60FPS vs 30FPS

[moldorf]

I imagine some versions of this question have already been asked and answered many times, so I apologize in advance for the cyber-clutter...

anyway, I've noticed if I record raw 4K video with my Mini 3 at 30fps there's around 100,000kbps and at 60fps there's about 150,000kbps. That would seem to indicate that 60fps has about 50% more data per second but about 25% less per frame. But I also understand that sometimes pure mathematical comparisons don't yield equal real world results

my question is this: almost always when I do landscape flyovers (of more than 200 feet) I will edit the video in Resolve. And I almost always 'speed up' the video anywhere from X2 to X4 (a 20 minute flyover is a lot less boring if it's 5 minutes). I try to keep whole integers as the multiplier. Will I see any final video quality difference between the raw input of 60fps vs 30fps? I try to archive my 'keeper' raw videos but those 60fps raw videos are huge files. I have a 28 minute flight that is a 32GB file. Archiving that just seems excessive

by the way, I have noticed, at least to my eyes, that I get the best results, with fewer artifacts, by distilling the raw video to 24fps regardless if the input is 60fps or 30fps

 

[JoelP]

This is a great question. I have been thinking along the same lines and hope one of the pro videographers would weigh in here. By which technique do you distill to 24 fps in Resolve?

 

[moldorf]

This is a great question. I have been thinking along the same lines and hope one of the pro videographers would weigh in here. By which technique do you distill to 24 fps in Resolve?

the FPS selection menu on the render window

 

[SethB]

If your final edited output is 24 or 30fps, filming in 60 is *only* helpful for slow motion.

I don’t think you’re getting any benefit from all those frames if editing fast mo and regular speed.

 

[JoelP]

While it isn’t so much of an issue for the gimbal
stabilized drone video I wonder if shooting at a higher frame rate allows Resolve or similar video editors to do a better job with image stabilization for other video.

 

[moldorf]

If your final edited output is 24 or 30fps, filming in 60 is *only* helpful for slow motion.

I don’t think you’re getting any benefit from all those frames if editing fast mo and regular speed.

I was thinking along those lines. That's why I mentioned data/frame vs data/second. If there was more data per frame I'd think that might yield better quality in a rendered video. But there's actually less data/frame

but I really don't know how the rendering engine of Resolve processes the data and constructs the finished frames

 

[SethB]

I was thinking along those lines. That's why I mentioned data/frame vs data/second. If there was more data per frame I'd think that might yield better quality in a rendered video. But there's actually less data/frame

but I really don't know how the rendering engine of Resolve processes the data and constructs the finished frames

h.264 has a few profiles with different features; I’ve not checked on what DJI is doing but I’d guess that they are using the main or high profiles which include bi-predictive frames aka b-slices. That is, a non-keyframe is based on changes from previous and following frames in a group of frames. I think h.265 does this by default.

Decoding that: one *can* look at bitrate for the same codec and frame size as determining quality. Data per second is higher with 60p, quality would be better.

With a couple caveats: 1) same content. 2) same frame size. 3) same codec. Note that h.265 can be about twice as efficient as h.264, meaning that for a given visual quality measure h.264 needs twice-ish the bitrate as h.265.

Then, Resolve will go by the codec spec to decode the b slices with all the qualities that went into them. These encoding methods create groups of frames, so Resolve must decode groups of frames to recreate frames.

But that this is true in theory does not mean that we could see such quality differences with our eyes. One would need to carefully test and evaluate to determine if any visual quality gains can be had, and if they are worth the various costs.

In theory, theory and practice are the same. In practice, they’re not!

For example, DJI may use slightly different codec settings at 60p than at 30p. Most people should base framerate choices on their intended use most of the time…