Weird vertical seperation in BMPCC4K image?

Shooting a casting with my BMPCC4K i noticed a vertical seperation in some of the footage. I grabbed a still in Davinci Resolve and graded it so that you can clearly see the seperation in the image below.

vertical_line.jpg (670.36 KiB) Viewed 3052 times

The BMPCC4K settings are:
BRAW Q0 – 2.6K – ISO 1000 – 24 FPS – 180° Shutter Angle – Firmware 6.9.5

I guess it is not fixed-pattern noise, because there is not pattern, just a seperation.

Do you know what caused this?

EDIT 19.01.2021
I did some testing: Different frame rates (24, 25), different sensor modes (4K "Full Sensor" & 2.6K "Windowed Sensor"), different ISO settings (400-25600). Results:
The frame rate and sensor mode (resolution) do not have an effect on it. But the ISO seems to have. It seems to appear only in ISO 1000 and below.

Do you see that vertical lines in other shots with different settings? Do you see it with ISO 400? Only in 2.6K?

Try updating to the latest firmware by downloading the most recent Camera Setup update from the BMD Support page.

Thanks for your answer. I don't have the camera here do do some more testing right now.

Here is one more still, which has the separation. First with applied LUT, then without:

vertical_separation_LUT.jpg (698.3 KiB) Viewed 2987 times

vertical_separation_RAW.jpg (830.85 KiB) Viewed 2987 times

I will update to the new firmware.

uhm, try to rest camera, recalibrate sensor and test again with cap, with gray paper and see if there is the problem, if it s still present, contact assistance bmd by web site from support section, scroll down the page and use the form to request assistance.

uhm, try to rest camera, recalibrate sensor and test again with cap, with gray paper and see if there is the problem, if it s still present, contact assistance bmd by web site from support section, scroll down the page and use the form to request assistance.

Thanks, I will do that as soon as I have the cam here to test. Then I will post the results here.

Carlo, there is an option to Remap Pixels, but there is no Calibrate Sensor function on the BMPCC4K.

Carlo, there is an option to Remap Pixels, but there is no Calibrate Sensor function on the BMPCC4K.

Right Rick, i forget that
i had ump and pockets, and forget that calibration is ursa only.

Same problem here since I got my P4K three years ago . Reported and send it to be serviced in the UK. Blackmagic was politely supportive but never solved the problem. In my case the only solution — which only disguises this anomaly — is to shoot at 1250, where in shadows to light transition situations. As soon as ISO is superior to 400 it only gets worst and nothing can be done. The same can be said at ISO superior to 1250, but less noticeable. If this is not an expected behaviour from the sensor, Blackmagic should have done better in supporting those who reported it.

I updated the firmware – still the vertical seperation.
I remapped the pixels – still the vertical seperation.
I reset camera settings – still the vertical seperation.

Tormorrow I will do a few more test, to define at which exact settings the seperation appears.

I will also contact the support.

I did some testing: Different frame rates (24, 25), different sensor modes (4K "Full Sensor" & 2.6K "Windowed Sensor"), different ISO settings (400-25600). Results:
The frame rate and sensor mode (resolution) do not have an effect on it. But the ISO seems to have. It seems to appear only in ISO 1000 and below.

Support will be contacted tomorrow – I will post updates here.

That’s very strange that the ISO 3200 range is fine. Not the best solution but good you have that option working well.

I sent the camera to BMD. It will take 3-6 weeks until I have it back. Again, I'll post updates here.

Today I got an answer:

We believe the behaviour you are observing, only occurs with a certain type of scene where there is a high contrast and certain angles of light, and this is then exacerbated where exposure is below a level where the camera is able to make a useful picture. This is really creating a condition beyond the capabilities of the camera for filming, but the same behaviour would not be seen under normal exposure and recording conditions.

It is possible to expose noise and patterns in any camera sensor when it is pushed beyond its capabilities, and this is not something limited to our cameras. In this sort of scene, it would be possible to achieve the desired shots, but it may need to be approached differently.

For example, you may need to look at increasing the exposure in other areas through lighting, and look to use different ISO settings, such as using the second bank of ISO in a dual ISO camera like Pocket Cinema Camera 4k. Using a higher ISO in but in the second bank of ISOs, may produce lower noise than using a lower ISO that is at the top range of the first ISO bank.

In this example, ISO 1250 (2nd ISO bank) is almost certain to be less noisy than ISO 1000 (1st ISO bank) and we have seen that this completely eliminates the behaviour as it allows more data to be encoded below middle grey (the darker parts of the image).

For more information on this please see page 50 of the manual below:
https://documents.blackmagicdesign.com/ ... 5983611000

But if you look at an actual shot, I don't this it is too unusual.
Also I do not complain about the noise, but about the vertical seperation line in the right eighth of the frame. That's two different things, if I am not mistaken.

The first still is plain with LUT, just the blur. Second still is slightly graded. Third still has neat video desoiser + added resolve film grain. The seperation line on the right is visible in all three variations (Right click –> View Image for larger scale). Do I have to live with this from now on when I shoot @ ISO1000? I am sure this line isn't there in my previous shot footage and I shot against strong light in ISO1000 before. I mean, that's what ISO1000 is made for, it has the most stops of dynamic range above middle grey.

1-plain.jpg (707.37 KiB) Viewed 2417 times

2-grade.jpg (757.39 KiB) Viewed 2417 times

3-grade+denoise+grain.jpg (566.25 KiB) Viewed 2417 times

It may be frustrating that you can’t use ISO 1000 on that shot, but you must agree that the right side of the frame is quite underexposed and would be a better shot perhaps as an ISO 400 shot to darken the subject in silhouette or shot at ISO 1250 which doesn’t show the vertical line problem.

I never shoot ISO 1000 due to the noise (and I’ve never had that vertical line) and on the BMPCC4K, usually shoot either 400 or 3200 but in this shot it might be good to go with ISO 1250 and adjust highlights in post.

Sorry I haven’t got a solution for you, but suggest a workaround that will get your shot. And a recommendation simply to avoid ISO 1000 and anything above ISO 6400. You have probably seen images that go higher than a stop above the ‘native’ ISO in a range, but you can be sure they’re carefully selected and you’re not seeing the failures.

Same problem here since I got my P4K three years ago . Reported and send it to be serviced in the UK. Blackmagic was politely supportive but never solved the problem. In my case the only solution — which only disguises this anomaly — is to shoot at 1250, where in shadows to light transition situations. As soon as ISO is superior to 400 it only gets worst and nothing can be done. The same can be said at ISO superior to 1250, but less noticeable. If this is not an expected behaviour from the sensor, Blackmagic should have done better in supporting those who reported it.

I only just now read your post, must have slipped my attention before.

This is so unfortunate!! I use ISO1000 a lot due to its high dynamic range and advantage towards highlights.

Hi,

Rick is right.

Based on my own experience using the P4k camera...

1. A combination of Shadows and ISO 1000 is a danger zone.

2. 1000 is very usable when purely filming high mids to Highlight/whites

3. If you’re using BRAW, You can get the full Dynamic range with 100-1000 or 1250 - 6400. You don’t need to shoot the highest ISO. Use Zebra (90/95%) or false colour to monitor clipping.

4. Like what Rick said,400 (Bank 1) or 3200 (Bank 2) are good starting points. If exposure is good, I’d start at 200 or 1250 to keep noise gain low - but you may like the look of noise, so more gain is ideal.

5. Using fill to Bounce back light on to the subject can go a long way, especially with the large windows you have.

6. Maybe your pocket 4K can’t handle low light situations better than other pocket 4ks cameras.

I hope this helps

Thanks Daniel, I got some questions:

3. If you’re using BRAW, You can get the full Dynamic range with 100-1000 or 1250 - 6400. You don’t need to shoot the highest ISO. Use Zebra (90/95%) or false colour to monitor clipping.

You are basically saying, that if shooting BRAW, which I always do, I get the full dynamic range of either one of the ISO scales, independent of the ISO value within one scale, right? But how does this affect the tolerance towards shadows and highlights? If you may please look at this chart:

Screenshot 2021-02-17 at 13.26.01.png (171.91 KiB) Viewed 2280 times

You see that the the tolerance towards shadows and highlights (or let's say the dynamic range in relation to middle grey) changes depending on the selected ISO value within one ISO scale. How could BRAW possibly retain this?
Also, if you say that the selected ISO value isn't affecting the recording, the vertical seperation should also appear, if I shoot ISO400 and then brighten up the image in post, right?

I would do tests by myself to check, but my camera is still not back, yet.

EDIT: Wow, it has already been tested:

And impressively and luckily you are right, it doesn't matter in which ISO value within one given ISO scale I record. Great news!

5. Using fill to Bounce back light on to the subject can go a long way, especially with the large windows you have.

Yeah, but what if I want this high contrast? Is your suggestion, that I bounce to brighten the dark parts of the image and then in post I make then darker again? This seems very incovenient also I try do avoid "fixing things in post", naturally.

6. Maybe your pocket 4K can’t handle low light situations better than other pocket 4ks cameras.

I guess this is just sarcasm?

This video will explain it more elaborately

https://m.youtube.com/watch?v=hV7ixU2AmLE

As for contrast, I would first use a light meter to get an accurate Ratio. Since You care about the shadow area, I would blowout the window by using the lowest iso I can get and then use whatever necessary to fill in the shadows until you get the result you desire.


This will give you a more cleaner, dynamic image.

Of course this is one way of doing it, others could do it differently.

You can use micro adjustments in post - I wouldn’t call it fixing, I’d say: tightening the screws.

With Raw: More exposure to the sensor, the more information you capture.

This video will explain it more elaborately

https://m.youtube.com/watch?v=hV7ixU2AmLE

It doesn't as Gerald is only using one clip all the time on which he changes the ISO value in the Resolve Camera RAW settings and explains the functions of the Highlight Recovery checkbox. This has not much to to do with the tests from the video I posted, where the thesis is confirmed, that different recorded ISO values on different clips are indifferent in the recorded image's dynamic range when using the Resolve Camera RAW settings as ISO within one scale is merely metadata.

I found Gerald’s video to be very useful in understanding how to use lower ISO with extra highlight information when needed...

Not to drag it on, but looking at your shots you need to protect the shadow area, so a lower ISO would be ideal with this camera and use lights.

I found Gerald’s video to be very useful in understanding how to use lower ISO with extra highlight information when needed...

For that it is useful

Not to drag it on, but looking at your shots you need to protect the shadow area, so a lower ISO would be ideal with this camera and use lights.

I know what you mean; these are just shots from a casting without any lights, not from a an actual production.

The main point I want to make is that this vertical separation did not appear once in the entire footage of a feature film I shot with 95% natural/available light and a lot of shots which I exposed to ISO1000. Yes, there is noise. But I do not care so much because neat video does an excellent job and I add film grain later anyway. Noise is not the issue. The vertical separation is, which only just now appeared, and never before did.

Okay, I’ll be as direct as I can here so you can stop going round in circles

Some pocket cameras deal with lack of exposure better than other pocket cameras...

Your camera clearly reacts violently to lack of exposure when using High ISO.

If I was you, walk away from the camera if you can’t live with its personality. You’ll never be completely satisfied and always paranoid when you see Dark scenes without any artificial light at hand

Going by the their messages, Blackmagic UK won’t replace it.

Buy a new pocket 4K or try a different brand.

It is what it is.

6. Maybe your pocket 4K can’t handle low light situations better than other pocket 4ks cameras.

So this wasn't sarcasm, but instead you imply that there are differences between several BMPCC4Ks, which might be so huge that you see different low light behavior between them?

Still leaves the question unanswered, why the separation didn't appear before.

Conclusion
In the end Blackmagic Design re-evaluated the case.
They confirmed, that there is a reproducible vertical separation appearing on the sensor if you shoot in the ISO400 range with underexposed areas.
But this behavior is in the tolerance range of the camera's performance.
I have to live with it.

You have been very patient through this process and I really think you’ll be able to work around that limitation shooting ISO 3200 or ISO 1250 to which you may or may not require some additional noise reduction. Having that additional ISO range is a very powerful tool that even the UMP12K lacks.

From someone else picture of a (completely broken) bmpcc4k, it seemed obvious the sensor is actually split into 4 vertical stripes so my guess is that those can drift independently yielding a slightly different noise floor which becomes visible when digital gain is applied (such as using the 400 iso native and digitally boosting to 1000).

I would love to see the datasheet and conditioning stage schematic to learn more about all those stuff works on a real cinema camera, but ... no way I'd ever get to see those.

From someone else picture of a (completely broken) bmpcc4k, it seemed obvious the sensor is actually split into 4 vertical stripes so my guess is that those can drift independently yielding a slightly different noise floor which becomes visible when digital gain is applied (such as using the 400 iso native and digitally boosting to 1000).

I would love to see the datasheet and conditioning stage schematic to learn more about all those stuff works on a real cinema camera, but ... no way I'd ever get to see those.

Interesting information.

I'd like to add that the seperation also visibly appeared when I shot ISO400 and underexposed – not exclusively on ISO1000.