Jump to content

What's wrong with this motion blur


Heath Orchard

Recommended Posts

Hi Everyone, 

Would love to get your thoughts on this. Here's a screen grab from a short dance film I recently shot. Once it got into post the editor noticed there was a "choppiness" to the motion blur. My first thought was it was some sort of frame rate conversion happening unintentionally. But everything checks out there and it is also seen on the original clip. It's very noticeable (though only to me and the editor, no one else has noticed it) when the dancers are moving quickly. Sort of a trailing effect. I thought perhaps it was a bad LED source somewhere, but there is no flicker anywhere like I'd expect. My current hypothesis is that all the point sources may cause something like this but I would expect them to be blended together enough to play as a single source, especially since there weren't a hundred different shadows on the ground. 

Any ideas?

-Heath

dance film_PS.jpg

Link to comment
Share on other sites

If it's the LED's wouldn't I have been able to see flicker either through the camera or at least in playback? And when I refer to flicker I'm not referring to the weird choppy thing above but the flicker we all expect. I shot in this location for two nights and never saw a thing. 

Assuming it is LED flicker, what are the chances that a deflicker software would clean this up?

-Heath

 

Link to comment
Share on other sites

I’ve had this happen with mag ballast HMIs and other 60hz sources like sodium vapor etc. 

It’s 60hz lighting and 24fps so that's something like 2.5 discrete sine waves per frame or I don’t know the terminology. But the brightness (which I think correlates with area under the sine wave–chop up 60 sine waves into 24 pieces and I think the area under the wave would be consistent per sample regardless of where you start and end) is consistent frame per frame but there are 2.5 “samples” of motion blur per frame rather than a “smooth” blur like you’d get with a high frequency ballast or tungsten source (which is also pulsing but heat keeps it more consistent).

Which is to say there's triple-edged (or 2.5-edged) motion blur, but it's consistent frame-per-frame.

In theory a very very dim tungsten light might do this, too. I'm not sure if that's the case in practice, but this is why for HFR you want bigger tungsten units.

Anyway it usually looks fine to me.

If you’re worried about it, add some pixel motion blur or reelsmart motion blur but that will look more smooth in motion more like a 360º shutter angle. I would leave it as-is.

Edited by M Joel W
Link to comment
Share on other sites

  • Premium Member
1 hour ago, Heath Orchard said:

If it's the LED's wouldn't I have been able to see flicker either through the camera or at least in playback? And when I refer to flicker I'm not referring to the weird choppy thing above but the flicker we all expect. I shot in this location for two nights and never saw a thing. 

Think about it. If it's pulsing on and off a reasonably consistent number of times per frame, it won't manifest as a variation in brightness in the recorded image, you'll just get this. I've heard it referred to as "segmented motion blur." Yes, it can absolutely happen on film or digital.

(As an aside, if the camera and the lighting are slowly running out of phase with one another, you can sometimes see a slow dimming and brightening of mains-powered lights if you scrub through quickly on an NLE timeline. Shooting 24fps in a 50Hz region creates a four-second cycle with full-wave rectified lighting as well as segmented motion blur.)

This is more likely if you're shooting at a frame rate that's near to, or a clean multiple of, the local mains power frequency. LEDs sometimes do it because they're run from cheaply-converted mains power and flicker at some multiple of the frame rate. These days, though, all kinds of weird and wonderful approaches are used to drive LEDs and the results are really in the whim of whoever designed the electronics.

The situation M Joel describes is the same thing happening for the same reason, just with HMIs. 24 multiplied by 2.5 is 60, which might explain why there's almost three samples of motion blur there.

No, a deflicker filter won't fix it, because it's not manifesting as flicker. I'm not aware of any fix for this.

Link to comment
Share on other sites

  • Premium Member

it can become even weirder if one has flickering lights on different phases of the input power. it can fool flicker reduction and cause other interesting effects :) 

 

making a high speed pulse width modulation circuit to drive leds or tungsten lights is relatively simple. if using very high frequencies or high voltages it would become trickier but for example having a 12v light run at 50kHz or 100kHz at, say, 200w output is pretty easy to build from scratch (not that there would not be any eBay solution for this but just saying that it is easy to design a circuit like this from scratch if needed)

Link to comment
Share on other sites

Thanks everyone! This is super helpful and I feel like I generally get it now. I always thought this was more of an issue when shooting internationally, but now I see that its just LEDs continuing to be a blessing and a curse. I'm curious to know now if there could have been a way to prevent this. Could I have changed my shutter speed or some other camera setting to be more in sync with the pulsing light? We were essentially forced into the location the morning of our shoot and our entire creative had to change on the fly due to covid madness. That being said there was no real time for anything other than embracing the light that was there, which fortunately was quite beautiful aside from this issue. And the scene is a fantasy so the trailing effect can at least sort of be justified in the story. But really, nobody else seems to notice it and the film premiered tonight regardless. So it goes. 

Oh, and we were shooting on the Venice at 23.98 and 180 shutter if that informs anything. 

-Heath

Edited by Heath Orchard
Link to comment
Share on other sites

13 minutes ago, Heath Orchard said:

Thanks everyone! This is super helpful and I feel like I generally get it now. I always thought this was more of an issue when shooting internationally, but now I see that its just LEDs continuing to be a blessing and a curse. I'm curious to know now if there could have been a way to prevent this. Could I have changed my shutter speed or some other camera setting to be more in sync with the pulsing light? We were essentially forced into the location the morning of our shoot and our entire creative had to change on the fly due to covid madness. That being said there was no real time for anything other than embracing the light that was there, which fortunately was quite beautiful aside from this issue. And the scene is a fantasy so the trailing effect can at least sort of be justified in the story. But really, nobody else seems to notice it and the film premiered tonight regardless. So it goes. 

Oh, and we were shooting on the Venice at 23.98 and 180 shutter if that informs anything. 

-Heath

Sometimes its easier to deal with shutter angles than degrees .. at 180 degrees 23.98 your, shutter would be 1/48th ..  in the US the mains is 60hz..so the "safe" shutter is 1/60th  1/120th etc..ie  divisible by the mains 60hz..  in the UK the safe shutter speed would be 1/50th to match the 50hz mains ..  the difference is tiny and no one will notice any difference in motion blur .. neons this doesn't work .. always check but thats the rule of thumb for flicker / shutter ..  but I don't think thats your problem .. that would just give you a generally flicker over everything ..  this looks different to me ..  you see it in all forms of playback ?   

Link to comment
Share on other sites

We see it in all the playback now but we must have just missed it on the day. It was a pretty hectic shoot and we didn't have a DIT. It's on the original clips, in the timeline and now in the final piece. Not sure what other forms of playback there could be. It's a weird problem to have for sure. 

Link to comment
Share on other sites

3 minutes ago, Heath Orchard said:

We see it in all the playback now but we must have just missed it on the day. It was a pretty hectic shoot and we didn't have a DIT. It's on the original clips, in the timeline and now in the final piece. Not sure what other forms of playback there could be. It's a weird problem to have for sure. 

yeah .. if it doubt on locations with practical mains lights I always follow the "safe" shutter speed thing .. fluorescents are the usual culprits .. but yeah I don't think this is what is causing this funny frame thing personally..  .. 

Link to comment
Share on other sites

15 hours ago, Heath Orchard said:

If it's the LED's wouldn't I have been able to see flicker either through the camera or at least in playback?

 


Screens themselves have their own refresh rates, and I believe that they are not always the same as what is being shot.

E.g. if you're shooting at 100fps the monitor might only be displaying at 25Hz, and not showing you 3 out of every 4 frames.

It might be that the frames you are seeing are all the same brightness, so no flickering, but those frames in between are a different brightness, so when you playback slowed down to 25fps back at base it flickers.


Don't think that's what's happening here though.

 

Link to comment
Share on other sites

  • Premium Member

Since everyone has thrown out ideas that probably aren't the problem, how about this one. Maybe just maybe the camera was setup for a different imager frame rate than project frame rate. I know on the Venice you can force the camera to run a "variable" frame rate and retain the same project settings. So imagine the camera running at 60 for instance and trying to create a 23.98 file. That would cause a very strange motion blur effect. 

Edited by Tyler Purcell
Link to comment
Share on other sites

@Phil Rhodes is right it's the LEDs. It usually happens with those cheap string LED lights used for Christmas trees or outdoor decoration. In fact it's not just cameras that pick this up, but rather the human eye picks this up as well. For those who don't believe me, buy a set of LED string lights, set them up in a dark room, then wave your arm left and right in front of yourself and you will actually see this ghosting yourself in real time.

  • Like 1
Link to comment
Share on other sites

I have a contact at ReVisionFX who also feels it's the LED lights pulsing as Phil described. We're going to see if his software can fix this. Mostly just out of curiosity at this point. I'll let you all know how it goes. Thanks for all the feedback. 

-Heath

Link to comment
Share on other sites

  • Premium Member
7 hours ago, Tyler Purcell said:

Since everyone has thrown out ideas that probably aren't the problem, how about this one. Maybe just maybe the camera was setup for a different imager frame rate than project frame rate. I know on the Venice you can force the camera to run a "variable" frame rate and retain the same project settings. So imagine the camera running at 60 for instance and trying to create a 23.98 file. That would cause a very strange motion blur effect. 

I don't know any current camera which would do pulldown processing in this type of situation. They will just store the captured image frames as is an mark the metadata info differently to the file. So for example you will get the same frames shot at 100fps sensor framerate whether the camera stores them to a "50fps" file or a "24fps" file. The files differ in that the 24fps file is marked as 24fps and the 50fps file marked as 50fps so that one knows in playback how it is played "correctly". If you would change the metadata later to say that the clip is now "25fps" it would not change the captured frames in any way, the playback software is just instructed to play them back at different rate.

So the sensor framerate and shutter speed are important but the project framerate is not unless it affects monitoring or compressing options. For example in nature etc.documentaries I work with the sensor framerate can be anything high-ish which gives the best performance in that shooting situation with the subject. The project framerate can be anything depending on the camera model. I think it can be 8 or 10 project framerates in a movie like this. Then everything conformed to 24fps for editing and post. No frames or quality lost because it is only metadata which is changed and no one is touching or altering the actual frames in the image stream

Link to comment
Share on other sites

  • Premium Member

Don't know though if venice has internal pulldown option to ruin the image but it does not sound practical at all (because you would lose all the benefits of the framerate alteration if playing back the shot at normal speed doing in camera pulldown. That would be just a stupid design flaw and you should just alter the shutter angle instead to get the same effect without the pulldown artifacts)

Link to comment
Share on other sites

  • Premium Member
4 hours ago, aapo lettinen said:

Don't know though if venice has internal pulldown option to ruin the image but it does not sound practical at all (because you would lose all the benefits of the framerate alteration if playing back the shot at normal speed doing in camera pulldown. That would be just a stupid design flaw and you should just alter the shutter angle instead to get the same effect without the pulldown artifacts)

It's the only thing I can think of, considering it's not the LED's since there is no flicker in the shot. It has to be some sort of mismatch with the frame rate and its blurring along the way as a consequence. 

Link to comment
Share on other sites

  • Premium Member
2 hours ago, Tyler Purcell said:

It's the only thing I can think of, considering it's not the LED's since there is no flicker in the shot. It has to be some sort of mismatch with the frame rate and its blurring along the way as a consequence. 

If this stuff was shot at, say 24 frames, and it's a 60Hz region, there's a stable relationship between the LEDs and the camera, so the total illumination available to each frame is constant. We don't see flicker, but we do see the discontinuous illumination on moving objects. Various other phase relationships are possible between various lighting technologies and camera taking rates.

  • Like 1
Link to comment
Share on other sites

19 hours ago, Matthew J. Walker said:

@Phil Rhodes is right it's the LEDs. It usually happens with those cheap string LED lights used for Christmas trees or outdoor decoration. In fact it's not just cameras that pick this up, but rather the human eye picks this up as well. For those who don't believe me, buy a set of LED string lights, set them up in a dark room, then wave your arm left and right in front of yourself and you will actually see this ghosting yourself in real time.

But then wouldn't you see the source itself also flickering .. it would be over thew whole image the whole time ..? the OP says it only happens when they move quickly , maybe some shutter "thing".. ?

Link to comment
Share on other sites

11 minutes ago, Robin R Probyn said:

But then wouldn't you see the source itself also flickering .. it would be over thew whole image the whole time ..? the OP says it only happens when they move quickly , maybe some shutter "thing".. ?

You only see it when they move quickly, because when they move slowly, the discrete samples blend together, which is what's happening with the LEDs themselves.

I suspect if he whip panned over the LEDs they would show similar artifacts. I have had this exact same thing happen before with 60hz HMIs.

  • Upvote 2
Link to comment
Share on other sites

  • Premium Member
9 hours ago, Phil Rhodes said:

If this stuff was shot at, say 24 frames, and it's a 60Hz region, there's a stable relationship between the LEDs and the camera, so the total illumination available to each frame is constant. We don't see flicker, but we do see the discontinuous illumination on moving objects. Various other phase relationships are possible between various lighting technologies and camera taking rates.

Yea but everyone does this all the time in America. I shoot with shitty LED's all the time, Christmas tree lights, hand held's, flash lights, I've never once seen this phenomena unless there is something else wrong. I'm not completely discounting lights, but that's quite a bit of strobing, it would be very noticeable in other ways. 

  • Like 1
Link to comment
Share on other sites

9 hours ago, M Joel W said:

You only see it when they move quickly, because when they move slowly, the discrete samples blend together, which is what's happening with the LEDs themselves.

I suspect if he whip panned over the LEDs they would show similar artifacts. I have had this exact same thing happen before with 60hz HMIs.

So here's a couple stills pulled from the original clip that seem to support this. Notice the point sources read as sets of three as the camera is panning into position but once it settles for the take everything becomes one. But there is still no flickering anywhere. What's interesting is that we didn't see it on the day, which makes me think Mei Lewis has a point with the refresh rate of the monitor, which was a SmallHD 1703 P3x. 

Screen Shot 2021-01-16 at 3.33.28 PM.jpg

Screen Shot 2021-01-16 at 3.46.48 PM.jpg

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...