What speaks against recording every video at 60fps vs 30fps with a drone?

[brett8883]

Keep in mind that 60fps reduces and amount of data for every available frame in comparison to 30fps

What are you basing this on? Even if this is the case I don’t believe that’s how it works. At 60 FPS the image in each frame changes less from frame to frame so it should be easier to compress each frame without losing visual quality.

 

[TerryToast]

How are you coming up with 4k 30fps is 100Mbps? All I see in the specs is that it’s max 120 Mbps.

More DJI Mavic Air 2 camera problems

From limited resolution in Quickshots to the lack of autobracketing in 48MP photo mode, the DJI Mavic Air 2 has some frustrating camera problems.

dronedj.com

 

[swoermck]

One advantage of 60 fps is the smoothness of the video. Its more like viewing it with your own eyes for motion. 30fps is what most of us are used to, but you start seeing jerkiness when moving faster.

Exactly the reason why I'm planning to get a MA 2 as well - the smooth video, as quoted above ".. like viewing it with your own eyes."

With my Mavic Pro 1 even if I do very very slow turns, the video quality get's worse. So therefore I always take videos this way: fly to starting point - start recording - fly a straight line - stop recording - do a turn - start recording - fly a straight line - stop recording - do a turn and so on. And in post combing the pieces for the final video.

So with 60fps I hope can continue filming while making (slow) turns.

In addition I have the DJI Action cam, and often I combine videos from the Mavic Pro 1 with the one from the action cam (obviously 4K 30fpts). But as the Action cam is capable of filming 4K 60fps as well, plan is to combine those videos with 60fps from the MA 2 in the future.

Why I don't have a MA 2 yet? Here in Europe, we won't get the version with AirSense (ADS-B in receiver), due to supply shortages only the U.S will get this version right from the start.. so I hope by the summer we can get the MA 2 with AirSense here in Europe as well ?

 

[allenface]

all settings the same, no, the framerate has zero effect on the image.

two videos, one at 30fps and one at 60fps...
1/120th second shutter
100 iso
f/2.8
ev+/-0

the exact same amount of light is going to hit the sensor, for the exact same amount of time. one setting will simply create 30 exposures while theother creates 60 in the same second.

This isn't a matter of individual understanding or preference. It's science/fact/reality.

what you choose to actually shoot at, that's preference.

i shoot primarily high action stuff (mountain biking) and may have used 60 (or higher) fps once or twice. i just dont find myself wanting or needing the extra frames often enough to justify the overhead. and premiere's ability to fill in the blank did work quite nicely the one time I wished for higher fps (during one of my crashes.)

Also tried 24 and that just didnt lookright to my eye for the activity at hand. so 30fps it is all the time.

I'm on your page. 30fps for me in my production work and in most cases on my drone. But sometimes, as you mentioned, 60 is nice to use if you know you will be doing some post manipulation. Some get shutter and frame rate confused and this is not the forum to get into that wicket. That said, these little cameras seem to use shutter as their exposure mechanism which is why ND's or polarizers are handy. I do have to take exception with your comment saying the same amount of light is entering the sensor at both 60 and 30. Your math says it's fact. My camera says otherwise. With my Sony Fs7 large sensor camera, if you shoot 24fps, I get more light gathering as the frame is running slower, gathering more light. The higher frame rate, the more light you loose. I may have misunderstood your assertion and if so, I apologize but coming from a film background, the math as always been the slower the framer rate, the more light enters the plane, as does motion blur. What am I missing here?

 

[brett8883]

I'm on your page. 30fps for me in my production work and in most cases on my drone. But sometimes, as you mentioned, 60 is nice to use if you know you will be doing some post manipulation. Some get shutter and frame rate confused and this is not the forum to get into that wicket. That said, these little cameras seem to use shutter as their exposure mechanism which is why ND's or polarizers are handy. I do have to take exception with your comment saying the same amount of light is entering the sensor at both 60 and 30. Your math says it's fact. My camera says otherwise. With my Sony Fs7 large sensor camera, if you shoot 24fps, I get more light gathering as the frame is running slower, gathering more light. The higher frame rate, the more light you loose. I may have misunderstood your assertion and if so, I apologize but coming from a film background, the math as always been the slower the framer rate, the more light enters the plane, as does motion blur. What am I missing here?

Your missing the fact that the sensor is only collecting light so long as the shutter is open.

At 30FPS 1/60, the shutter is open for 1/60 of a second per frame. At 60 FPS 1/60, the shutter is also open for 1/60 of a second per frame etc

 

[Geekinions]

What are you basing this on? Even if this is the case I don’t believe that’s how it works. At 60 FPS the image in each frame changes less from frame to frame so it should be easier to compress each frame without losing visual quality.

Actually, that's a very good question and one I'm more than happy to answer. It could be beneficial to a lot of people new to video. I'll try to keep it basic and not get into the nitty gritty because there is a lot to consider.

First off, I base it off of 27 years experience. I've worked in television since 1993. That's my background. ? Newb to drones, but video is my thing.

There are two important considerations. Bitrate and framerate. I'm not sure how the MA2 handles this in different recording situations, but I'll be testing it out the moment I've got a full charge on the battery to see exactly what each recording mode gives. But for today, without that info available to me, I'm going to assume that the UAV records at 120 megabits per second in both 60fps and 30 fps. 120 megabits of information are recorded each second. Not each frame. That means at 60fps, each frame only has an average of 2 megabits of information. At 30fps, that average goes up to 4 megabits of information per frame. Note I didn't say each frame HAS only 2 or 4 megabits of information, because it doesn't work that way. Its an average. That's where the codec comes in and calculates what information from one frame can be carried over to the next one (and further along) to save on space. That's as basic as I can make how bitrates and frame rates work together along with the compressor/decompressor. Keep in mind that the container file (MOV or MP4 in this case) has zero bearing on any of this. Its simply a container for the data, which is the codec (H.265 or H.264).

In 4K, H.265 is considerable more efficient at making a better picture from lower bitrates compared to H.264.) You can google exactly how much and why if you're curious, I'm just trying to keep it basic here, like I said. So if the MA2 gives you the option to record 4K30 using H.265 or H.264, I would always pick H.265 because of the better compression algorithm which will result in a better picture to the eye. Its a smart move that DJI only allows H.265 at 4K60 because even at 120 megabits a second, H.264 would start falling apart a bit.

Essentially the bitrate is the amount of information (a cup full of water) and the frame rate is how often that information is taken. (How often do you drink from that cup.) The faster you drink from it, the faster it will be empty. But if you have a larger bitrate (larger cup), it will take longer to empty at a faster frame rate.

*Unless* 30fps is limited to a lower bitrate than 60fps at the appropriate ratio for the H.265 codec that would result in things being equal. I don't have the MA2 yet, so I can't do a side by side comparison or get the exact numbers for the bitrates.

This of course also doesn't take account variable bitrate... I'm assuming it's CBR (constant bitrate). It may very well be VBR. In which case, things get even more complex and interesting.

Even with VBR, the answer is essentially the same, but if there's less motion, the bitrate goes down and if there's more motion it goes up to the theoretical max allowed. Either way, the more information you have per frame, the crisper it will be.

Ever wonder why your slow motion videos look janky compared to 30, 24 or 60 fps even though it says it's 120 megabits a second? There's less info to go around in those 240 frames a second.

So if you're color grading using an 8-bit codec, you want as much data per frame available and as such, want as high of a bitrate as possible. If you record a slower fps and can maintain that bitrate, you will have more information to process those grading choices.

Anyway, I did my best to keep it basic and explain it in laymen terms. Hopefully it helps you out a bit to understand why filming in 60fps result in lower quality video compared to 30.

 

[brett8883]

Actually, that's a very good question and one I'm more than happy to answer. It could be beneficial to a lot of people new to video. I'll try to keep it basic and not get into the nitty gritty because there is a lot to consider.

First off, I base it off of 27 years experience. I've worked in television since 1993. That's my background. ? Newb to drones, but video is my thing.

There are two important considerations. Bitrate and framerate. I'm not sure how the MA2 handles this in different recording situations, but I'll be testing it out the moment I've got a full charge on the battery to see exactly what each recording mode gives. But for today, without that info available to me, I'm going to assume that the UAV records at 120 megabits per second in both 60fps and 30 fps. 120 megabits of information are recorded each second. Not each frame. That means at 60fps, each frame only has an average of 2 megabits of information. At 30fps, that average goes up to 4 megabits of information per frame. Note I didn't say each frame HAS only 2 or 4 megabits of information, because it doesn't work that way. Its an average. That's where the codec comes in and calculates what information from one frame can be carried over to the next one (and further along) to save on space. That's as basic as I can make how bitrates and frame rates work together along with the compressor/decompressor. Keep in mind that the container file (MOV or MP4 in this case) has zero bearing on any of this. Its simply a container for the data, which is the codec (H.265 or H.264).

In 4K, H.265 is considerable more efficient at making a better picture from lower bitrates compared to H.264.) You can google exactly how much and why if you're curious, I'm just trying to keep it basic here, like I said. So if the MA2 gives you the option to record 4K30 using H.265 or H.264, I would always pick H.265 because of the better compression algorithm which will result in a better picture to the eye. Its a smart move that DJI only allows H.265 at 4K60 because even at 120 megabits a second, H.264 would start falling apart a bit.

Essentially the bitrate is the amount of information (a cup full of water) and the frame rate is how often that information is taken. (How often do you drink from that cup.) The faster you drink from it, the faster it will be empty. But if you have a larger bitrate (larger cup), it will take longer to empty at a faster frame rate.

*Unless* 30fps is limited to a lower bitrate than 60fps at the appropriate ratio for the H.265 codec that would result in things being equal. I don't have the MA2 yet, so I can't do a side by side comparison or get the exact numbers for the bitrates.

This of course also doesn't take account variable bitrate... I'm assuming it's CBR (constant bitrate). It may very well be VBR. In which case, things get even more complex and interesting.

Even with VBR, the answer is essentially the same, but if there's less motion, the bitrate goes down and if there's more motion it goes up to the theoretical max allowed. Either way, the more information you have per frame, the crisper it will be.

Ever wonder why your slow motion videos look janky compared to 30, 24 or 60 fps even though it says it's 120 megabits a second? There's less info to go around in those 240 frames a second.

So if you're color grading using an 8-bit codec, you want as much data per frame available and as such, want as high of a bitrate as possible. If you record a slower fps and can maintain that bitrate, you will have more information to process those grading choices.

Anyway, I did my best to keep it basic and explain it in laymen terms. Hopefully it helps you out a bit to understand why filming in 60fps result in lower quality video compared to 30.

Thanks for putting the time to go over that, the only issue is that we are assuming 120Mbps at 60 FPS and 100Mbps at 30 FPS. It’s also definitely VBR.

My issue is I don’t think we can assume that 60 FPS is somehow lower visual quality. Not only is 60 FPS actually higher bit rate, the frames play back faster with each frame being played back at a faster rate, AND due to the details of how H.265 reduces bit rate it should be expected that a higher frame rate with less difference between frames would be less than double the bit rate of 30 FPS without a reduction in visual quality.

How HEVC reduces bit rate is the key here. It saves the DIFFERENCE between frames using b-frames and p-frames. I-frames are full frames.

So what HEVC is really recording looks like this



This is one of the ways HEVC is able to reduce data rates so much without losing quality over H.264.

“FIGURE 2: EXPLOIT COMMONALITY AND REDUCE DATA BY IDENTIFYING THE DIFFERENCES BETWEEN FRAMES

With high motion video, such as sports content, the information that stays the same from frame to frame is significantly less important than predicting the motion of objects from frame to frame. With motion compensation algorithms implemented in the encoding process, the codec is able to take into account the fact that most of what makes up a new frame in a video sequence is based on what happened in previous frames. So at a block by block level, the encoder can simply code the position of a matching object in the frame and where it is predicted to exist in the next frame via a motion vector. The motion vector takes fewer bits to encode than an entire block and thereby saves bandwidth on the encoded stream.
(https://www.e-idsolutions.com/wp-content/uploads/2016/10/HEVC_Demystified.pdf)”

So my point is that since the difference between frames is less at 60 FPS, the encoder can compress the video much more efficiently than at 30 FPS. I don’t believe we can make the assumption that because the average bits per frame is less that means the visual quality is automatically less. To use your analogy, H.265 sips from the cup less if the difference between I-frames is small.

The question in my mind is if the DJI encoder uses a different method to compress 60 FPS than 30 FPS and if there’s a perceived quality difference when decoded.

 

[dronecrasher]

Thanks for putting the time to go over that, the only issue is that we are assuming 120Mbps at 60 FPS and 100Mbps at 30 FPS. It’s also definitely VBR.

My issue is I don’t think we can assume that 60 FPS is somehow lower visual quality. Not only is 60 FPS actually higher bit rate, the frames play back faster with each frame being played back at a faster rate, AND due to the details of how H.265 reduces bit rate it should be expected that a higher frame rate with less difference between frames would be less than double the bit rate of 30 FPS without a reduction in visual quality.

How HEVC reduces bit rate is the key here. It saves the DIFFERENCE between frames using b-frames and p-frames. I-frames are full frames.
View attachment 102006
So what HEVC is really recording looks like this
View attachment 102007
View attachment 102008
View attachment 102009
This is one of the ways HEVC is able to reduce data rates so much without losing quality over H.264.

“FIGURE 2: EXPLOIT COMMONALITY AND REDUCE DATA BY IDENTIFYING THE DIFFERENCES BETWEEN FRAMES

With high motion video, such as sports content, the information that stays the same from frame to frame is significantly less important than predicting the motion of objects from frame to frame. With motion compensation algorithms implemented in the encoding process, the codec is able to take into account the fact that most of what makes up a new frame in a video sequence is based on what happened in previous frames. So at a block by block level, the encoder can simply code the position of a matching object in the frame and where it is predicted to exist in the next frame via a motion vector. The motion vector takes fewer bits to encode than an entire block and thereby saves bandwidth on the encoded stream.
(https://www.e-idsolutions.com/wp-content/uploads/2016/10/HEVC_Demystified.pdf)”

So my point is that since the difference between frames is less at 60 FPS, the encoder can compress the video much more efficiently than at 30 FPS. I don’t believe we can make the assumption that because the average bits per frame is less that means the visual quality is automatically less. To use your analogy, H.265 sips from the cup less if the difference between I-frames is small.

The question in my mind is if the DJI encoder uses a different method to compress 60 FPS than 30 FPS and if there’s a perceived quality difference when decoded.

Wow I have to admit this is all very informative and I definitely learned a something new but that was also very technical and caused major brain damage Thanks for taking the time you two for explaining the more technical aspect of it

 

[Hiflyer808]

Wow I have to admit this is all very informative and I definitely learned a something new but that was also very technical and caused major brain damage Thanks for taking the time you two for explaining the more technical aspect of it

Don't sweat the quality thing. Whether better or worse quality it's by a tiny amount you wouldnt notice. If final product going to be 30fps you might as well do 60fps to have more options. For tracking, hdr or super low light switch to 30fps.

 

[Phantomrain.org]

You should always try to fly 60 frames you never know when the extra frame that you grab on the time line is going to offer something really nice. If you have the editing power its worth it. Slow Motion is nicer because you have a smoother transition. Plus 60 frames might be your only chance of catching a UFO pic





Phantomrain.org
Gear to fly your Mavic in the Rain.
Coal

 

[Ron Caballero CAP]

what speaks against h.265? Isn't' it the same quality as 264 just with a smaller file size? If I don't care about HDR, what else?

You don't care about HDR? What a shame. Try it in situations that have high dynamic range...that is "high contrast" where some parts of the image/s are brighter and others darker. Both will not normally be equally visable at the same time. I believe you will be pleased.

 

[Ron Caballero CAP]

You should always try to fly 60 frames you never know when the extra frame that you grab on the time line is going to offer something really nice. If you have the editing power its worth it. Slow Motion is nicer because you have a smoother transition. Plus 60 frames might be your only chance of catching a UFO pic

View attachment 102268
View attachment 102269
View attachment 102270

Phantomrain.org
Gear to fly your Mavic in the Rain.
Coal

Clearly, 60fps is benefidial when filming in a lightening storm.