I carried out an experiment to find the limits of motion blur with a Mavic Pro in bright sunlight. I had sunlit sea behind me and scrubland in the oposite direction.
I discovered the ND 8 & ND 16 filters I had failed to reduce exposure enough with 24fps/1/50th shutter/100 ISO. I was unable to compare the blur reduction as a result.
However, the test shows conclusively that if the drone is further away from the moving subject than 22 metres, (Paces) then ND filters are not needed.
It raises the question, how often do you video with a drone that close? Motion blur was negligeble at just 12 metres for most situations.
I then questioned the practicality of an ND filter for use on a drone, simply because you cannot fit a different one during flight.
To answer that question, I noted the shutter speed I needed for correct exposure at 24fps by rotating the drone.
I needed 1/1600 towards the scrubland, 1/2500 at 90 degrees and 1/8000 pointing at the sea. Not into the direct sun.
This suggested I would need to change the ND filter at least 3 times if filming those typical extremes.
I saw a video on here that used an ND filter during the flight over land and water. It was a simple circuit. To control the exposure, in-flight adjustments were made. I presume these would be shutter speed. If so, what was the purpose of the ND filter? the shutter speed would not remain 24fps, would it?
In addition, the videographer claimed that the ND filter improved the dynamic range of the vid. I though the colouring was acceptable though slightly under exposed and saturated. Could that have been the result of using an ND filter? If so, the same result would have been achieved with TrueColour and slight under exposure. Certainly the pic looked more normal as the exposure was optimum, before the shutter speed was increased.
I also believed that an ND filter would hav no evvect on dynamic range. A polarising filter set at the right angle to the sun would do that. What do you think?
Perhaps you experts will be able to clarify my conclusions.
Video link to show the test.
PS. Please do not comment on the over exposure of the vid. I started taking the test facing the other way and ran out of distance. I just corrected for the land part of the scene.
Thanks for making and sharing this.
What I am curious about is the same BUT now when the drone moves.
Not only 0.1% of the frame (your arm)
Most of the motion will come from the drone moving (thus the whole frame moving).
Like when you fly rather slow and make both forward and turning motion at the same time.
I'm curious at what speed you will notice the stuttering.
Montfrooij Posted at 2018-4-5 10:00
Thanks for making and sharing this.
What I am curious about is the same BUT now when the drone moves.
Not only 0.1% of the frame (your arm)
Great points - that is precisely when you notice it the most - yaw and pitch of the drone.
To be fair the whole 180 degree rule is a bit OTT on consumer grade drones in my experience but very fast shutter speeds are noticeable. I would say I tend to get stuttering at speeds of 1/250th or higher and generally below 100' AGL - the more detail the more pronounced so rail lines, roof tops, buildings when yawing the drone. Light and air speed also play a big part too.
A CW Posted at 2018-4-5 10:13
To be fair the whole 180 degree rule is a bit OTT on consumer grade drones in my experience but very fast shutter speeds are noticeable. I would say I tend to get stuttering at speeds of 1/250th or higher and generally below 100' AGL - the more detail the more pronounced so rail lines, roof tops, buildings when yawing the drone. Light also plays a big part too.
That last one makes sense.
The more contrast, the more sharpness to get stuttering.
A CW Posted at 2018-4-5 10:13
To be fair the whole 180 degree rule is a bit OTT on consumer grade drones in my experience but very fast shutter speeds are noticeable. I would say I tend to get stuttering at speeds of 1/250th or higher and generally below 100' AGL - the more detail the more pronounced so rail lines, roof tops, buildings when yawing the drone. Light also plays a big part too.
Exactly - if recording at 30fps you will notice hardly any difference between setting the shutter speed at 1/60th or 1/80th. That said, if you need to set your shutter to 1/1000th you will notice it and that is when ND filters are important to bring down the shutter without over exposing the image. The more control you have over the footage the better.
Montfrooij Posted at 2018-4-5 10:00
Thanks for making and sharing this.
What I am curious about is the same BUT now when the drone moves.
Not only 0.1% of the frame (your arm)
I have noticed stuttering with my vids when running from an SD card (I don't do that anymore) and with VLC, currently (I use the win10 video thing) so that is cured. I have also seen brilliant drone vids by experts that stutter, but I put that down to streaming from YouTube.
None of my humble vids stutter at 4K from the hard drive.
May I suggest you take a peek at a test vid I did when I tried different scene shots to see how the look. It is not intended to be a proper video for the likes of you guys, but it doesn't stutter on YouTube and has some very close moving scenes that defy the nd filter argument about the frame movement rather than arm waving.
AlansDronePics Posted at 2018-4-5 10:20
I have noticed stuttering with my vids when running from an SD card (I don't do that anymore) and with VLC, currently (I use the win10 video thing) so that is cured. I have also seen brilliant drone vids by experts that stutter, but I put that down to streaming from YouTube.
None of my humble vids stutter at 4K from the hard drive.
May I suggest you take a peek at a test vid I did when I tried different scene shots to see how the look. It is not intended to be a proper video for the likes of you guys, but it doesn't stutter on YouTube and has some very close moving scenes that defy the nd filter argument about the frame movement rather than arm waving.
Thanks for sharing this.
In some shots you do see some motion stutter. Mainly when turning the drone around.
Like in 1:12
Also the background in the opening shot has a little (but the foreground won't move fast enough to stutter). .
So indeed. (my best guess is you did not have the slow shutterspeed) it is only noticeable when the movement is fast enough.
A CW Posted at 2018-4-5 10:20
Exactly - if recording at 30fps you will notice hardly any difference between setting the shutter speed at 1/60th or 1/80th. That said, if you need to set your shutter to 1/1000th you will notice it and that is when ND filters are important to bring down the shutter without over exposing the image. The more control you have over the footage the better.
Yes, unless your movement is not too fast.
That is my assumption at least. I have not tested it. (I can't....)
I tried that vid on youtube before suggesting it here. There is no stutter on the original at the points you mention. If there was a place for stuttering it would be the very low level shot over the grass and quarry edge. Even that is stutter free. The setting were auto with iso 100. Your guess is as good as mine on shutter speed.
Montfrooij Posted at 2018-4-5 10:42
Yes, unless your movement is not too fast.
That is my assumption at least. I have not tested it. (I can't....)
To be honest I look at how bright the sky is, throw a filter on to compensate and take off then manually adjust my settings and simply fly the thing. Works for me.
Won't be long until you get another drone mate.
A CW Posted at 2018-4-5 10:50
To be honest I look at how bright the sky is, throw a filter on to compensate and take off then manually adjust my settings and simply fly the thing. Works for me.
Won't be long until you get another drone mate.
Indeed.
Although I am not to hopefull about this.
Gave him the task of repairing it 'good'.
So no quick fix, but a complete overhaul of these roofs.....
Better to get it done good now and be ok for a long time.....
Montfrooij Posted at 2018-4-5 11:49
Very true.
I am also renovating some old shutters from our house.
A lot of work, but very therapeutic to do beside my day job
AlansDronePics Posted at 2018-4-5 10:20
I have noticed stuttering with my vids when running from an SD card (I don't do that anymore) and with VLC, currently (I use the win10 video thing) so that is cured. I have also seen brilliant drone vids by experts that stutter, but I put that down to streaming from YouTube.
None of my humble vids stutter at 4K from the hard drive.
May I suggest you take a peek at a test vid I did when I tried different scene shots to see how the look. It is not intended to be a proper video for the likes of you guys, but it doesn't stutter on YouTube and has some very close moving scenes that defy the nd filter argument about the frame movement rather than arm waving.
To me this is excellent quality footage how I wish mine to look like!
I thought the Reservoir video was good.
Only saw a blur when the speed picked up or the turns were sharp. As long as you flew slower the blur was barely noticeable ,if at all.
AlansDronePics Posted at 2018-4-5 10:20
I have noticed stuttering with my vids when running from an SD card (I don't do that anymore) and with VLC, currently (I use the win10 video thing) so that is cured. I have also seen brilliant drone vids by experts that stutter, but I put that down to streaming from YouTube.
None of my humble vids stutter at 4K from the hard drive.
May I suggest you take a peek at a test vid I did when I tried different scene shots to see how the look. It is not intended to be a proper video for the likes of you guys, but it doesn't stutter on YouTube and has some very close moving scenes that defy the nd filter argument about the frame movement rather than arm waving.
No one has yet mentioned the other questions I posed in my OP.
1/ I then questioned the practicality of an ND filter for use on a drone, simply because you cannot fit a different one during flight...
2/ To answer that question, I noted the shutter speed I needed for correct exposure at 24fps by rotating the drone.
I needed 1/1600 towards the scrubland, 1/2500 at 90 degrees and 1/8000 pointing at the sea. Not into the direct sun.
This suggested I would need to change the ND filter at least 3 times if filming those typical extremes.
3/ I saw a video on here that used an ND filter during the flight over land and water. It was a simple circuit. To control the exposure, in-flight adjustments were made. I presume these would be shutter speed. If so, what was the purpose of the ND filter? the shutter speed would not remain 24fps, would it?
4/ In addition, the videographer claimed that the ND filter improved the dynamic range of the vid.
I would like to hear views on these unanswered points.
No ND filter on a drone will provide 100% perfect exposure/shutter speed correlation when the drone is orientated to and from a bright sun and recording over reflective surfaces such as large bodies of water in the same flight - which is obvious as the dynamic range will clearly change as the drone is moving.
The key is to use an ND filter that, worked in conjunction with gimbal tilts and yaw will at least allow more control over the shutter speed/motion blur. If, upon turning on my drone I need a shutter speed of 1/800th to create an exposure value of -.7 to zero then placing an ND8 on the lens will enable me to reduce the shutter to 1/60th to double my frame rate and create noticeable motion blur in the footage.
If I turn toward the sun and my footage becomes over exposed then I will need to pitch the gimbal down to compensate by removing as much of the light from hitting the sensor directly in that shoot or increase the shutter speed but having my shutter at 1/100th is still more effective than 1/1000th without the using the filter. On my next flight I may place an ND16 or ND32 on the lens to increase the stop of light hitting the sensor further and do the shoot again if I specifically need to capture something facing that direction on my next flight.
Yes, ND filters are important when using drones with fixed apertures in bright conditions.
Stills are not the best way to demonstrate ND filters as obviously no motion blur but to put the exposure value alone into perspective the top image was without a filter and the bottom (same day) was taken with an ND32 - on the first shoot I could not yaw my drone anywhere near the direction of the sun unless I massively increased the shutter speed and it looked terrible - I clearly could on the second flight with the ND32 applied and reduced my shutter speed from 1/2000th to 1/100th - that reduced the stuttering at these altitudes in my videos and as for the stills, well - you can decide which one you prefer!
A CW Posted at 2018-4-6 02:42
Stills are not the best way to demonstrate ND filters as obviously no motion blur but to put the exposure value alone into perspective the top image was without a filter and the bottom (same day) was taken with an ND32 - on the first shoot I could not yaw my drone anywhere near the direction of the sun unless I massively increased the shutter speed and it looked terrible - I clearly could on the second flight with the ND32 applied and reduced my shutter speed from 1/2000th to 1/100th - that reduced the stuttering at these altitudes in my videos and as for the stills, well - you can decide which one you prefer!
Hi A CW,
I hear what you say with the previous reply and respect your opinion.
Regarding this reply, I particularly liked the second picture, perhaps a little dark for my taste, but that is easy to say without the original to tweak with. What I am trying to say, this might be the ultimate setting, all circumstances considered.
Digressing from nd filters for a moment, sometimes you need to be flying into the sun and tilting the camera down more is an effective solution. You are the first person to have mentioned it in all the posts and comment I have seen. Obviously, I haven't seen them all...
Thank you for your insight.
Dynamic range is a specific term for the ratio between the strongest signal and the weakest signal in a specific frame of data. In videography and photographic terms that usually means how much chromatic detail is resolved from the brightest to the darkest portions of a frame. In video capture we also have the notional higher data rate of frames per second, which the human eye integrates and perceives 'more' picture information - as compared to a static still. The use of an ND filter alone will only drop the exposure value of the incoming light on the camera sensor, it will not change dynamic range as such. However allowing a slightly longer capture time per frame due to a slower shutter speed will allow slower photons to build up in the photoreceptors just that bit more - which if properly exposed should just a tad reduce high frequency noise from each photocell (it's an integrative process). Using a combined ND and PL filter will 'enhance' the chromatic contrast in a scene by cutting to a degree random light reflections which 'muddy' the light. So in the later case there is a perceived increase in dynamic range but in fact it is just chromatic contrast. In the same way more detail can appear to be in a frame just because we are seeing more high frequency contrast changes across an image. For myself I'm using an ND32PL all the time as the sunlight is now bright enough .... Merry O'l England and all...!
I want to clarify that what you call 'motion blur' in your video is in fact 'motion stutter' caused by the LACK of motion blur. ND filters would allow you to expose the image longer thus creating the pleasant motion blur that smoothens the 'transition' between frames for your eyes.
It's true that the need for such blur is not immediately evident in slowly moving shots, but even if you cannot rationally tell for sure when motion blur is missing, a shot is perceived as more pleasant if it is present. Also, mecanical shutters traditionally covered 180 degrees in cinema, so shutter speeds close to that (double the framerate) help give video a more classic look. A bit like how 24 frames per second looks way more cinematic than 50 or 60 despite not being technically superior.
You're right in that if you want to stick to the 180º rule, your options for controlling exposure are very limited (because of the fixed aperture and the bad performance of high ISOs on the Mavic). I usually do very short flights with one or two shots in mind, so that's not a big problem for me. I can measure the light and choose the right ND value between shots, but of course everyone has their own workflow and style.
@Lamplighter55 gave some good insight on the dynamic range topic. I don't use polarisers all the time unless I know which reflection or light source I want them to affect, as I find them a bit unpredictable. (and they add a layer to the filter, so there's more risk of losing sharpness, creating reflections or imperfections....)
0Kajuna0 Posted at 2018-4-6 06:35
I want to clarify that what you call 'motion blur' in your video is in fact 'motion stutter' caused by the LACK of motion blur. ND filters would allow you to expose the image longer thus creating the pleasant motion blur that smoothens the 'transition' between frames for your eyes.
It's true that the need for such blur is not immediately evident in slowly moving shots, but even if you cannot rationally tell for sure when motion blur is missing, a shot is perceived as more pleasant if it is present. Also, mecanical shutters traditionally covered 180 degrees in cinema, so shutter speeds close to that (double the framerate) help give video a more classic look. A bit like how 24 frames per second looks way more cinematic than 50 or 60 despite not being technically superior.
Thank you for the explanation. Something to think about.
0Kajuna0 Posted at 2018-4-6 06:35
I want to clarify that what you call 'motion blur' in your video is in fact 'motion stutter' caused by the LACK of motion blur. ND filters would allow you to expose the image longer thus creating the pleasant motion blur that smoothens the 'transition' between frames for your eyes.
It's true that the need for such blur is not immediately evident in slowly moving shots, but even if you cannot rationally tell for sure when motion blur is missing, a shot is perceived as more pleasant if it is present. Also, mecanical shutters traditionally covered 180 degrees in cinema, so shutter speeds close to that (double the framerate) help give video a more classic look. A bit like how 24 frames per second looks way more cinematic than 50 or 60 despite not being technically superior.
0Kajuna0 Posted at 2018-4-6 06:35
I want to clarify that what you call 'motion blur' in your video is in fact 'motion stutter' caused by the LACK of motion blur. ND filters would allow you to expose the image longer thus creating the pleasant motion blur that smoothens the 'transition' between frames for your eyes.
It's true that the need for such blur is not immediately evident in slowly moving shots, but even if you cannot rationally tell for sure when motion blur is missing, a shot is perceived as more pleasant if it is present. Also, mecanical shutters traditionally covered 180 degrees in cinema, so shutter speeds close to that (double the framerate) help give video a more classic look. A bit like how 24 frames per second looks way more cinematic than 50 or 60 despite not being technically superior.
I want to clarify that what you call 'motion blur' in your video is in fact 'motion stutter' caused by the LACK of motion blur. ND filters would allow you to expose the image longer thus creating the pleasant motion blur that smoothens the 'transition' between frames for your eyes.
Thank you for bringing that up. I thought I must be confused for awhile and just didn't think he had his hat on straight.
Like someone else said, I'm usually running ND32 here, and have been tempted to put the 32PL on to see how it does. There IS the problem of scene choosing and using the right polarization for each scene though.