H!

Johnnie Behiri
Admin
September 12th, 2019
Reply to  Brandon Moye

Thank you Brandon. We are actually on it as we speak. A short article mentioning the new price is coming up.

 Brandon Moye
Brandon Moye
Member
September 12th, 2019
Reply to  Johnnie Behiri

Apparently it changed back? Not sure if it was a glitch on BH, but the price has reverted back to 2495. I wish they would stop playing with my heart cause I was just about to purchase it at that new price.

 Manuel Manuel
Manuel Manuel
Member
September 7th, 2019

Thank for this amazing article. I’m afraid it explains why my bmpcc 6k shows so much motion jitters when people mi se in the frame…

 Tom Roper
Tom Roper
Member
September 6th, 2019

It seems Dual ISO is no free lunch if it costs 2 stops DR at 3200 ISO?

 Tom Roper
Tom Roper
Member
September 6th, 2019
Reply to  Gunther Machu

If I shot at 800 and pushed it two stops in post, isn’t it the same thing? I bought the camera BTW from your B&H link, will have it Tuesday with free shipping, on the basis of the latitude test you did. :-)

 Brandon Moye
Brandon Moye
Member
September 12th, 2019
Reply to  Tom Roper

Tom, See if B&H will give you $500 dollars back, as it’s now listed at $1996

 Claus Meyer
Claus Meyer
Member
September 6th, 2019

Wow, this is the kind of test, I want to see… please do one on the orginal pocket 1080, I want to see how close to 13 stops it was :D

 A.c. Concepcion
A.c. Concepcion
Member
September 5th, 2019

Hi Gunther – Very informative stuff. Do you think adding the metabones ultra speed booster .71x to the bmccp4k will improve it’s latitude test? Thank you!

Dalton Randall
Dalton Randall
Guest
October 13th, 2019
Reply to  Gunther Machu

Ah, but increasing the light hitting the sensor means a potential decrease in ISO needed to compensate – which in certain situations would increase the DR, all else being equal. :)

 Wayne S
Wayne S
Member
September 4th, 2019

Hi Gunther. It would be good to have a comparison between the Xyla tiles images, and raw Imatest images, for the Pocket 4k and 6k (and 4.6k Pro)? The Imatest of the 6k is strikingly similar to that of 4.6k.

Usable stops of latitude is a very subjective term. For various purposes, the lower quality stops maybe worth recovering. Seeing the comparative test images of the tiles we can judge how usable the lower stops can be for our own needs.

Thanks.

 Wayne S
Wayne S
Member
September 5th, 2019
Reply to  Gunther Machu

Sorry, the waveform result, and step chart.

The curves in the result and the height spacing at the ends. Looking at it, you could use processing to extract a similar amount of stops.

 Wayne S
Wayne S
Member
September 7th, 2019
Reply to  Gunther Machu

Thanks for those Gunther. I’m trying to make a crude jpeg with all the relevant pictures in one (the 3200 iso ones I’m leaving out).

 Wayne S
Wayne S
Member
September 9th, 2019
Reply to  Gunther Machu

OK Gunther. I put together a crude jpeg on one of my phones. I had to install adobe lightroom on the $39 phone just to get to view the 6k tiff, that was something.

The jpeg got a bit messed up, but doing comparisons, the 4k got an extra stop of something discernible at a solid tile, over the 6k. But the 6k result seems marred in these last stops at sensor level, with an odd dark patch. Assuming the dark patch might not normally be there, that would give one stop better then the 4k, and similar to the 4.6k, of extractable fairly crude basic detail stops. Temporal noise reduction definitely desirable.

I guess the zcam was not in HDR mode, as it scored the worse of the lot. The bluish noise of the 4.6k was obvious, but the still if the zcam seemed to be lifted (got to go back and check the noise.

I use this phone on such reviews, because the basic LCD just needs to be tilted to brighten the dark areas and see what is going on.

Thanks for the I images again. I will have to view them on a desktop to get a proper idea ;f noise.

But, the reality here is that the 4k, 6k, and 4.6k are incredibly close in discernable stops, despite what the instantly usable stops may be. Actually, similar to test charts I have seen on the iPhone se, but I don’t trust those particular charts. However, as far as phones tested with those charts, I phone se did well in latitude compared to all top phones I looked at. Some models doing an extra discernible stop or so. Would be interesting to see what a se could do on a proper setup like thus compared to the pocket 4k (likely 2+ discernible stops down).

One last thing, is the zcam an actual known Sony sensor? You can improve look with a different colour filter material. So, I wonder if the 6k and zcam are using that s yo get their images?

 Alexander Kolomiec
Alexander Kolomiec
Member
September 4th, 2019

6k, many people want to see a short version of RF. They managed to collect unique optics, and throwing it all away will be a betrayal …

 Rob Jackson
Rob Jackson
Member
September 3rd, 2019

The DR increases slightly with UHD ProRes. So in theory would it increase slightly even more recording in 1080p ProRes HQ?

I believe that’s a full sensor readout. I have both pockets and resolution isn’t a priority as much as DR is for me. A majority of my projects are exported in 1080p and as much as I love editing BRAW, 1080p ProRes is a much more efficient editing workflow for me and suffice for most work that I do.

 Jacob Stencil
Jacob Stencil
Member
September 3rd, 2019

Gunther, in the underexposed 5 stops 4K image I see two faint vertical lines: one on the right quarter and a second on the left third. I see the same fixed lines when panning across darker areas using ISO 640-1000 on my camera, but these disappear when using ISO 1250+. Any insight as to what these are?

Blackmagic has told me that they are normal camera behavior in underexposure, but I don’t understand how two fixed, unmoving lines, unrelated to noise, are “normal behavior”.

 Mike Courian
Mike Courian
Member
September 3rd, 2019

Would love to see the lattitude results for over exposure also! Great information here! Thank you for sharing.

Member
September 2nd, 2019

Please blackmagic release av version of the 6K camera with canon RF, SOny E, L-mount or Z mount

 Alex Ribeiro
Alex Ribeiro
Member
September 3rd, 2019

I use mine with Nikon lenses through a cheap Chinese adaptor. Meter markings are not accurate but it works. The only advantage of actually using Canon lenses is iris control in camera. Shake reduction doesn’t work and autofocus sucks really bad.

 Paul Corneille
Paul Corneille
Member
September 3rd, 2019

But why? The BMPCC 6K Is not a vblogger camera.
The only mount can be useful is ‘new’ARRI LPL.
The autofocus is unusable on the BMPCC why you want an expensive auto focus lens mount?
Better put money to buy a nice prime Cinema lens kit.umless you really need autofocus but on this case buy a Sony.

 Rastko Vukovic
Rastko Vukovic
Member
September 2nd, 2019

Wonderful and trustworthy test and review as usual. There are some concerns, in my opinion, with this camera. Highlight recovery option in DaVinci resolve is, as I understand, an artificial thing that is bringing back channels that are not clipped and therefore not reliable for a professional shooting and not an actual dynamic range for all three channels? If so I certainly hope it wasn’t included in the test. It would be like cheating. Also camera malfunctioning possibility is too real for this brand, it should be mentioned in every single part of the review, seems like an important thing. For the 4K version third of the people (half joke approximation) who bought it on B&H had to return it, you can read it in the early B&H comments yourself. If so it shouldn’t just be mentioned it should be on top of the page with big red letters. Thanks again for the review.

 Rastko Vukovic
Rastko Vukovic
Member
September 3rd, 2019
Reply to  Gunther Machu

Thanks.

 Wayne S
Wayne S
Member
September 4th, 2019
Reply to  Gunther Machu

I think testing the difference highlight recovery makes is important. The tests probably show clipping on channels at different times at the end. It might be interesting to test using primaries on peak RGB frequencies of the sensor, to see the latitude of each. But it also means that different light sources, of different colour temperature of white, could affect the latitude.

Thanks again.

 Wayne S
Wayne S
Member
September 7th, 2019
Reply to  Gunther Machu

Thanks Gunther. I just meant as an add on test, it’s useful to see how far you can stretch a camera, like with detail preserving noise reduction also. There is a guy over at the bm forums, Dmitry, who has done a workflow to extend dynamic range on older camera. So post processing is useful to get to the end stops, and it is going vary between cameras based on their colour filter and sensor response curves, and noise characteristics. That’s why the stepchart and waveform can show just what is there to work with. I could easily see that while one pocket could reach up towards the 4.6k, it’s last two stops were so polluted, they were not worth trying to recover, and the ones between that and your usable stops result, were not exactly great either. It’s the sort of stops you still might try to use in ENG on occasion though (for something important the team can establish an auto workflow for the camera model).

 Paul Corneille
Paul Corneille
Member
September 3rd, 2019
Reply to  Rastko Vukovic

We didn’t have any issue with BMD cameras since the very first Cinema Camera many many year ago and since then we had many many BMD camerae and models and the only issue we had is when we try to test not certified gears.
The reliability of BPD cameras is not inferior than other famous brand, (some one say RED??).
We travel few times for months for documentaries with one or more cameras and we never had an issue.
The malfunctioning if is present normally appear just out of the box and not month later like other brands.
IMO your propaganda to try to attempt BMD reputation is unjustified.

 Rastko Vukovic
Rastko Vukovic
Member
September 3rd, 2019
Reply to  Paul Corneille

They have great cameras and allows people to do great things. And yes, compared to Red they are reliable. But don’t tell me that what I said is not true. If it’s just out of the box then quality control, literally any quality control in the factory would catch malfunctioning units. You turn it on and it doesn’t work..taadaa. So it’s left to the people to do the quality control and send units back on their own cost. Also bad reviews are from the customers, you can go and recheck the fact that I stated. You obviously didn’t even do that or you work for black magic man (being paid or not):) Being critical with a reason is productive, being brand loyal- not really.

Mirko Jankovic
Guest
September 2nd, 2019

the bigger the sensor the bigger the DR.

1
1
- 2
Filter:
all
Sort by:
latest

Take part in the CineD community experience