Welcome, Autel Pilots!
Join our free Autel drone community today!
Join Us

Why I NEVER use ND Filters With Drone Cameras

herein2021

Well-Known Member
Joined
Jan 26, 2021
Messages
592
Reaction score
568
Location
Florida
I can create the exact same effect with my Red Epic. In fact in certain situations I’ve done that because the effect works. Like shooting whitewater rapids, where a 1/4000 shutter creates a sparkly effect in the water that really works for that subject. It’s not a flaw, like HMI flicker, it’s a characteristic of high shutter speed footage. I don’t like that look for most subjects.

The NFL is really not relevant here. They’re shooting at higher shutter speeds and frame rates so that slo-mo playback will look better and their signal path includes devices capable of frame blending and interlace for 1080i playback.

I think part of the problem here is that no one has really stated their definition of strobing. Strobing to me is a rapid increase and then decrease in exposure. Using a higher shutter speed when combined with the rest of the proper settings of the exposure triangle will not affect the exposure.

Even the effect you just described would not be defined as strobing, the exposure did not change, there was simply no motion blur to make the transitions of the water drops as they fell look less than hyper realistic. I have agreed many times 100%....higher shutter speeds definitely reduce or eliminate motion blur that is a well known fact, but as long as the exposure remains constant which it will when using daylight lighting, then strobing will not occur.

If people simply do not like the reduced motion blur of higher shutter speeds then sure, that is a good use case to use an ND filter to lower the shutter speed until the motion blur that is required is achieved, but to state that "strobing" as defined by rapid changes in exposure is caused by high shutter speeds is still a false statement when using daylight as the light source and this is why there are no reputable articles and no one can provide a single link to a reputable source that states flickering or strobing is caused by high shutter speeds; hyperrealism sure, possible camcorder look sure....strobing/flickering....no.
 

AirStreamPictures

Active Member
Joined
Mar 1, 2021
Messages
38
Reaction score
14
Location
Maine
Website
airstream.pictures
I think part of the problem here is that no one has really stated their definition of strobing. Strobing to me is a rapid increase and then decrease in exposure. Using a higher shutter speed when combined with the rest of the proper settings of the exposure triangle will not affect the exposure.

Even the effect you just described would not be defined as strobing, the exposure did not change, there was simply no motion blur to make the transitions of the water drops as they fell look less than hyper realistic. I have agreed many times 100%....higher shutter speeds definitely reduce or eliminate motion blur that is a well known fact, but as long as the exposure remains constant which it will when using daylight lighting, then strobing will not occur.

If people simply do not like the reduced motion blur of higher shutter speeds then sure, that is a good use case to use an ND filter to lower the shutter speed until the motion blur that is required is achieved, but to state that "strobing" as defined by rapid changes in exposure is caused by high shutter speeds is still a false statement when using daylight as the light source and this is why there are no reputable articles and no one can provide a single link to a reputable source that states flickering or strobing is caused by high shutter speeds; hyperrealism sure, possible camcorder look sure....strobing/flickering....n
I think part of the problem here is that no one has really stated their definition of strobing. Strobing to me is a rapid increase and then decrease in exposure. Using a higher shutter speed when combined with the rest of the proper settings of the exposure triangle will not affect the exposure.

Even the effect you just described would not be defined as strobing, the exposure did not change, there was simply no motion blur to make the transitions of the water drops as they fell look less than hyper realistic. I have agreed many times 100%....higher shutter speeds definitely reduce or eliminate motion blur that is a well known fact, but as long as the exposure remains constant which it will when using daylight lighting, then strobing will not occur.

If people simply do not like the reduced motion blur of higher shutter speeds then sure, that is a good use case to use an ND filter to lower the shutter speed until the motion blur that is required is achieved, but to state that "strobing" as defined by rapid changes in exposure is caused by high shutter speeds is still a false statement when using daylight as the light source and this is why there are no reputable articles and no one can provide a single link to a reputable source that states flickering or strobing is caused by high shutter speeds; hyperrealism sure, possible camcorder look sure....strobing/flickering....no.
I’d say you’re right about that. (That part of the problem is semantic.) What you call strobing, I would call flicker. (As in HMI flicker or “FLICKERfree” lighting.) I call it strobing because it reminds me of the staccato motion you get with strobe lights in a dark room. Strobing is not changes in light level, it’s discontinuous light—light/no light/light… but it’s semantics, and as far as I know neither of us works for Webster’s, so I guess it doesn’t matter as long as we’ve figured out what the other is talking about.

It is a pretty common problem in photography forums these days. Terms like “banding,” “strobing,” and “flicker” aren’t very universally defined. But you see it all the time with terms like DOF where people just haven’t learned the basic terminology of the craft. (Not meaning you here at all.) Terminology causes a lot of confusion on the net.
 
Last edited:
  • Like
Reactions: herein2021

herein2021

Well-Known Member
Joined
Jan 26, 2021
Messages
592
Reaction score
568
Location
Florida
I’d say you’re right about that. (That part of the problem is semantic.) What you call strobing, I would call flicker. (As in HMI flicker or “FLICKERfree” lighting.) I call it strobing because it reminds me of the staccato motion you get with strobe lights in a dark room. Strobing is not changes in light level, it’s discontinuous light—light/no light/light… but it’s semantics, and as far as I know neither of us works for Webster’s, so I guess it doesn’t matter as long as we’ve figured out what the other is talking about.

It is a pretty common problem in photography forums these days. Terms like “banding,” “strobing,” and “flicker” aren’t very universally defined. But you see it all the time with terms like DOF and just haven’t learned the basic terminology of the craft. (Not meaning you here at all.) Terminology causes a lot of confusion on the net.

I definitely agree with you. so I think we went full circle and may have decided we are saying the same thing...that ND filters will lower the shutter speed which will add motion blur if desirable but if not desirable (or required) not using an ND filter will reduce motion blur to a state that some may find undesirable depending on the final shutter speed.

For the type of projects that I shoot I have never needed or used an ND filter with my drone footage; I've also never had a client complain that the footage did not have enough motion blur or that it had flicker/strobing/pulsing/stuttering but I am genuinely curious to see exactly what footage that people using ND filters would find undesirable if they did not use an ND filter. So the video below is one that I put together with nothing but high shutter speed footage. All of the footage in the video was shot at 4K30FPS 4:2:0 10bit with manual exposure and the ALOG profile. The shutter speed for the camera for every clip in the video ranged from 1/1000s - 1/2000s. I did not add any motion blur in post.

Keep in mind that YouTube macro blocking and compression artifacts will impact the footage in the video especially for the faster moving footage. This would have occurred regardless of what shutter speed was used.

 

Quad808

Well-Known Member
Joined
Dec 19, 2018
Messages
423
Reaction score
460
Age
109
Location
Hawaii
This has been a great thread. I am not an expert by any means, but have done commercial work as a photog and videographer. I know what works for me and what doesn't in those situations. I also know the danger of adding yet another piece of glass between the sensor and the subject matter. I just re-read the whole thread and very much appreciate the "experts" point of views based largely on their experience....that is always the best. What works for them will probably work for us "normal" people, as long as you have a good understanding of shutter speeds, ISO settings and aperture settings and how they all work together.
I will admit I starting using ND filters for the original reason that herein2021 started this thread with. It never made sense to me to do so, other than to keep the shutter speeds double the frame rate on drones that can't be set on full manual settings. The Evo II is not such a drone, and can be set on full manual settings.

I have some experimenting to do, and shoot drone footage for my own enjoyment only. I will bet for the area and type of video I shoot, using an ND filter as compared to not using one will result in video you can't tell apart. I don't care about motion blur one bit, and actually prefer super sharp video as compared to lots of motion blur video. Personal taste.
Exposure is always a balancing act with a moving drone filming different subject matter in the same shots.
Thanks all for taking the time to share your knowledge!
 

WilsonLR

Member
Joined
Aug 26, 2021
Messages
8
Reaction score
2
Age
65
Total drone noob here from video/photo world but looking at drones in photogrammetry application and took the plunge with an EVO II Pro V2.

This is a great thread. After reading it, I was all set to return the Freewell filters I had bought based on "some article I read on the internet". Then I came across this paper entitled
"The effect of Neutral Density Filters on drones orthomosaics classifications for land-use mapping"
that tested ND filters on a P4P specifically for the use case of making orthomosaics where the name of the game is capturing detail. FWIW
 
  • Like
Reactions: Rubik3x

tvwxman

Well-Known Member
Joined
Jul 23, 2020
Messages
94
Reaction score
107
Age
60
WOW! This is absolutely incredible!! Thanks for this post, simply amazing information... I'm sold!

Now if they would only do a similarly quantitative and scientific paper on comparing the sharpness of our drones with the various aperture settings!!!
 
Last edited:

tuxontodd

Well-Known Member
Joined
Sep 11, 2019
Messages
61
Reaction score
26
Age
67
If you have an EVO 1 ND filters are a must because of the props in your shot. It was horrible without them. Yes I know I can fly really slow, go backwards, or tilt the camera down but those are BS solutions for a bad design. I mostly use a ND16. If it gets really bright then a ND32.
 
  • Like
Reactions: tvwxman

The Autelian

Active Member
Joined
Feb 6, 2021
Messages
37
Reaction score
18
Age
63
I don’t care what Herein says but professional operators will always use ND filters! It’s a matter of simple physics. if you want professional results you have to use them to slow down the shutter speed. If you’re filming a friends wedding, then you can maybe get away with it.
End of story!
 
  • Like
Reactions: tvwxman

Rubik3x

Well-Known Member
Joined
Nov 4, 2018
Messages
313
Reaction score
187
Location
Washington State
So the video below is one that I put together with nothing but high shutter speed footage. All of the footage in the video was shot at 4K30FPS 4:2:0 10bit with manual exposure and the ALOG profile. The shutter speed for the camera for every clip in the video ranged from 1/1000s - 1/2000s. I did not add any motion blur in post.

Thanks for posting the video. I see what you mean that the fast shutter speeds do not interfere with pleasant videos at 30fps. Even the jet ski clips looked fine. However, the sped up sections looked very jerky to me, hard to watch. Also, many of the early clips looked over exposed. Was that intentional?
I also shoot 4K/30 LOG 10-bit and produce in 4K/30 10-bit 4.2.0 using DaVinci Resolve Studio. However, I usually use ND16 filter in bright sunny days. I've attached an eyebrow to the filter to help shade the front lens. When I tried shooting without that ND filter using a fast shutter, I got prop shadow flickering at certain angles to the sun. I'm wondering if its the eyebrow that prevents the prop shadow flickering or the slower shutter speed. More testing...
 

herein2021

Well-Known Member
Joined
Jan 26, 2021
Messages
592
Reaction score
568
Location
Florida
Total drone noob here from video/photo world but looking at drones in photogrammetry application and took the plunge with an EVO II Pro V2.

This is a great thread. After reading it, I was all set to return the Freewell filters I had bought based on "some article I read on the internet". Then I came across this paper entitled
"The effect of Neutral Density Filters on drones orthomosaics classifications for land-use mapping"
that tested ND filters on a P4P specifically for the use case of making orthomosaics where the name of the game is capturing detail. FWIW

I will caveat this entire statement by saying I know next to nothing about photogrammetry and obviously the paper is very detailed in many areas. But the part that makes no sense to me is that the shots without the ND filters and with the ND32 filter were completely out of focus/lacking detail. ND filters have nothing to do with improving sharpness, detail, or focusing and if they did this they would be well documented in many other areas of photography.

ND filters perform a single function; reduce the light reaching the camera lens which allows the camera operator to reduce the shutter speed or open the aperture. The most important key detail was the one left out within the Materials and Methods section; how they maintained identical exposure between the tests with and without ND filters. If they adjusted the Aperture to achieve that then everything would make sense; without the ND filters the aperture was stopped down to something like F11 which is not the sharpest aperture for the camera; at ND32 the camera was shot wide open at F2,8 which also explains why ND32 was so devoid of detail. At ND4, ND8, and ND16, they were probably somewhere around F4.0, F5.6, and F6.3 which are going to resolve more detail.

On a bright sunny day in direct sunlight pointing directly at the ground for a photogrammetry project, if you shot at F4.0,F5,6, or F6.3 you will always have enough available shutter speed remaining to properly expose the image without an ND filter. I suspect the authors decided instead to adjust the Aperture for their tests vs the shutter speed; without this key piece of information every other aspect of the test conclusions are highly suspect to me.

I am intrigued enough to discover if they made such a simple error in an otherwise very detailed analysis that I emailed the author requesting additional information on how they maintained identical exposure from test to test. If they simply did not know about the effects of lens distortion they could have very well made this mistake.

I don’t care what Herein says but professional operators will always use ND filters! It’s a matter of simple physics. if you want professional results you have to use them to slow down the shutter speed. If you’re filming a friends wedding, then you can maybe get away with it.
End of story!

The first rule of photography/video is there are no rules so you don't have to do anything and there's definitely no rules about always doing anything; an entire feature length film was shot on nothing but GoPros and I am certain many people like you said you must always use a cinema camera and no "professional" would use a GoPro to film a feature length movie yet it grossed $16.8M. If you shoot and deliver footage that your client is willing to pay for then you have met the client's requirements and you also fall into the category of professional; and if it is a matter of simple physics I would definitely be open to reading which law of physics states that the only way to achieve proper exposure for a given scene is to use an ND filter.

Thanks for posting the video. I see what you mean that the fast shutter speeds do not interfere with pleasant videos at 30fps. Even the jet ski clips looked fine. However, the sped up sections looked very jerky to me, hard to watch. Also, many of the early clips looked over exposed. Was that intentional?
I also shoot 4K/30 LOG 10-bit and produce in 4K/30 10-bit 4.2.0 using DaVinci Resolve Studio. However, I usually use ND16 filter in bright sunny days. I've attached an eyebrow to the filter to help shade the front lens. When I tried shooting without that ND filter using a fast shutter, I got prop shadow flickering at certain angles to the sun. I'm wondering if its the eyebrow that prevents the prop shadow flickering or the slower shutter speed. More testing...

Actually every clip was sped up in the video with the exception of maybe one or two. Due to the camera to subject distance of typical drone footage and the fact that I typically only use a few seconds of it in a project that otherwise has none of it; I speed up the footage to get more motion. I will admit I spent less than 20min on the sample video because the only intent was to demonstrate high shutter speed footage.

The portions that probably look jerky to you are the ones where there was already a lot of motion such as with the waves on the water and the flags. For a paying client I would have left those at their default speeds due to that very reason.

The early clips were not overexposed, a Rec709 WFM will show all of the footage was graded to fit within the Rec709 color space but since I only spent about 20min on the whole thing I did not focus on bringing down the lows or adding contrast which would have countered the highs and given it more DR. My only focus really was just showing high shutter speed footage in different scenarios.

The problem with drone cameras is that they not only do not have lens hoods, they also have props which can come between the sun and the lens at certain angles. An ND filter could lower the shutter speed to the point to where the shadow is such a blur that it is nearly imperceptible or the eyebrow may add just enough shade to do the same thing. My own personal solution is to simply angle the camera more towards the ground, or change the relationship between the drone and the sun.

I do think that's a very valid use case for a drone ND filter.....if you need to reduce prop shadow and do not want to change the relationship between the drone and the sun or angle the camera further down then the ND filter could reduce that shadow effect. I am not entirely convinced that it would completely remove it however and I would not risk it for a paying client; I'd either wait for clouds, wait for the sun to change positions, angle the camera down more, or change the way the drone is pointed at the subject material. The eyebrow on the other hand would definitely completely remove it if you manage to also keep the eyebrow out of the image while it is also somehow long enough to shade the entire lens.
 
  • Like
Reactions: tvwxman

Rubik3x

Well-Known Member
Joined
Nov 4, 2018
Messages
313
Reaction score
187
Location
Washington State
The problem with drone cameras is that they not only do not have lens hoods, they also have props which can come between the sun and the lens at certain angles. An ND filter could lower the shutter speed to the point to where the shadow is such a blur that it is nearly imperceptible or the eyebrow may add just enough shade to do the same thing. My own personal solution is to simply angle the camera more towards the ground, or change the relationship between the drone and the sun.

I do think that's a very valid use case for a drone ND filter.....if you need to reduce prop shadow and do not want to change the relationship between the drone and the sun or angle the camera further down then the ND filter could reduce that shadow effect. I am not entirely convinced that it would completely remove it however and I would not risk it for a paying client; I'd either wait for clouds, wait for the sun to change positions, angle the camera down more, or change the way the drone is pointed at the subject material. The eyebrow on the other hand would definitely completely remove it if you manage to also keep the eyebrow out of the image while it is also somehow long enough to shade the entire lens.
The eyebrow on my ND16 filter was fashioned out of a soda bottle, glued to the filter and trimmed for maximum coverage without showing up in the video image. I've forwarded this idea to Autel and Freewell. I was initially concerned that the extra weight on the ND filter would overload the gimbal. But, happily, the gimbal doesn't seem to care, even at ludicrous speed, perfectly steady. Here's a picture of it: DSC02998compr.JPG
 

herein2021

Well-Known Member
Joined
Jan 26, 2021
Messages
592
Reaction score
568
Location
Florida
The eyebrow on my ND16 filter was fashioned out of a soda bottle, glued to the filter and trimmed for maximum coverage without showing up in the video image. I've forwarded this idea to Autel and Freewell. I was initially concerned that the extra weight on the ND filter would overload the gimbal. But, happily, the gimbal doesn't seem to care, even at ludicrous speed, perfectly steady. Here's a picture of it: View attachment 11416

That is definitely an interesting modification, I am surprised that it is not visible in the shot and that it does not affect the gimbal motors especially in ludicrous mode. Definitely innovative to say the least; you could probably 3D print a batch and start selling them.

I have certainly encountered the prop shadow issue over the years but only once has it completely prevented me from getting a specific shot and that was when I was filming a waterfall in Hawaii that fell into a deep hole with obstructions on all sides...I couldn't angle the camera down because it would lose sight of the waterfall and I could not reposition the drone because there were walls on all sides. I also did not have the luxury of waiting for the sun to move across the sky so I ended up having to skip that particular shot.
 
  • Like
Reactions: Rubik3x

gschulzuio

Well-Known Member
Joined
Dec 14, 2020
Messages
260
Reaction score
116
I regularly do mapping missions with DD and a P4P for recurring project progress reporting. Also do some with Evo II and Explorer. I prefer doing mapping on cloud days to cut down on shadows, however reality is that you dont always get cloudy days.

Thus would have been interesting to see their tests done on a cloudy day for comparisons, as well as control point. That might also qualify their results as being positive for sunny days, not for cloudy or something else.

Also interesting that they indicated flying from 200m or above 600', that could result in about 2.2 in / px vs. say 1.1 in / px at 300'. They also dont indicate how fast the aircraft was traveling. Generally the higher is up, the aircraft can fly faster and is taking fewer images, meaning doing a mission quicker, granted resolution can be impacted. Otoh, flying lower gives more resolution, also more images, takes more time and batteries.

It would also be interesting to see what camera parameters they gave litchi to use (e.g. 16:9, 4:3, 3:2) among other settings.
 

herein2021

Well-Known Member
Joined
Jan 26, 2021
Messages
592
Reaction score
568
Location
Florida
I regularly do mapping missions with DD and a P4P for recurring project progress reporting. Also do some with Evo II and Explorer. I prefer doing mapping on cloud days to cut down on shadows, however reality is that you dont always get cloudy days.

Thus would have been interesting to see their tests done on a cloudy day for comparisons, as well as control point. That might also qualify their results as being positive for sunny days, not for cloudy or something else.

Also interesting that they indicated flying from 200m or above 600', that could result in about 2.2 in / px vs. say 1.1 in / px at 300'. They also dont indicate how fast the aircraft was traveling. Generally the higher is up, the aircraft can fly faster and is taking fewer images, meaning doing a mission quicker, granted resolution can be impacted. Otoh, flying lower gives more resolution, also more images, takes more time and batteries.

It would also be interesting to see what camera parameters they gave litchi to use (e.g. 16:9, 4:3, 3:2) among other settings.

Although very detailed in some areas, a lot was also left out in others; but everything indicates they made a simple error by adjusting the Aperture vs the Shutter speed to maintain exposure which of course would yield the exact results they received with or without the ND filters. If anything you want a higher shutter speed when doing something as detailed as mapping because the higher shutter speed would reduce motion blur on windy days.

The study even concluded that ND8 was the optimal ND filter strength for photogrammetry projects which ironically probably aligns almost exactly around F5.6 where lens distortion is probably the least for the P4P. It is surprising that even the authors at some point did not realize their results defied logic; if ND filters increased sharpness to the degree demonstrated in their study every camera would ship with a built-in set of ND filters to prevent the footage from being the blurry mess they demonstrated in their analysis of footage without ND filters. There are dedicated mapping drones available where it is not even possible to attach an ND filter; yet I am sure they deliver commercially viable footage for their owners/operators.
 
Last edited:
  • Like
Reactions: WilsonLR

Paulemus

Member
Joined
Sep 17, 2020
Messages
13
Reaction score
5
Age
71
For still photos, anyway, you want the fastest shutter speed to reduce any camera movement; so no ND filter.
 

parkgt

Well-Known Member
Joined
Feb 11, 2020
Messages
417
Reaction score
244
For still photos, anyway, you want the fastest shutter speed to reduce any camera movement; so no ND filter.
Wrong, sometimes motion blur is wanted in timelapses or long exposure shots of moving water. Fastest shutter speed would be obtained by always shooting at F2.8; yet 2.8 is not the sharpest aperture; so again no, don't always want the fastest shutter speed.
 

herein2021

Well-Known Member
Joined
Jan 26, 2021
Messages
592
Reaction score
568
Location
Florida
Wrong, sometimes motion blur is wanted in timelapses or long exposure shots of moving water. Fastest shutter speed would be obtained by always shooting at F2.8; yet 2.8 is not the sharpest aperture; so again no, don't always want the fastest shutter speed.

While I agree that the fastest shutter speed is not always necessary with drone photography, I disagree that you can achieve commercially viable timelapses or long exposure shots of moving water with current drone platforms.

I have tested this on a perfectly windless day, and the EVO II 6K starts noticeably losing sharpness at around 1/20s shutter speed due to it's inability to hold perfectly still in the air. To get good results of moving water you need around 1/8s or slower depending on how fast the water is moving. For water that is not moving much at all, you will need around 1s or more. Timelapses of course also need a perfectly still platform which once again current drone technology is unable to provide.

Now if you were referring to regular (non drone) cameras then certainly, ND filters combined with long exposures is pretty common for waterfalls, ocean shots, scenic views, city lights, etc.
 

Latest threads

Members online

Forum statistics

Threads
8,049
Messages
75,906
Members
5,925
Latest member
LCummings99