DJI Mavic, Air and Mini Drones
Friendly, Helpful & Knowledgeable Community
Join Us Now

Please Explain This Video/Edit Issue to Me

moldorf

Well-Known Member
Joined
May 30, 2022
Messages
535
Reactions
421
Age
72
Location
Oregon
OK....I have a Mini 3

I recently returned from a 2 week camping trip to SE Oregon; some mountains of the Great Basin desert

I was capturing 4K video, 30fps, h-264. Now, lots of my flights were 12-18 minutes long. Usually, I was flying 200-350 feet above the landscape. The thing is, a 15 minute video where the view only gradually changes can be kind of boring. So, I was using Resolve to speed up the videos, on average probably 2.2-2.6 times. Condensing a 15 minute video to 5-6 minutes made for much more interesting video

an issue I encountered though, was that occasionally I was getting some flickering/shimmering/jumpiness on areas of the video that had rather uniform colors and textures. For example large patches of tan/grey sagebrush; or forests of green trees; or expansive cliff faces. It was pretty distracting. I went back and looked at the raw videos and while I noticed a slight flicker, every once in a while, it just wasn't as extreme as the edited video. I don't know how else to describe the artifacts other than flickering/shimmering/jumpiness

anyway, I tried a lot of different methods in Resolve (free version) to correct the issue. Converted to H-265; increased the bit rate; down-converting to 1080p. The artifacts were still there. What finally 'solved' the issue for me (when I say solved I mean it mitigated the artifacts down to a level where they were gone rarely noticeable) was to down-convert from 30fps to 24fps while keeping the bitrate at about the same level as the raw captures (around 95,000kbs). I even down-converted to 1080 at 24fps and 30,000kbs and the video was fine

my question(s) is what is introducing this shimmering to the video? Is it just the camera quality? I was 'filming' a lot while flying in sport mode; could that be a problem? And why would the process of rendering at a higher speed significantly expand the shimmering/jumpiness? And why would dropping the frame rate by 20% mitigate that shimmering so significantly?
 
What frame rate is your project? Your source video? When you started the project, loading the first video, did Resolve ask you if you wanted to update the project frame rate?
 
Good questions, retiredNH. Frame rate conversions will frequently introduce artifacts. This is especially problematic with non-integer speed changes.

It's always best to speed up or slow down in increments of 2 or 4 or 0.5. Any changes other than whole numbers will force the software to duplicate or delete uneven numbers of frames to achieve the editor's requests. This is unlikely, I think to introduce the "shimmering" effects you report.

Fine textures (like distant views of millions tiny dots like those presented by sagebrush) can interfere with the scanning frequency or line structure of the video. That may be what you're seeing. "Scintillation" is the computer graphics term for that issue.

When you view the camera originals in VLC (or some other media player), do you see the same artifacts?

Posting a link to a small sample of the camera original and the problematic post-edit footage would be very helpful.
 
Good questions, retiredNH. Frame rate conversions will frequently introduce artifacts. This is especially problematic with non-integer speed changes.

It's always best to speed up or slow down in increments of 2 or 4 or 0.5. Any changes other than whole numbers will force the software to duplicate or delete uneven numbers of frames to achieve the editor's requests. This is unlikely, I think to introduce the "shimmering" effects you report.

Fine textures (like distant views of millions tiny dots like those presented by sagebrush) can interfere with the scanning frequency or line structure of the video. That may be what you're seeing. "Scintillation" is the computer graphics term for that issue.

When you view the camera originals in VLC (or some other media player), do you see the same artifacts?

Posting a link to a small sample of the camera original and the problematic post-edit footage would be very helpful.

What he said...he just saved me a lot of typing :)

I ran into a similar situation and it had to do with setting the framerate incorrectly in the project so it didn't match the framerate of my clips. Somehow I forgot to set that. Additionally I sped things up and found that using whole number increments or at least .5 helped as well.
 
  • Like
Reactions: AlbionDrones
What he said...he just saved me a lot of typing :)

I ran into a similar situation and it had to do with setting the framerate incorrectly in the project so it didn't match the framerate of my clips. Somehow I forgot to set that. Additionally I sped things up and found that using whole number increments or at least .5 helped as well.
Had the same issue myself with diferent frame rates, annoying, but can be overcome when you know what the cause is :)
 
  • Like
Reactions: maggior
Good questions, retiredNH. Frame rate conversions will frequently introduce artifacts. This is especially problematic with non-integer speed changes.

It's always best to speed up or slow down in increments of 2 or 4 or 0.5. Any changes other than whole numbers will force the software to duplicate or delete uneven numbers of frames to achieve the editor's requests. This is unlikely, I think to introduce the "shimmering" effects you report.

Fine textures (like distant views of millions tiny dots like those presented by sagebrush) can interfere with the scanning frequency or line structure of the video. That may be what you're seeing. "Scintillation" is the computer graphics term for that issue.

When you view the camera originals in VLC (or some other media player), do you see the same artifacts?

Posting a link to a small sample of the camera original and the problematic post-edit footage would be very helpful.
thanks for the reply, Lister...appreciate it

* generally, I have set the speed increment set at 2.5. I think 4 would be too fast. I guess 2 would work. I'll have to try both just to test. But it looks like you are saying 2.5 could be problematic

* and yes, as I said, this 'scintillation' seemed to occur on fine textures....rather uniform colors and texture, and seen from a distance away, or above.

* and no, I did not really notice the artifacts on my PC when played in VLC. Maybe a little more grainy at points. Of course, that playback is on an older 23 inch monitor, and I then watch on a brand new 43 inch 4K TV. That might make a difference

* here is something I just noticed, and I double checked my drone thru the Fly app: I have the 4K recording on the Mini 3 set at 30fps. But when I check the downloaded raw video properties tab in video details on my win10 PC, it says the frame rate is 29.97. Wut?

as far as linking/posting video samples, I don't really know how to do that. Seems like a 10-15 second video would be pretty large
 
Last edited:
66-78fps in a 30fps timeline will come out like crap. You shot 30fps, so use 30, 60, 90.
I had to read that a couple of times to understand what you were referring to. Correct me if I misunderstand you

I initially said I sped up the video 2.2-2.6 times. 2.2 = 66; 2.6 = 78. So, you are saying the same thing as Lister, that being that for best results keep the speed increase to a whole number. Essentially 60=2 and 90=3. In other words, best if I speed up by either 2 or 3 or 4, etc.

thanks for the reply by the way
 
here is something I just noticed, and I double checked my drone thru the Fly app: I have the 4K recording on the Mini 3 set at 30fps. But when I check the downloaded raw video properties tab in video details on my win10 PC, it says the frame rate is 29.97. Wut?
That’s pretty normal. I’m not sure why the standard frame rate is 29.97 instead of 30, but it is. But every camera I’ve used just says 30, even though it seems to be actually 29.97 once you get it onto a computer and inspect the properties.

Did you turn on any kind of noise reduction in Resolve? I’m not sure if that feature is available in the free version or not. I was editing a sky/sunset hyperlapse I made, and enabled some mild NR in Resolve. On this type of footage, it seemed to do the opposite of what I expected it to do no matter which method I picked. It basically induced flickering/shimmering/jumpiness. Took me a while to figure out that’s what was causing it.
 
I had to read that a couple of times to understand what you were referring to. Correct me if I misunderstand you

I initially said I sped up the video 2.2-2.6 times. 2.2 = 66; 2.6 = 78. So, you are saying the same thing as Lister, that being that for best results keep the speed increase to a whole number. Essentially 60=2 and 90=3. In other words, best if I speed up by either 2 or 3 or 4, etc.

thanks for the reply by the way
Best not to speed up by 3, either. Try 2X or 4X and see what happens. If the problem disappears, then you've found the problem.

Worst thing to do is select "fit to fill", which basically forces the software to play back at some weird, non standard fps.

29.97 is "sorta" the same as 30 fps.

Here's some info on the where and why of 29.97

 
I had to read that a couple of times to understand what you were referring to. Correct me if I misunderstand you

I initially said I sped up the video 2.2-2.6 times. 2.2 = 66; 2.6 = 78. So, you are saying the same thing as Lister, that being that for best results keep the speed increase to a whole number. Essentially 60=2 and 90=3. In other words, best if I speed up by either 2 or 3 or 4, etc.

thanks for the reply by the way
Yes. Increments of 15 may work as well. In Resolve you can change the clip speed as a percent or framerate. You can also force framerates of clips in the Media Pool prior to importing on the timeline of a different fps. For example, I shoot some 4K60. I make a 4K30 timeline. I go to the Clip Attributes in the Media Pool and change it to 30fps. Now when I bring it into the timeline, it's at 30fps so it's at half speed.
 
29.97 is due to "drop frame" system used in the US. It is purely done in how frames are counted, no frames are really discarded. Only applies to NTSC only, not PAL
 
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
134,442
Messages
1,594,815
Members
162,978
Latest member
dojin23