Mavic 3 lack of details in 4K 120fps video?

[SkywalkerFeng]

This may not come as a surprise to video professionals - but I shot mostly still images, and this is the first time I tried shooting 4K videos, in H.265 format, both 60fps and 120fps. I immediately noticed the lack of details in the 4K 120fps video. In comparison, 4K 60fps looks much sharper, with more detail.

I guess this is due to the 200 Mbps maximum Bitrate for H.264/265 video?

Does Apple ProRes solve this problem?

 

[Phantomrain.org]

Interesting, I am going to have to check this out tomorrow as i have not noticed it yet.

Phantomrain.org
Gear to fly in the Rain

 

[brett8883]

This may not come as a surprise to video professionals - but I shot mostly still images, and this is the first time I tried shooting 4K videos, in H.265 format, both 60fps and 120fps. I immediately noticed the lack of details in the 4K 120fps video. In comparison, 4K 60fps looks much sharper, with more detail.

I guess this is due to the 200 Mbps maximum Bitrate for H.264/265 video?

Does Apple ProRes solve this problem?

View attachment 144480

If those two images are unedited it looks like they’ve gone through different image pipelines. If that’s the case it may be that the image processor can’t keep up at 120 fps so you get a less processed image. If it was just sharpness I would have said the sensor readout mode is just different in 120p mode but it looks more saturated and vibrant in the 60p image. Hallmarks of different processing being used.

Typically that wouldn’t be a bad thing, most processionals would prefer a less processed image but the M3 relies heavily on image processing to make up for the underperforming main lens. It actually does a really good job which makes it probably a bad thing in this case that the image processing is dialed back for 120p mode.

 

[Bryce steiner]

The only difference I see is that the 120 is cropped more. Is that normal?

 

[SkywalkerFeng]

The only difference I see is that the 120 is cropped more. Is that normal?

Pay attention to the hotel building by the cliff. And the trees. They look sharper, with more details.

The screen capture of the videos also resulted in image files of two different sizes, which indicates one has more details.

4K 60fps - 14.8 MB
4K 120fps - 13.9 MB

 

[akdrone]

If those were individual frames exported from a timeline I would not be entirely shocked to learn that the very minor difference in sharpness might be due to movement of the camera during the capturing of the frame. The drone is not sitting perfectly still on a tripod so I would expect some minor variation in terms of sharpness from one frame to another. Even if you looked at 10 or twelve frames it would be hard to judge, not knowing if a minute earlier the wind was more or less strong the the drone was more or less stable. The difference in sharpness is, after all, pretty minor. I don't dismiss the possibility that the 120fps image shot by your drone is indeed a bit less sharp but one frame doesn't tell us that definitively. To my way of thinking it would take a fair bit more testing to say that the 120fps is not as sharp - at least from simple observation - since nobody has piped in here with the bit rate of 120 vs 60. The M3 manual states that the bitrate for H264 is 200Mbps and for 265 it is 140 so I would expect to see some difference there but I wouldn't expect different bit rates for the two fps variations. I'll be interested to learn what others have to say although the difference between the two, for my purposes, is negligible.

 

[SkywalkerFeng]

If those were individual frames exported from a timeline I would not be entirely shocked to learn that the very minor difference in sharpness might be due to movement of the camera during the capturing of the frame. The drone is not sitting perfectly still on a tripod so I would expect some minor variation in terms of sharpness from one frame to another. Even if you looked at 10 or twelve frames it would be hard to judge, not knowing if a minute earlier the wind was more or less strong the the drone was more or less stable. The difference in sharpness is, after all, pretty minor. I don't dismiss the possibility that the 120fps image shot by your drone is indeed a bit less sharp but one frame doesn't tell us that definitively. To my way of thinking it would take a fair bit more testing to say that the 120fps is not as sharp - at least from simple observation - since nobody has piped in here with the bit rate of 120 vs 60. The M3 manual states that the bitrate for H264 is 200Mbps and for 265 it is 140 so I would expect to see some difference there but I wouldn't expect different bit rates for the two fps variations. I'll be interested to learn what others have to say although the difference between the two, for my purposes, is negligible.

I re-did the comparison, this time on tripod, in no wind environment. For both video, the focus is on the body of the tree. I still get the same result. The captured frame from 4K 60fps video looks sharper with more details, when compared to 4K 120fps. It also retains more highlight, in high contrast scene.

 

[Dave Maine]

If the 2 files sizes are the same as you indicate, the 120 fps individual frame has only half the bits that the 60 FPS has. That necessarily degrades the still frame image significantly.

ProRes may or may not help, as we don't know where the limitation in the image pipeline is. ProRes will produce files at least 6x the size of the ones you are getting now.

 

[SkywalkerFeng]

If the 2 files sizes are the same as you indicate, the 120 fps individual frame has only half the bits that the 60 FPS has. That necessarily degrades the still frame image significantly.

ProRes may or may not help, as we don't know where the limitation in the image pipeline is. ProRes will produce files at least 6x the size of the ones you are getting now.

I captured both videos for 10 seconds. The resulted file size is not the same. 4K120fps video file size is 1.6 times bigger. But, given the double frame rate (120fps v.s. 60fps), I was expecting double the file size. Maybe there are more aggressive compression applied to 120fps video.

 

[Dave Maine]

Very likely. The amount of data captured I limited by the bit rate of the recording media - the speed of the SD card being written to. Generally, the fastest speed it 100 mbps for U3 cards. The P3 Cinema uses different media to use ProRes, as 100 mbps is way too slow for 4K 30 fps video, never mind higher pixel dimensions or faster frame rates.

All codecs that write 4K images at 100 bps, are encoding lossy images.

 

[akdrone]

I re-did the comparison, this time on tripod, in no wind environment. For both video, the focus is on the body of the tree. I still get the same result. The captured frame from 4K 60fps video looks sharper with more details, when compared to 4K 120fps. It also retains more highlight, in high contrast scene.

View attachment 144489

Now we're getting somewhere. And were both shot at the same fstop, shutter speed and ISO? That would be another important variable. The blown out sky on the 120 makes me question that...but each of those variables needs to be accounted for before we can attribute it to the 60vs 120....

 

[Kilrah]

It's common for cameras to have to skip some processing steps in the more demanding modes.

 

[SkywalkerFeng]

Very likely. The amount of data captured I limited by the bit rate of the recording media - the speed of the SD card being written to. Generally, the fastest speed it 100 mbps for U3 cards. The P3 Cinema uses different media to use ProRes, as 100 mbps is way too slow for 4K 30 fps video, never mind higher pixel dimensions or faster frame rates.

All codecs that write 4K images at 100 bps, are encoding lossy images.

For Mavic 3, the maximum Bitrate is 200 Mbps (25 MB/s) for H.264 video, and 140 Mbps (17.5 MB/s) for H.265 video.

The micro sd card I am using is SanDisk Extreme 128G v30 A2 microSDXC, rated at 160 MB/s read, and 90 MB/s write. So the card should not be the bottleneck in this case.

 

[brett8883]

It's common for cameras to have to skip some processing steps in the more demanding modes.

Yep looks pretty clear the 120fps is less processed.