Shutter encoder advice ?

Gid wrote on 6/3/2024, 4:13 PM

Hi, last couple of wks I've been editing a few hundred of my phone clips from 2022, UHD AVC MP4 29.97 Variable without any issue, a couple I've converted to constant because of the 'stutter' that can happen when the phone pans around, apart from that no problem.

Just as a bit of a test to see if it would make much difference to editing & rendering last night I left Shutter Encoder to convert all the remaining 62 clips of this folder to constant, as is often recommended on here, it did them without prob,

But I just sat down to edit & 😡 I only added 3 clips to the timeline, about 5mins worth & those first few paused very slightly for just a fraction of a second while playing, backtracking the cursor it would play fine but then a little bit later maybe 10-30secs or so it would do it again on a different part just playing the clip. 12frame transitions needed at least one loop play to play fluidly... I thought what! what's going on?

Looking at the GOP of the converted to constant clips using Shutter encoder it shows a GOP of 120, & looks like this. Long GOP has been mentioned on here before, ?

The orig Variable clips that don't have this 'pause' issue look like this with a GOP of 30.

Am I right in thinking I just need to change this setting in Shutter Encoder (pic below) ? Like I say my phone clips play fine normally in Vegas so I rarely have to do this, only occasionally do I change 1 or 2 to constant & normally use Handbrake, Shutter encoder seemed to do it quicker when I tested both on a handful of clips, & I didn't want to convert to ProRes because these orig 62 MP4 clips are already 49GB in total, I'd hate to think what size the ProRes total size would be.

Thanks.

---------------

Example of the converted to constant pausing

---------------------

Just as a comparison this is the orig variable playing without pauses

Comments

mark-y wrote on 6/3/2024, 4:50 PM

GOP length of 120 will not trigger a nag message in Vegas, and although you will get a variety of opinions, that is a nominal length for AVC that works well for me. It is that uber-compressed stuff from Zoom meetings that causes problems and triggers caution messages, with its 900-frame GOPs!

It would be worthwhile to test your output at 30 frames vs. 120 to see if you can spot a visual difference. In certain benchmarks (SSIM), you may get slightly different results, most notably with high-motion, high complexity source, such as boat racing.

I noticed your image shows encoding is set to Interlaced. Of course it will be progressive for delivery.

Gid wrote on 6/3/2024, 5:00 PM

@mark-y Thanks, is it just that setting in Shutter encoder I need to change to get 30GOP?

As you can see the orig that play fine are 30GOP but variable frame rate so I was just trying it out. I do sometimes swing the camera around so converting all before starting editing would be good, then there would be very little if any of the stutter I mentioned when panning & I wouldn't have to go find that clip, convert it to constant & then use Vegas's Replace.

john_dennis wrote on 6/3/2024, 5:44 PM

Try these settings and report.

Gid wrote on 6/3/2024, 6:14 PM

@john_dennis Thankyou, I'l test this tomorrow, midnight now so off to bed now 😊👍

john_dennis wrote on 6/3/2024, 6:24 PM

@mark-y said: "It would be worthwhile to test your output at 30 frames vs. 120 to see if you can spot a visual difference."

Comparing GOP30 to GOP120 encoded from Shutter Encoder on the Vegas Pro 21-315 timeline showed some difference in the Video Scopes. I'm not able to run scientific tests today.

mark-y wrote on 6/3/2024, 7:38 PM

Try these settings and report.

Those settings are almost an exact match of what I use for "highest quality" x264 output not only in Shutter Encoder, but also Voukoder, HOS, and Handbrake.

@john_dennis and I have reached consensus on a number of things over the years; there is no big downside to 30 or even 15 GOP except for file size increase and possibly longer encode times, and by using the zerolatency tune, almost matches MPEG-2 structure. It's essentially Baseline with CABAC.

Using CRF 18 (I use 20), sometimes results in soaring bitrate bursts at times of complex motion, which can mess with streaming reception and playback, causing skips and jumps, perhaps garbled playback. In those cases, an alternative to a higher CRF is to employ the vbv-bufsize and vbv-maxrate x264 flags in combination, which can tame the bandwidth bursts to manageable server / diner levels, without sacrificing the average bitrate during the rest of your program.

RogerS wrote on 6/3/2024, 8:27 PM

Interesting about this use for zerolatency- I thought the purpose was for streaming but it seems like it's also a workaround for a simple encode.

mark-y wrote on 6/3/2024, 9:42 PM

zerolatency is really a dumbed-down Main Profile, using fewer compression tricks and suppressing b-frames. The huge compression advantage of CABAC (20% or more) is retained. Older versions of Vegas couldn't even support 8x8dct and weightp

fastdecode goes a step further by disabling CABAC as well, which takes us back to the earliest days of AVC, and the very first iPod players. You had to be there.

The following information is given by typing the command x264 --fullhelp

john_dennis wrote on 6/3/2024, 11:01 PM

Long, long ago, in a land far, far away, I had a camera that recorded FHD AVC/PCM in a MOV wrapper onto SD cards. The video was easily by decoded Vegas versions at the time without Quicktime.

General
Complete name                            : \\Z790\2016-10-24 Hunting\MVI_3035.MOV
Format                                   : MPEG-4
Format profile                           : QuickTime
Codec ID                                 : qt   2007.09 (qt  /CAEP)
File size                                : 71.7 MiB
Duration                                 : 17 s 17 ms
Overall bit rate                         : 35.3 Mb/s
Frame rate                               : 23.976 FPS
Encoded date                             : 2016-10-24 18:16:56 UTC
Tagged date                              : 2016-10-24 18:16:56 UTC
Copyright                                : 2014
com.apple.quicktime.make                 : Canon
com.apple.quicktime.model                : Canon PowerShot G15
com.apple.quicktime.rating.user          : 0.000
com.apple.quicktime.author               : John Dennis

Video
ID                                       : 1
Format                                   : AVC
Format/Info                              : Advanced Video Codec
Format profile                           : Constrained Baseline@L4.1
Format settings                          : 1 Ref Frames
Format settings, CABAC                   : No
Format settings, Reference frames        : 1 frame
Format settings, GOP                     : M=1, N=12
Codec ID                                 : avc1
Codec ID/Info                            : Advanced Video Coding
Duration                                 : 17 s 17 ms
Bit rate                                 : 33.8 Mb/s
Width                                    : 1 920 pixels
Height                                   : 1 080 pixels
Display aspect ratio                     : 16:9
Frame rate mode                          : Constant
Frame rate                               : 23.976 (24000/1001) FPS
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Progressive
Bits/(Pixel*Frame)                       : 0.679
Stream size                              : 68.5 MiB (96%)
Language                                 : English
Encoded date                             : 2016-10-24 18:16:56 UTC
Tagged date                              : 2016-10-24 18:16:56 UTC
Color range                              : Full
Color primaries                          : BT.709
Transfer characteristics                 : BT.709
Matrix coefficients                      : BT.709
Codec configuration box                  : avcC

Audio
ID                                       : 2
Format                                   : PCM
Format settings                          : Little / Signed
Codec ID                                 : sowt
Duration                                 : 17 s 17 ms
Bit rate mode                            : Constant
Bit rate                                 : 1 536 kb/s
Channel(s)                               : 2 channels
Channel layout                           : L R
Sampling rate                            : 48.0 kHz
Bit depth                                : 16 bits
Stream size                              : 3.12 MiB (4%)
Language                                 : English
Encoded date                             : 2016-10-24 18:16:56 UTC
Tagged date                              : 2016-10-24 18:16:56 UTC

I tried to replicate those characteristics as a means of creating files that were easily edited from weird source material that Vegas would not have otherwise been able to handle or handle well. One characteristic of the files was the absence of B-Frames. I posited that Canon tried to get the best video they could with the encoding hardware that was available at the time. I stumbled on the zerolatency setting without the depth of analysis that @mark-y described. I'm sure I read the FFMPEG documentation, but have long since forgotten it all along with many other things worth remembering.

Gid wrote on 6/4/2024, 7:09 AM

@john_dennis Thanks, I've tagged your comment as the Solution, I've had a quick play & as far as I can tell the file it produces plays as well as the orig, no pausing like before & should be better than the orig as with it being Constant it should stop any stuttering I mentioned when the camera pans, any faltering on the timeline is I suspect Vegas not the file itself 👍

This is after conversion with Shutter Encoder 👍

@mark-y

Those settings are almost an exact match of what I use for "highest quality" x264 output not only in Shutter Encoder, but also Voukoder, HOS, and Handbrake.

I tried setting up Handbrake in a similar way to Shutter Encoder, I got 30 GOP by adding the keyint=30,

This orig 1.15min clip -

Shutter encode - 3.08mins

Handbrake encode - 0.43secs 😮👍

So would both of you have a look at these & tell me if there's anything else I should change? These settings produce a file that to me looks fine but is half the file size of the orig & the encode from Handbrake is significantly faster, should that bother me?

Thankyou both 👍

gorGaram wrote on 6/4/2024, 7:19 AM

I'm sorry, but Shutter Encoder only shows the FIRST ONE GOP. Everyone else may not be like that at all. Just need to keep in mind.

john_dennis wrote on 6/4/2024, 7:33 AM

I don't see anything askew with your Handbrake settings. The difference in encode time is unexpected.

RogerS wrote on 6/4/2024, 8:00 AM

Does the resulting MediaInfo for each rendered file offer any clues?

Gid wrote on 6/4/2024, 8:07 AM

@gorGaram Thankyou I will 👍

@john_dennis Thankyou 👍

Having a problem with Shutter Encoder at the mo, I saved those Shutter Encoder settings as a preset but now the two files I've been trying, the 3min drilling one & the 1.15min one in the pics above both produce a 5.01min file from Shutter Encoder that plays slow mo video & sound. I deleted the preset & manually put those settings back in, didn't save as a preset but the same 5min slow mo is produced, I think Shutter Encoder is grumpy with me,

I need to go out so I'll clean & restart my PC then report back 👍

Thanks again.

-------------

@RogerS Hi sorry I deleted that comment,

These are before & after, I'm hoping it's just a cache,temp file thing..

Orig

5min slow mo encoded with Shutter Encoder.

& these were the settings used that created that slow mo, I'm hoping as I say the PC just needs a clean

-----------

This is that file on Google Drive

https://drive.google.com/file/d/1CJqf5sYhYUmjsO2xhppq1_OA2BAOk8Vl/view?usp=sharing

gorGaram wrote on 6/4/2024, 8:08 AM

@Gid
Incorrect setting in Shutter Encoder . If you remove it, everything is fast and correct. With your settings 1:34 and the video lasts for 5 minutes. With the correct settings, 25 seconds of rendering and the video\audio is correct.
Handbrake 1:09 correct and video\audio correct.
(I think 18 CQ is a bit much, but you know better)

gorGaram wrote on 6/4/2024, 8:13 AM

@Gid

The rest of the settings are yours

gorGaram wrote on 6/4/2024, 8:22 AM

Gid wrote on 6/4/2024, 8:28 AM

@gorGaram Sorry I deleted that comment asking to @ tag people, I'm going out & was just being a bit petty, I didn't know if you were talking to me or @john_dennis,

What does that picture mean?

gorGaram wrote on 6/4/2024, 8:30 AM

@gorGaram Sorry I deleted that comment asking to @ tag people, I'm going out & was just being a bit petty, I didn't know if you were talking to me or @john_dennis,

What does that picture mean?

How do I make a correct link to the person I'm talking to?

Dexcon wrote on 6/4/2024, 8:33 AM

@gorGaram

Hit the @ key on your keyboard and then select the recipient from the list that opens.

Gid wrote on 6/4/2024, 8:36 AM

@Gid
Incorrect setting in Shutter Encoder . If you remove it, everything is fast and correct. With your settings 1:34 and the video lasts for 5 minutes. With the correct settings, 25 seconds of rendering and the video\audio is correct.
Handbrake 1:09 correct and video\audio correct.
(I think 18 CQ is a bit much, but you know better)

@gorGaram Like you have in the comment I've quoted, why yours doesn't show as blue could be if you type @ & there is another letter immediately after it it won't work, like this @GorGaram. There needs to be a space after @ for it to work.

john_dennis wrote on 6/4/2024, 8:44 AM

@Gid

SWAG

We would probably get more predictable results if we were working with the same file(s). My conversion was on a file with a relatively constant frame rate even though it was marked as variable at 30 fps.

/SWAG

john_dennis wrote on 6/4/2024, 8:51 AM

@gorGaram

Gid wrote on 6/4/2024, 10:21 AM

Try these settings and report.

@john_dennis My OP with the video showing the pausing was a couple of clips on the timeline where I mainly showed the drilling one playing , that drilling one is an orig straight from the phone, everything in the comments after that about the encoder producing 5min clips was about the one at the start of the timeline (the orig unedited from the phone) I shared it in the comment 10 above this one named 20220530_141322 showing the room - This is that same link - https://drive.google.com/file/d/1CJqf5sYhYUmjsO2xhppq1_OA2BAOk8Vl/view?usp=sharing but this 5min thing seems to happen to both clips.

In your pic I've quoted above it says 8K, I shared that quite a while ago for whatever reason 🤷‍♂️, I never use 8K so I presumed you were using that just because you didn't at that time have a sample of what I'm showing now.