@Dexcon Yeah, I've got a Samsung S21 Ultra, 30, 60 it says in the phone camera but 29.97 variable is what it creates,

I understand that 'Variable' says it all, it's in the label & that it can bring a whole array of rates, but like you say I have seen little difference between variable & constant. When i'm messing around with the Boris effects it's pretty much always on clips that have already been edited & rendered as constant, I see just as many crashes with those as i do with my unedited phone clips,

Seeing recent iPhone adverts here described the phone as "Hollywood in a pocket" or something similar maybe it's time phone manufacturers stopped using variable framerate. To my knowledge Hollywood doesn't distribute movies with variable framerate. If they did I wonder how players including your TV would handle them.

@Dexcon Yeah, I've got a Samsung S21 Ultra, 30, 60 it says in the phone camera but 29.97 variable is what it creates,

I understand that 'Variable' says it all, it's in the label & that it can bring a whole array of rates, but like you say I have seen little difference between variable & constant. When i'm messing around with the Boris effects it's pretty much always on clips that have already been edited & rendered as constant, I see just as many crashes with those as i do with my unedited phone clips,

 

@Former user I had a look at my Samsung footage, 1 thing to note, Vegas is the most sensitive with dropping frames compared to other NLE's like Resolve and Premiere, but I did my testing in Vegas, it's possible these results aren't seen in other NLE's to the same extent. EDIT: Identical results in Resolve

duplicated frames

4,5

21-25

26-28

I checked the next minute, there are no dropped frames in Vegas, then checked 30frame increments throughout the video, no dropped frames. I've heard a similar conjecture saying variability with phones happens at beginning and end. I had a look at the end few seconds on a couple of videos, but don't see any dropped frames.

I then cut off the first few seconds 'Cut without re-encoding' however media info still reports VFR only now there is no VFR at the beginning. I searched 30frame samples throughout video, did not see dropped frames, but they're probably there. One guess is phones processors get overloaded due to the image it's recording if the next frame is vastly different to the one before such as static camera then bump it with force.

Other idea is it's not related to what the camera sees but background processes that can swamp the processor. If you could be bothered in checking an entire phone video for dropped/dupe frames I'd be interested in the results.

@Former user Hi n thanks, I'm interested in all this but still learning so a fair bit of this goes over my head, If you want tho you're welcome to 10 of my clips, you'd prob do a better job of analyzing them, 🤦‍♂️🤣👍

https://drive.google.com/file/d/1_-ObFyXj8Onx3PeRmFS0dH42BNDNDHJo/view?usp=sharing

I had a look at a couple of your files, dropped frames as expected, and Vegas reacts badly with Vegas dropping many more frames where the dropped frame/frames is. play it over enough times it will play smoothly, correctly only displaying duplicated frames where it should, and no further lagging, but this is caching, so it is having a problem with playback as you say. Your phone files are the simplest available, AVC, low bitrate, no B frames, and a GOP of 30. Higher GOP's with B frames at higher bitrates are likely going to be more of a problem for Vegas

Resolve played back perfectly, Premiere had a problem with 1 of the dropped frames, I would guess it lagged for double the dropped frames, so 2 dupe frames played back as 4 dupe frames. After a single playback it played back perfectly. I don't have caching turned on but Premiere looks to be doing the same sort of ram caching as Vegas does. Vegas had a problem with every dropped frame, showing small lags 4 or 5 times in the same video that Premiere lagged once and Resolve didn't lag.

Vegas does a poor job in comparison and I agree, it's wrong to say 'file is VFR, you have no right to complain' it depends on the variability and for your average phone video, it's not that great.

Hi @Former user

I use phone footage quite often too. My poor Galaxy A40 records 1080p 30fps, variable frame rate, GOP of 30, 1 reference frame, 17Mbps.

It works quite well with Vegas. Most of the time I don't even bother to convert the videos to constant frame rate. When I do I don't see any difference in performance between the original variable frame rate file and the converted constant frame rate one.

On some occasions the original file plays back better than file converted with Handbrake or Shutter encoder despite being variable frame rate. Don't know, maybe it's due to changes to GOP structure and reference frames during conversion.

@misohoza I think Gid may be looking at it from a perfectionist point of view. not insinuating it's so bad as to not be editable. If the phone camera drops 1 frame (2 duplicate frames on timeline) Vegas should play 2 duplicate frames, not lag out for 10.

It's a different story with maybe OBS where the CPU gets completely bogged down at various points when encoding CFR, or for an screen recorder setup for power point presentations etc where VFR may make more sense (except for editing) creating vastly smaller files with very high efficency. All editors will have more of a problem with those files over phone cameras.

My impression is that slight variations are no problem and are averaged out. It's the huge variations sometimes encountered that cause problems. I wonder what is the reason for phone's using VFR?

I wonder what is the reason for phone's using VFR?

I recall that someone on the forum a few years ago suggested that it was to keep video file sizes to a minimum. If there's not much movement in what's being filmed, then the phone reduces the frame rate thus reducing the file size for that low movement period of recording. It makes sense but maybe there's another reason.

Now that many phones allow for an additional larger capacity micro-memory card, why is VFR still necessary? They should give a choice of VFR and CFR along with 25/50 fps. Surely it's not that hard to do. Surely coding for CFR would have to be simpler than the variables needed for VFR.

In comparison, I think that camera operators in the early silent era did much better with their hand-cranked cameras when they got 16 fps fairly consistently.

VFR is also used to compensate for lighting. I know cheap cameras (like ones on inexpensive drones) do not have variable lens openings, like f-stops in still cameras, so VFR is used it darker situations for a better exposure.

I wonder what is the reason for phone's using VFR?

I recall that someone on the forum a few years ago suggested that it was to keep video file sizes to a minimum. If there's not much movement in what's being filmed, then the phone reduces the frame rate thus reducing the file size for that low movement period of recording. It makes sense but maybe there's another reason.

 

Hi @Dexcon That makes sense, i use a tripod & handheld, but either way there's always some movement in my clips,

I don't know if this says anything ...

I pointed the camera at the wall for 3mins, no movement or light change.

& this is one of my clips with lots of movement in shot, most of my clips look like this, or a variation of, whether it's on a tripod with me working in the shot, or if it's me holding the camera & panning around, 🤷‍♂️

I guess it would need deeper analysis 🤷‍♂️

 

Hi @Dexcon That makes sense, i use a tripod & handheld, but either way there's always some movement in my clips,

I don't know if this says anything ...

Modern phones don't use VFR, as you showed you get the lowest frame rates as reported by media info with high motion high light. I have done similar tests in the past, given it perfect VFR conditions, bright light perfectly still for 5 minutes of a floor board. this is the result

• Bit rate                                 : 54.0 Mb/s
• Width                                    : 3 840 pixels
• Height                                   : 2 160 pixels
• Display aspect ratio                     : 16:9
• Frame rate mode                          : Variable
• Frame rate                               : 30.000 FPS
• Minimum frame rate                       : 29.960 FPS
• Maximum frame rate                       : 30.060 FPS
• FrameRate_Real                           : 30.000 FPS
•  

If it was recording in VFR we'd expect to see a massive saving in bitrate and very long frame rates, but infact it's variability is close to none. Phones don't work like dashcams or security cameras that often do use VFR to save space on limited recording medium. Media player saying VFR, means variability in frame, rate not that it was recorded in VFR. In the case of phone cameras it appears they drop frames, usually to do with high motion and when they start recording.

There's also a strangeness in the first GOP as far as variable bitrate that doesn't correspond to the video. It can't detect variable frames, but it most likely happens here as recording begins, causing the perceived variable bitrate , much more variable compared to the following GOP's

Dexcon wrote on 9/28/2022, 5:03 AM

I wonder what is the reason for phone's using VFR?

I recall that someone on the forum a few years ago suggested that it was to keep video file sizes to a minimum. If there's not much movement in what's being filmed, then the phone reduces the frame rate thus reducing the file size for that low movement period of recording. It makes sense but maybe there's another reason.

That is exactly correct.

Unfortunately, relying on MediaInfo to tell one if a file is Constant or Variable Frame Rate is a sticky wicket.

MediaInfo has many false positives as a result of clock jitter, which is not a frame rate variation, just an oscillating number sequence trying to express an irrational number (refer to middle school math).

30/1001, a common frame rate that cannot be expressed as a rational number, divided by 90kHz, a common clock frequency used by x264/265 among others, is an irrational number, inexpressible by decimal math, which is what MediaInfo prints out.

To determine if MediaInfo is telling the truth, there are a couple of tests:

1. Look at the spread between high and low numbers -- is it less than one? Chances are it is CFR being expressed as two rational numbers.
2. Average the high and low numbers to determine the irrational mean. Is it somewhere near a common NTSC frame rate (29.970, 23.974, 59.940)? Again, chances are it is CFR.
3. Real VFR can vary by as much as 70% with motion video between high and low sample points. Talking heads with a stationary phone don't count. CFR is almost always less than a one frame variation.

If MediaInfo reported frame rates in fractional, rather than decimal math, there wouldn't be this problem. We've seen a need to repeat this information from time to time for the benefit of newer hobbyists.

does anyone know what tool it is that is shown here:

https://www.vegascreativesoftware.info/us/forum/variable-frame-rate-being-an-issue--137374/#ca859193

( Former user wrote on 9/28/2022, 10:51 AM ... )

@Former user I'm curious as to how a TV would handle VFR. Can you plug your phone into your TV and play the camera files. If so what is the result?

@EricLNZ Don't know, I would expect so but won't know till i try, the TV has a USB port, I could put some files on a stick & see , tomorrow tho, it's 3.41am now 🥱

I wonder what is the reason for phone's using VFR?

I recall that someone on the forum a few years ago suggested that it was to keep video file sizes to a minimum. If there's not much movement in what's being filmed, then the phone reduces the frame rate thus reducing the file size for that low movement period of recording. It makes sense but maybe there's another reason.

Possibly.
But the real reason is the ability to extend the frame time (and exposure) in low light. The frame is longer, but at least you can see something.

But the real reason is the ability to extend the frame time (and exposure) in low light. The frame is longer, but at least you can see something.

My 12 year old Canon AVCHD camera does the same thing in low light but in a different way. It drops the shutter speed way below the framerate. It achieves this by duplicating frames across the exposure time. The result on viewing will be the same as phones with VFR but the difference is the framerate is preserved.