Capturing 24fps with a 1080p30fps capture card, will I miss frames?

[nMaib0]

I have an old capture card, AverMedia LGP Lite and I have noticed that when I capture 720p60fps the video transitions are fluid and smooth, despite getting some serious duplicates, but when I set the card to 1080p60fps I get fewer duplicates and the transitions stutter a bit.

My question is, from the 60hz that the card is receiving and the 30 that it's actually capturing, am I missing important frames even though the original frame rate of the video is 23.976? if so which avisynth filter could help smooth the motion where it really shows, thanks.

So far I have been able to remove the dupes with tdecimate() but I'm not quite sure how to make the video smoother.

[jagabo]

If you capture a 24p movie at 30 fps you won't be missing frames (assuming the capture worked correctly) but you will have a duplicate frame every 5th frame. Ie, 4 frames becomes 5 by duplicating one of those frames. Hence 24 frames becomes 30. You can eliminate those extra frames with TDecimate(). That should make it as smooth as 24 fps ever is. After TDecimate() step through a medium speed panning shot and look at the motion at every frame. Are there duplicates or missing frames (jumps in the motion)?

[nMaib0]

Yeah I notice the duplicate every 5 frames and the motion compared to when I capture 60fps 720p is more robotic, sometimes it looks good and then for a moment it stutters.

Also if the card is getting 60hz and the result is 30fps, aren't I losing 30 frames and maybe within these frames there are several good ones? I mean, when I record at 60fps 720p the motion is flawless even though it has way more duplicated frames of course.


Edit: Here's a clip from the video, still not decimated

https://mega.nz/#!LIYWlA4C!hDUex5pTDAdW4Fmfx7bLPwKLF6BuytoIPtYw755SXp8

[jagabo]

24p material is normally converted to 60p by using 3:2 duplicates:

1 1 2 2 2 3 3 4 4 4 5 5 6 6 6 7 7 8 8 8...

If you capture that at 30 fps you should get every other frame:

1   2   2   3   4   5   6   6   7   8...

You can see the result that one of every 4 frames is duplicated (2, and 6 here). TDecimate() by default removes 1 duplicate out of every 5 frames:

1   2       3   4   5   6       7   8...

Leaving you with the original 24p film frames. If you're not getting smooth playback from that there is probably something wrong with the capture. Or maybe the source wasn't really 24p.

[nMaib0]

Alright, thanks for helping me with my question. is there a way to make the motion smoother with avisynth?

[jagabo]

I recommend you post a short sample with a long medium speed panning shot for analysis.

[Brad]

Don't know whether this one works the same way as some of AVerMedia's other capture devices, but:

The ExtremeCap U3 has a ridiculous bug where 1080p59.94 is captured as 1080p60.00 with duplicate frames added. I guess it's a bug across all of AVerMedia's hardware, as the C127 card I have is supposed to capture "1080p60 as 1080p30" but instead of just dropping half the frames from 1080p59.94 it creates a 1080p30.00 video. The problem may be unnoticeable to most in direct viewing, but for example if you have a PS3 playing back 23.976 -> 59.940 -> 30.000 it's impossible to recover the original 23.976 frame rate. (I mention this because my own reason for buying the card was recording Vudu. I can just record 1080p23.976 Vudu direct from my WD TV, but I can't download my purchases to hard drive prior to capturing like I can on PS3. So I have to rely on my internet connection to keep up with streaming.)

https://originaltrilogy.com/topic/How-to-capture-HDCP-encrypted-HDMI-sources-Vudu-Netf.../page/5#717216

[nMaib0]

I recommend you post a short sample with a long medium speed panning shot for analysis.

Sorry for the delay, see, in these videos you can see that there is a little bump every certain amount of time, as if some frames were missing. And I can't help but wonder how this device can detect out of 60 fps which ones are duplicated and which ones are not and produce a 30fps video that includes all the frames that matter. my guess is that out of the 60 that the signal carries 30 are taken in and other 30 ignored, and these 30 carry these missing frames, because it's not an issue of quality, I have captured at 10 mbps and at 40 and the result is the same, a slight bump noticeable in scenes with slow motion scenes. on the other hand, I have used tdecimate() with avisynth and the resulting video with no dupes is a perfect 23.976 fps video.

https://mega.nz/#F!fVJHHIxS!fCAoihzUyQUSmT4KU5AqbA


I don't understand why it is smoother when I capture at 60fps if the video I am capturing is 23.976 fps. The problem is my capture card captures 60fps only at 720p.

[poisondeathray]

I recommend you post a short sample with a long medium speed panning shot for analysis.

Sorry for the delay, see, in these videos you can see that there is a little bump every certain amount of time, as if some frames were missing. And I can't help but wonder how this device can detect out of 60 fps which ones are duplicated and which ones are not and produce a 30fps video that includes all the frames that matter. my guess is that out of the 60 that the signal carries 30 are taken in and other 30 ignored, and these 30 carry these missing frames, because it's not an issue of quality, I have captured at 10 mbps and at 40 and the result is the same, a slight bump noticeable in scenes with slow motion scenes. on the other hand, I have used tdecimate() with avisynth and the resulting video with no dupes is a perfect 23.976 fps video.

https://mega.nz/#F!fVJHHIxS!fCAoihzUyQUSmT4KU5AqbA

These were already processed, 10bit 4:2:0 HEVC, 23.976fps

How did you get from original to that ?

Check in the undecimated capture that the frames are really missing, and it's not the result of wrong decimation or improper processing afterwards

If they are present, maybe you need to use a bigger "window" for the decimation ? such as 2 in 10, 3 in 15, 4 in 20, etc... This can help if the duplicate is mistimed or slightly out of order

[nMaib0]

This is a clip from the original video
https://mega.nz/#!rRYFxABT!aaQ99mvvq8mZHlGrcNo5h4SlHHXUdRX3ABagp5xjhWc

this I used this script

LoadPlugin("C:\AvsPmod\tools\mvtools2.dll")
LoadPlugin("C:\AvsPmod\tools\TIVTC.dll")
LoadPlugin("C:\AvsPmod\tools\neo-f3kdb.dll")
LoadPlugin("C:\AvsPmod\ffms2.dll")
s="F:\1y2.mp4"
v=FFVideoSource(s)
a=FFAudioSource(s)
audiodub(v, a)
trim(809,123180)
tdecimate()
Sharpen(0.2)
super = MSuper(pel=2, sharp=1)
backward_vec2 = MAnalyse(super, isb = true, delta = 2, overlap=4)
backward_vec1 = MAnalyse(super, isb = true, delta = 1, overlap=4)
forward_vec1 = MAnalyse(super, isb = false, delta = 1, overlap=4)
forward_vec2 = MAnalyse(super, isb = false, delta = 2, overlap=4)
MDegrain2(super, backward_vec1,forward_vec1,backward_vec2,forward_vec2,thSAD=400)
neo_f3kdb()

[jagabo]

Your video was a 25p PAL source (24p film sped up to 25p) captured at 29.97p. One out of every 6 frames is a duplicate (except every 1001 frames). Use TDecimate(Cycle=6, CycleR=1) to return it to 25p. Actually that returns it to 24.975 fps but that remaining "one out of every 1001 frames" can be elusive. You can see the error in this clip just after the camera starts panning over the newspaper clippings (one jerk). Usin mode 2 and specifying 25 fps instead: TDecimate(mode=2, rate=25.0). But over the entire movie you may still end up with occasional jerks either way.

[nMaib0]

So by using the default mode I was removing more frames than I actually needed right? Thanks I am encoding the video now again and will check it in a few hours.

by the way how could you tell?

[nMaib0]

I don't mind not having it at 25p exactly, if you tell me TDecimate(mode=2, rate=25.0) is better then I will use that one instead.

[nMaib0]

Your video was a 25p PAL source (24p film sped up to 25p) captured at 29.97p. One out of every 6 frames is a duplicate (except every 1001 frames). Use TDecimate(Cycle=6, CycleR=1) to return it to 25p. Actually that returns it to 24.975 fps but that remaining "one out of every 1001 frames" can be elusive. You can see the error in this clip just after the camera starts panning over the newspaper clippings (one jerk). Usin mode 2 and specifying 25 fps instead: TDecimate(mode=2, rate=25.0). But over the entire movie you may still end up with occasional jerks either way.

which one should I use?

[jagabo]

So by using the default mode I was removing more frames than I actually needed right?

Yes.

by the way how could you tell?

First I took your word for it that the clip was originally 24p and encoded it with just TDecimate(). I saw it was jerky. So I looked at the original clip in an editor and saw that there was a duplicate every 6 frames (5 frames became 6 by duplicating one), not every 5 frames (4 frames became 5 by duplicating one). That meant the original 25p not 24p.

I don't mind not having it at 25p exactly, if you tell me TDecimate(mode=2, rate=25.0) is better then I will use that one instead.

In theory the latter is better for 25 sped up to 59.94p. But reliably detecting that one frame difference between 59.94p and 60p is difficult. It may or may not work better than the former. What that short clip it did.

[nMaib0]

Jagabom thanks for everything man you really helped me.