BMPCC: "Ghosting" with fast moving objects in low-light?

I don't know if this is normal, but I have never come across the same kind of ghosting (I don't know if that's the right word for it) than in a couple of BMPCC shots. I first noticed it in Auckland In My Pocket clip in the basketball scene and then in Wooden Camera's low-light test (fast moving car in the beginning). It kinda looks like the same with too few motion blur samples in 2D/3D animation (if you're familiar with it). I uploaded a closeup of the car's tire. Can someone give an answer from a technical point of view?

framegrab.jpg (84.41 KiB) Viewed 10983 times

Since that image was taken from a low-light shot, that may just be motion blur from using a high shutter angle (e.g. 360 degrees). Doing so will provide the sensor with more light, but you'll also get a lot of motion blur. It's not ideal, but if you're already at 1600 ISO and don't have any additional lighting available, sometimes it's the only solution to prevent underexposure.

See http://www.red.com/learn/red-101/shutter-angle-tutorial for more info on shutter angles / shutter speeds.

Since that image was taken from a low-light shot, that may just be motion blur from using a high shutter angle (e.g. 360 degrees). Doing so will provide the sensor with more light, but you'll also get a lot of motion blur. It's not ideal, but if you're already at 1600 ISO and don't have any additional lighting available, sometimes it's the only solution to prevent underexposure.

See http://www.red.com/learn/red-101/shutter-angle-tutorial for more info on shutter angles / shutter speeds.

I would also add that it's not only the long shutter speed but also the pulsing from the light sources.

Wooden Camera says that they shot the scene with 180. I understand the shutter angle terminology, but I don't see how it explains that stepping in motion blur. Shouldn't the motion blur be "smooth"? I watched John Brawley's Vivid Pocket and didn't notice the same kind of effect. Maybe it's because of artificial lights pulsing in the wrong frequency or some filter creating reflections?

Oddly enough, that looks just like the kind of fake stepped motion blur you get in software like nuke or Maya with a low sample setting.

Exactly!

You sure its not your NLE adding a kind of frame blending?

Some add it automatically.

Here is an example from Auckland In My Pocket. It's like three hands layered together. I'm watching the footage from Quicktime Player, After Effects and Vimeo. Two different computers (though Macs both).

And that framegrab is from Vimeo.

Vimeo does process and recompress video before displaying it on the site. Not just that, I'm sure the upload format was h.264 and perhaps frame reordering or blending was enabled. I'd be more worried if this was from sample footage directly from the camera (which I don't think it is at all).

I'm seeing that same effect from raw Prores in After Effects, frame blending off... I hope there's something wrong with my computer.

I noticed this on the Auckland footage and also in the original pro Rez footage I downloaded for the wooden camera website. It's deffo on the footage! So far I'm seeing too many issues with the pocket footage and am gonna pay the extra and get the BMCC instead!

I checked the footage on my PC and it's the same. I was wondering if Wooden Camera made some retiming, but there's no mention on their web site, so I think it's straight from the camera. It's funny that there's no issue in John Brawley's low-light scenes even though you're pixel-peeping fast-moving objects the entire clip. I'd like to get some information about the issue straight from Blackmagic before going for the BMCC instead...

Magnified frame grab from New Zeland footage, natural light. No ghosting in motion blurred areas.

B01_2013-07-18_1602_C0000 (1443350).jpg (133.74 KiB) Viewed 10838 times

This is from NZ footage, Magnified. Natural light. No "ghosting" in motion blurred areas.

I'm also not feeling the amount of grain the pocket seems to be pumping out even in well lit scenes.

From woodencamera footage. Daylight

wc_IR_test_skate (108437).jpg (200.42 KiB) Viewed 10822 times

Same from wooden camera footage, daylight. No ghosting/pulsing

Natural well-lit is ok, but artificial low-light is not (in fast motion).

Magnified woodencamera footage.

wc_low_light (87024).jpg (215.69 KiB) Viewed 10812 times

In the same scene, ghosting and no ghosting. Conclusion: camera is ok. Behaves just like any other camera in similar situation.
Lightsource is important and affects appearance of light pulsing, motion blurred areas ghosting.

Is it because street lights pulse differently than "other lights"? I think the question is answered here, so no problem with the camera... But is the effect worse than in other DSLRs/BMCC/etc?

This effect should be noticeable in most cameras, when shooting the same framerate and shutter angle. It's obviously not a faulty sensor or some kind of in-camera-denoising but the stroboscopic effect of pulsating light sources. Shooting the same scene on the CC 2k5 through the same lens should yield identical effects; maybe cameras with lower DR won't show it that much because the pulsating lights will only change from "clipping" to "really clipping", which would both look just white.

[CaptainHook]

I'll take a closer look at the original files for the basketball stuff later tonight. I can share them too if people want. Pretty sure it's the strobing from the lights at the park but maybe i should try test at night with a different light source. Did you notice any problems with the night stuff around the churros etc? This was just up the road from the basketball on the same night, but different light sources i think.

I was definitely at 25fps/180 degrees and i believe iso400 for the basketball stuff.

I'll take a closer look at the original files for the basketball stuff later tonight. I can share them too if people want. Pretty sure it's the strobing from the lights at the park but maybe i should try test at night with a different light source. Did you notice any problems with the night stuff around the churros etc? This was just up the road from the basketball on the same night, but different light sources i think.

I was definitely at 25fps/180 degrees and i believe iso400 for the basketball stuff.

I don't think there was anything wrong with your shooting mr hook. How do you find the grain on the pocket. The pro res footage from it looks so grainy. A little worrying.

[CaptainHook]

I don't find it grainy but maybe that's preference. Are you referring to the prores original files i uploaded, or the graded vimeo file, or both?

I don't find it grainy but maybe that's preference. Are you referring to the prores original files i uploaded, or the graded vimeo file, or both?

The original pro Rez, but to be fair I didn't see it as much on yours. More JBs and wooden cameras.

I'm also not feeling the amount of grain the pocket seems to be pumping out even in well lit scenes.

Seriously waiting for RAW for this camera (and for the camera to actually be available) The prores Ive seen so far just doesnt do it, like you said, spend another g for a proper BMCC, sell the included Resolve, and have a superior 2nd or third cam for not much more $

The prores Ive seen so far just doesnt do it, like you said, spend another g for a proper BMCC, sell the included Resolve, and have a superior 2nd or third cam for not much more $

I don't believe there's a difference in dynamic range or color science between the Pocket CC and the older CC 2k5, as they share the same sensor. Oh, but wait - i remember you don't care what i believe.

It seems odd that so many distinctive "pulses" can be seen in some of the frames, if this was caused solely from artificial light then at most you should expect just about 3 distinctive copies. Our mains frequency is 50Hz here in New Zealand so during a single frame exposure at 25fps, 180 degrees shutter angle the worst case scenario is 3 full intensity peaks and 2 zero crossover points depending on the lamp technology. Depending on the speed of motion (object and camera), focal length and distance from camera this can definitely become visible. More pulses than 3 is only possible if AC lighting is out of phase with other lights in the scene, or if the distance the artificial light is far enough that the lights travel time has the same effect as a phase shift but that is negligible here. Sodium vapour lamps and fluorescents are different though and are regulated by a ballast but they do flicker, I'm sure many of you have experience at shooting under fluorescent lights at shutter speeds that don't match the mains frequency.

In the cropped frames posted above, it can be seen that there is 3 distinct pulses in the basketball frame. Here there is only or two one major lights at the scene. I know this location in Auckland and the intensity and distance of other lights in the area do not have enough effect to be seen. Those 3 pulses "ghosts" could be explained by the stuff mentioned above.

In the car frame it is likely that there are more lights in the scene and there is more opportunity out of phase lights in different lighting circuits. A phase shift can occur between these due to different inductive loads on the transformers because of the number of lights on a string. From what I can see, there are approximately 5 "ghosts" of the car. This is likely because of AC lighting again, a combination of one light approximately 90 degrees out of phase with the shutter and another near 90 degrees further.

In the US this is less likely at a grid frequency of 60Hz or perhaps it'll look a little more like motion blur instead.

Given the right information some simple calculations could be performed, and if you know the distance between each of the ghosts my guess is that it would match up perfectly with a pulse frequency at 100Hz

It seems odd that so many distinctive "pulses" can be seen in some of the frames, if this was caused solely from artificial light then at most you should expect just about 3 distinctive copies. Our mains frequency is 50Hz here in New Zealand so during a single frame exposure at 25fps, 180 degrees shutter angle the worst case scenario is 3 full intensity peaks and 2 zero crossover points depending on the lamp technology. Depending on the speed of motion (object and camera), focal length and distance from camera this can definitely become visible. More pulses than 3 is only possible if AC lighting is out of phase with other lights in the scene, or if the distance the artificial light is far enough that the lights travel time has the same effect as a phase shift but that is negligible here. Sodium vapour lamps and fluorescents are different though and are regulated by a ballast but they do flicker, I'm sure many of you have experience at shooting under fluorescent lights at shutter speeds that don't match the mains frequency.

In the cropped frames posted above, it can be seen that there is 3 distinct pulses in the basketball frame. Here there is only or two one major lights at the scene. I know this location in Auckland and the intensity and distance of other lights in the area do not have enough effect to be seen. Those 3 pulses "ghosts" could be explained by the stuff mentioned above.

In the car frame it is likely that there are more lights in the scene and there is more opportunity out of phase lights in different lighting circuits. A phase shift can occur between these due to different inductive loads on the transformers because of the number of lights on a string. From what I can see, there are approximately 5 "ghosts" of the car. This is likely because of AC lighting again, a combination of one light approximately 90 degrees out of phase with the shutter and another near 90 degrees further.

In the US this is less likely at a grid frequency of 60Hz or perhaps it'll look a little more like motion blur instead.

Given the right information some simple calculations could be performed, and if you know the distance between each of the ghosts my guess is that it would match up perfectly with a pulse frequency at 100Hz

Hello anonymous (real name forum here), I am curious about your math. I am not very good at it, could you explain more, how you figured out the pulses/crossover frequency?

Hello anonymous (real name forum here), I am curious about your math. I am not very good at it, could you explain more, how you figured out the pulses/crossover frequency?

Sorry I forgot about the rules here.

I have made some pictures to show how this works. Check them out here: https://app.box.com/s/zwhf4ktw244bvff6a646

One is for 50Hz supply frequency and the other for 60Hz, the same principles apply for both all that has changed is the timing. The vertical axis represents magnitude or light intensity (not necessarily directly! Dependent on type of bulb), ignore the negative and positive amplitudes as it makes no difference here. I thought about adding a duplicate waveform but the picture gets a little too busy but if you request it I will add it to show you the effect of out of phase lighting that can generate more pulses. Just imagine the waveform there duplicated and shifted 90 degrees to the right and overlayed on top and you will get an idea of how 5 pulses can be generated in 1/50th of a second. And that's only from 2 lights out of phase! adding another can produce another 2, and another can produce another 2 but eventually the pulses will turn into a motion blur and you cannot recognize any pulsing light. The chances of that many lights being out of phase is not likely but when you consider industrial buildings or even city blocks drawing a lot of power they can be out of sync by a surprising amount. There are so many factors that come into play that it becomes too complicated to explain, but something as simple as a large water pump underground can put enough inductive load on the local grid that the phase of a circuit will shift, and if there are lights on it they will also follow.

It's all very simplified but it gives one possible explanation...

That seems to make sense. Thank you for the explanation, Joel!