Editing lossless (or near lossless) video without losing quality

ItsClint wrote on 8/18/2017, 3:38 PM

Hello everyone, this is Clint, first post on these boards. I record videos from my screen via OBS, using NVENC, and the resulting quality videos are pretty high. Also Overwatch introduced a function that records your best moments in a game (highlight) at the settings you want, even if your system can't run them. 4k60 all ultra for example.

 

Point being, I have a bunch of very high quality videos that I'd like to edit. But I don't really wanna lose too much quality. Which project / render settings should I use in order to keep quality as close as my source as possible? I don't mind if the filesize gets even higher than my source, whatever. I want to keep that quality though (also unrelated, if any of you know any good handbrake settings to keep quality but still reduce filesize a bit, let me know).

 

TL;DR: How to edit near-lossless videos and render them without losing (too much) quality? Preferably close to indistinguishable from the source.

 

Info:

Vegas Pro 14

Windows 7 64bit Ultimate

i5 3570k @3.9 - 4 GHz

MSi GTX 670 Power Edition OC 2GB

Comments

Musicvid wrote on 8/18/2017, 5:17 PM

But you haven't told us your source video properties or your intended use; do you want lossless video or playable video? Pick one.

Back up one page and follow the FAQ to provide EXACT Vegas and MediaInfo properties, and tell us what you plan to do with your edited material.

 

ItsClint wrote on 8/18/2017, 7:01 PM

But you haven't told us your source video properties or your intended use; do you want lossless video or playable video? Pick one.

Back up one page and follow the FAQ to provide EXACT Vegas and MediaInfo properties, and tell us what you plan to do with your edited material.

 

I am terribly sorry. You are absolutely right. First of all I should specify the video(s) aren't recorded as actually lossless, I am using a CQP of 15 (not really distinguishable by eye from lossless), 1080p60 resolution and framerate. I plan on uploading to YouTube, while having the absolute best possible quality that YouTube can output. Lots of times, YouTube re-encodes the video and makes it lose quality. I'd like to avoid that.

I have seen absolutely gorgeous YouTube videos, with quality actually 1:1 to what I'd be seeing on my screen if I were playing the game myself, and I've never ever been able to reproduce nothing even closely comparable.

Now onto the MediaInfo properties:

This is a video that is automatically created by the game Overwatch; I didn't record it myself, there is an option to directly output an .mp4 file with the settings I desire. I will post a second MediaInfo with a recorded video too.

General
Complete name                            : D:\Tutto\Multimedia\Video\MP4\Misc\widow hs + mine kill no sound_17-08-17_02-02-29.mp4
Format                                   : MPEG-4
Format profile                           : Base Media / Version 2
Codec ID                                 : mp42 (mp41/isom)
File size                                : 170 MiB
Duration                                 : 24 s 650 ms
Overall bit rate                         : 57.8 Mb/s
Movie name                               : Widow hs + mine kill no sound

Video
ID                                       : 1
Format                                   : AVC
Format/Info                              : Advanced Video Codec
Format profile                           : Main@L4.2
Format settings, CABAC                   : No
Format settings, RefFrames               : 1 frame
Format settings, GOP                     : M=1, N=30
Codec ID                                 : avc1
Codec ID/Info                            : Advanced Video Coding
Duration                                 : 24 s 650 ms
Source duration                          : 24 s 650 ms
Bit rate                                 : 57.6 Mb/s
Width                                    : 1 920 pixels
Height                                   : 1 080 pixels
Display aspect ratio                     : 16:9
Frame rate mode                          : Constant
Frame rate                               : 60.000 FPS
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Progressive
Bits/(Pixel*Frame)                       : 0.463
Stream size                              : 169 MiB (100%)
Source stream size                       : 169 MiB (100%)
mdhd_Duration                            : 24650

Audio
ID                                       : 2
Format                                   : AAC
Format/Info                              : Advanced Audio Codec
Format profile                           : LC
Codec ID                                 : 40
Duration                                 : 24 s 650 ms
Source duration                          : 24 s 661 ms
Bit rate mode                            : Constant
Bit rate                                 : 192 kb/s
Channel(s)                               : 2 channels
Channel positions                        : Front: L R
Sampling rate                            : 48.0 kHz
Frame rate                               : 46.875 FPS (1024 SPF)
Compression mode                         : Lossy
Stream size                              : 589 KiB (0%)
Source stream size                       : 590 KiB (0%)
mdhd_Duration                            : 24661

Now onto the other type of file, meaning my NVEnc recordings; CQP should be around 15-20 for this one (can't remember). I record in .mkv format and remux the recording via the built-in OBS remux tool to .mp4 format. It's possible that 1 of the audio sources is muted; that is intended. I record 3 different tracks, 1 for game audio, 1 for my microphone, 1 for Discord / Teamspeak / whatever I am using to talk to people, or other things in general. So don't worry about that if it comes up in those infos :)

General
Complete name                            : D:\Tutto\Multimedia\Video\Recording\Overwatch\2017-08-18 06-54-15.mp4
Format                                   : MPEG-4
Format profile                           : Base Media
Codec ID                                 : isom (isom/iso2/avc1/mp41)
File size                                : 4.45 GiB
Duration                                 : 9 min 26 s
Overall bit rate                         : 67.5 Mb/s
Writing application                      : Lavf57.66.102

Video
ID                                       : 1
Format                                   : AVC
Format/Info                              : Advanced Video Codec
Format profile                           : High@L4.2
Format settings, CABAC                   : No
Format settings, RefFrames               : 2 frames
Codec ID                                 : avc1
Codec ID/Info                            : Advanced Video Coding
Duration                                 : 9 min 26 s
Bit rate                                 : 67.2 Mb/s
Width                                    : 1 920 pixels
Height                                   : 1 080 pixels
Display aspect ratio                     : 16:9
Frame rate mode                          : Variable
Frame rate                               : 60.000 FPS
Minimum frame rate                       : 58.851 FPS
Maximum frame rate                       : 1 024.000 FPS
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Progressive
Bits/(Pixel*Frame)                       : 0.540
Stream size                              : 4.43 GiB (100%)
Color range                              : Limited
Matrix coefficients                      : BT.470 System B, BT.470 System G

Audio #1
ID                                       : 2
Format                                   : AAC
Format/Info                              : Advanced Audio Codec
Format profile                           : LC
Codec ID                                 : 40
Duration                                 : 9 min 26 s
Bit rate mode                            : Constant
Bit rate                                 : 165 kb/s
Channel(s)                               : 2 channels
Channel positions                        : Front: L R
Sampling rate                            : 44.1 kHz
Frame rate                               : 43.066 FPS (1024 SPF)
Compression mode                         : Lossy
Stream size                              : 11.2 MiB (0%)
Default                                  : Yes
Alternate group                          : 1

Audio #2
ID                                       : 3
Format                                   : AAC
Format/Info                              : Advanced Audio Codec
Format profile                           : LC
Codec ID                                 : 40
Duration                                 : 9 min 26 s
Bit rate mode                            : Constant
Bit rate                                 : 2 089 b/s
Channel(s)                               : 2 channels
Channel positions                        : Front: L R
Sampling rate                            : 44.1 kHz
Frame rate                               : 43.066 FPS (1024 SPF)
Compression mode                         : Lossy
Stream size                              : 145 KiB (0%)
Default                                  : No
Alternate group                          : 1

Audio #3
ID                                       : 4
Format                                   : AAC
Format/Info                              : Advanced Audio Codec
Format profile                           : LC
Codec ID                                 : 40
Duration                                 : 9 min 26 s
Bit rate mode                            : Constant
Bit rate                                 : 70.4 kb/s
Channel(s)                               : 2 channels
Channel positions                        : Front: L R
Sampling rate                            : 44.1 kHz
Frame rate                               : 43.066 FPS (1024 SPF)
Compression mode                         : Lossy
Stream size                              : 4.75 MiB (0%)
Default                                  : No
Alternate group                          : 1

You asked me if I want a playable or lossless video, I'll say playable of course, I don't mind if the video is considered as lossy, I'd just like the image quality to be as close to an actually lossless video, with YouTube limitations in mind of course (which I am not fully aware of, to be honest). I am in your hands.

I will update with the Vegas 14 properties of the 2 aforementioned videos in a sec as an edit to this post.

Musicvid wrote on 8/18/2017, 7:22 PM

You need not worry about losing quality during editing and rendering.

YouTube will mangle it sufficiently through upstream processing to make anything you do seem like the tail trying to wag the dog.

1080p 60 at 12-16 Mbps is plenty considering what YT is going to do with it.

 

ItsClint wrote on 8/18/2017, 9:37 PM

You need not worry about losing quality during editing and rendering.

YouTube will mangle it sufficiently through upstream processing to make anything you do seem like the tail trying to wag the dog.

1080p 60 at 12-16 Mbps is plenty considering what YT is going to do with it.

 

First of all, I apologize for the late reply. My internet was gone. These are the properties of the first video (Overwatch output).

General
  Name: widow hs + mine kill no sound_17-08-17_02-02-29.mp4
  Folder: D:\Tutto\Multimedia\Video\MP4\Misc
  Type: AVC
  Size: 173,85 MB (178.023.916 bytes)
  Created: giovedì 17 agosto 2017, 02:02:29
  Modified: giovedì 17 agosto 2017, 02:03:45
  Accessed: giovedì 17 agosto 2017, 02:02:29
  Attributes: Archive

Streams
  Video: 00:00:24,633, 60,000 fps progressive, 1920x1080x12, AVC
  Audio: 00:00:24,650, 48.000 Hz; Stereo, AAC

ACID information
  ACID chunk: no
  Stretch chunk: no
  Stretch list: no
  Stretch info2: no
  Beat markers: no
  Detected beats: no

Other metadata
  Regions/markers: no
  Command markers: no

Media manager
  Media tags: no

Plug-In
  Name: compoundplug.dll
  Folder: C:\Program Files\VEGAS\VEGAS Pro 14.0\FileIO Plug-Ins\compoundplug
  Format: AVC
  Version: Version 14.0 (Build 270) 64-bit
  Company: MAGIX Computer Products Intl. Co.

 

While these are the OBS recorded video properties:

 

General
  Name: 2017-08-18 06-54-15.mp4
  Folder: D:\Tutto\Multimedia\Video\Recording\Overwatch
  Type: AVC
  Size: 4,67 GB (4.780.089.049 bytes)
  Created: sabato 19 agosto 2017, 01:32:45
  Modified: sabato 19 agosto 2017, 01:34:36
  Accessed: sabato 19 agosto 2017, 01:32:45
  Attributes: Archive

Streams
  Video: 00:09:26,750, 60,000 fps progressive, 1920x1080x12, AVC
  Audio 1: 00:09:26,776, 44.100 Hz; Stereo, AAC
  Audio 2: 00:09:26,776, 44.100 Hz; Stereo, AAC
  Audio 3: 00:09:26,776, 44.100 Hz; Stereo, AAC

ACID information
  ACID chunk: no
  Stretch chunk: no
  Stretch list: no
  Stretch info2: no
  Beat markers: no
  Detected beats: no

Other metadata
  Regions/markers: no
  Command markers: no

Media manager
  Media tags: no

Plug-In
  Name: compoundplug.dll
  Folder: C:\Program Files\VEGAS\VEGAS Pro 14.0\FileIO Plug-Ins\compoundplug
  Format: AVC
  Version: Version 14.0 (Build 270) 64-bit
  Company: MAGIX Computer Products Intl. Co.

 

Now that that's out of the way, I'd like you to check out this timestamp on this particular video: and watch for about 10 to 15 seconds if you could (edit: I linked it at about 15 seconds, but linking a YouTube video shows the preview apparently, so it starts at 0. Well, basically the same thing). The quality of the borders (antialiasing?) looks impeccable, it's all very detailed, the image quality is glorious. The only minor issue is the greenscreen, but that's only if you go fullscreen and even then, it's barely noticeable (and even then, I don't think he applied the filter via Vegas, there's a built in chroma key in OBS).

 

> YouTube will mangle it sufficiently through upstream processing to make anything you do seem like the tail

> trying to wag the dog.

 

I do see what you mean, but I really wanna try and optimize every setting I can. You mentioned the bitrate, that's ok, I hoped YouTube could handle more, but oh well. What about other settings though?

Render settings, for instance. I don't know which codec to use. I don't have a preference, I'd just want to use the one for best YouTube delivery. I see the generally recommended ones are Sony AVC/MVC with the Internet 1080p preset (but everyone gives mixed recommendations about the actual settings within the preset) or the MainConcept AVC/AAC Internet 1080p preset (same thing as the Sony preset, mixed recommendations).

I know about disabling the resample, but what should I use for project properties and render settings? More specifically which codec / preset / internal settings to customize the preset?

 

Thanks in advance.

Musicvid wrote on 8/18/2017, 11:20 PM

"Optimizing" settings works up to a point -- unfortunately with YouTube that's a pretty low ceiling.

Getting the basic settings right is the most important thing.

ItsClint wrote on 8/18/2017, 11:30 PM

"Optimizing" settings works up to a point -- unfortunately with YouTube that's a pretty low ceiling.

Getting the basic settings right is the most important thing.

Currently on my phone; are you suggesting I should post my current settings? If so, lend me a couple minutes, turning on my machine to grab some screenshots :)

Otherwise, please explain.

Here's a pic of my settings (project on the left, mainconcept render settings other 3 pictures. Skipped system tab, CUDA is available)

I used the Internet 1080p preset and edited that one to what I feel would be good settings. But since I'm quite a noob in this field, I'd love to ask to make absolutely sure I'm doing the right thing (I checked some guides and tried to make the best of them, but I'm still very confused).

 

EDIT: I found this: https://support.google.com/youtube/answer/1722171?hl=en

Those are the recommended settings that youtube provides. Do you think these would yield the best quality for youtube? Assuming it's still accurate, this should reduce the amount of shenanigans youtube does to the video, so it would resemble my original video more, isn't it? Apart from bitrate I suppose, which should stay on the high side (I guess?).

Musicvid wrote on 8/19/2017, 8:13 AM

The "amount of shenanigans" YouTube does is not under your control to any determinable extent. The recommended settings are a fine place to start.

Please feel free to conduct your own tests and post back when you have questions of substance.

 

NickHope wrote on 8/19/2017, 10:13 AM

https://www.vegascreativesoftware.info/us/forum/faq-how-can-i-improve-the-quality-of-my-avc-h-264-renders--104642/

ItsClint wrote on 8/19/2017, 6:26 PM

The "amount of shenanigans" YouTube does is not under your control to any determinable extent. The recommended settings are a fine place to start.

Please feel free to conduct your own tests and post back when you have questions of substance.

 

Alright, will do.

https://www.vegascreativesoftware.info/us/forum/faq-how-can-i-improve-the-quality-of-my-avc-h-264-renders--104642/

Thanks for that link! I noticed it says to use 59.94 fps. I've always set "60" not "59.94". I've never really understood 29.97 / 59.94, but is that relevant for "digital" content too? I since changed to 59.94 just to make sure; until I know more, at least. Please enlighten me, I can't find anything via google.

Also, I am currently uploading two videos, one is just the Overwatch output (function to export to highest quality .mp4, still not lossless but as close as it gets), the other is the Overwatch output rendered via Vegas Pro with the settings I'm using right now (with no editing, just to test image quality difference). I will report later with the results (and youtube links too if you would like to check them out yourselves and see if I made any mistakes with the rendering, which I would appreciate).

Thanks again with all the help thus far.

Musicvid wrote on 8/19/2017, 6:44 PM

It depends on your source. Set your frame rate the same as your source.

Record your video at constant, not variable framerate ( use your controls to determine this, not MediaInfo.)

Your source properties all say 60.000 fps. That is probably because it was recorded off a screen app rather than a camera.

If you set it at 59.940, which is NTSC, one out of 1000 frames will be duped, blended, or interpolated, depending on how your render properties are set. You don't particularly want that, although you'll never notice which is the surrogate frame.

Please, match your media settings in your project and render. That will save a lot of questions and confusion down the road.

 

ItsClint wrote on 8/19/2017, 7:45 PM

It depends on your source. Set your frame rate the same as your source!

Your source properties all say 60.000 fps. That is probably because it was recorded off a screen app rather than a camera.

If you set it at 59.976, which is NTSC, one out of 1000 frames will be duped, blended, or interpolated, depending on how your render properties are set. You don't particularly want that, although you'll never notice which is the surrogate frame.

Please, match your media settings in your project and render. That will save a lot of questions and confusion down the road.

 

I see, that explains it! Sorry if that sounded silly, I really couldn't find any info on whether I should use one or the other. Glad it's settled. Changed settings back to 60.000 fps.

I'd like to mention something; I noticed that the two MediaInfo reports for the two files (direct Overwatch "raw export", Vegas Pro 14 render) have different properties; namely, GOP. It's not there in the Vegas render's info, and it's m=1 n=30 in the raw vid's info.

How do I set GOP in Vegas' "Render As" menu? Is it even possible via Vegas? YouTube recommends a "closed" GOP.

I was also reading into using Vegas2Handbrake, since I normally use Handbrake to re-encode for smaller filesize. Is that still a good practice, or are there better ways / programs? I should have no problems installing it, just asking if it's still relevant or there are better ways.

 

This is the raw footage (from Overwatch) youtube link:

mega link: https://mega.nz/#!hkYFRBjS!Jqb7lsj7om4cXxeXZ8MKP5uvyCLXM9sLNcqiw9R8YEI (190MB)

With raw footage, I mean the file that gets spit out from Overwatch, not me recording the screen. Not gone through Vegas. The file itself looks gorgeous, youtube on the other hand... My rendered video with these (http://i.imgur.com/0vvTath.png) settings is almost the same as the source file while viewing it on my computer, I haven't tried uploading to youtube but I'd assume it's going to look basically the same as the raw footage upload, which still isn't that good. Is there a way of re-encoding this youtube upload in a way that would make it resemble the original file (on my pc) more, or is this a no-no? Is this the best quality I'm ever gonna get? I feel it's not, compared to, for example, this:

Is it just me, or does that look better? It could be me imagining things, that's the reason I'm asking. I just want to rest peacefully knowing that's the best I can do, or work hard to achieve a better quality.

Quick recap: Source "raw footage" and source rendered/encoded via Vegas look very very similar, if not the same. Uploaded "raw footage" looks like the bold a bit above. I haven't uploaded the rendered/encoded source file, but it's going to look very similar (I uploaded a different one with the same settings, but editing it a bit. The quality looks the same as the raw uploaded footage.)

I'm sorry if I'm being annoying with these questions, I just really want to learn, and hopefully help others that struggle with the same issue; I might (read: will) record a tutorial for this after we're done, to help other fellow Overwatch players in their efforts to upload good quality videos.

Thanks in advance.

TL;DR is bolded.

Musicvid wrote on 8/19/2017, 9:27 PM

It's time for you to go out there and make some mistakes. Lots of them - - they're cheap!

Best of luck. I get paid to teach, and it starts all over again in two days. My style is explorational, not directorial, if you catch my meaning.

NickHope wrote on 8/19/2017, 10:42 PM

I see, that explains it! Sorry if that sounded silly, I really couldn't find any info on whether I should use one or the other. Glad it's settled. Changed settings back to 60.000 fps.

The 2nd link in the link I posted above points to this: https://www.vegascreativesoftware.info/us/forum/faq-why-does-my-rendered-video-look-bad-troubleshooting-quality--103361/ The first section is all about matching your project properties to your media properties, which is very important.

I'd like to mention something; I noticed that the two MediaInfo reports for the two files (direct Overwatch "raw export", Vegas Pro 14 render) have different properties; namely, GOP. It's not there in the Vegas render's info, and it's m=1 n=30 in the raw vid's info.

How do I set GOP in Vegas' "Render As" menu? Is it even possible via Vegas? YouTube recommends a "closed" GOP.

GOP stands for "group of pictures". It's a termed used for inter-frame compression. The first frame of a GOP is fully described and known as an I frame. The remaining frames in the GOP describe the only differences from that I frame, so that file size can be saved. The converse of this is intra-frame compression (sometimes known as "all intra") whereby each frame is fully described. That's a very over-simplified explanation which doesn't touch on P frames, B frames etc.. If you're not doing "all intra" compression then I suggest not interfering with GOP settings and letting the encoder work out the best settings for you. Just concentrate on the quality value (usually termed "CRF" or just "RF" for x264 encoding) and encoder preset value, or the bitrate if you choose to take that approach.

I was also reading into using Vegas2Handbrake, since I normally use Handbrake to re-encode for smaller filesize. Is that still a good practice, or are there better ways / programs? I should have no problems installing it, just asking if it's still relevant or there are better ways.

There are some alternative x264 encoders listed in the 2nd paragraph of section 4 of that link. The advantage of Vegas2Handbrake is that it's quicker/simpler to execute a render directly from the timeline, and it deals with the glitch at the end of audio streams introduced by Frameserver. So at this time it's still a good solution, but a bit fiddly to set up. If you want to render directly from media, rather than the timeline, for example by rendering out a lossless intermediate file, then you could use Handbrake directly or try one of those other front ends for x264 instead.

I'm a little confused with the rest of your post about exactly what videos you'd like us to compare. Anyway I suggest you give Vegas2Handbrake a try. Set the "Encoder Preset" slider to "slower" and RF to 18 and see how that looks. You can of course go less lossy than that but you're into diminishing returns and it won't really help the final result on YouTube. Read my link and the other posts it links to again carefully so you don't drop a cod somewhere in your workflow.

Also you could try the 2 "tricks" explained in Walsh's comment here to try and force VP9 renders of your video by YouTube, which allegedly look better than H.264 renders. I haven't tested those tricks myself and they may be red herrings, but you never know. Lots of discussion about that on Reddit etc. if you go fishing (example).

ItsClint wrote on 8/20/2017, 1:29 AM

It's time for you to go out there and make some mistakes. Lots of them - - they're cheap!

Best of luck. I get paid to teach, and it starts all over again in two days. My style is explorational, not directorial, if you catch my meaning.

Thing is, mistakes for me aren't that cheap. Uploading takes a long, long time, and I can't really see a difference if the file isn't on YouTube. So yes, I can try lots of different presets and settings (which I have done, I am doing and will continue to do) but it won't really matter, because I wouldn't be able to upload them all in a reasonable amount of time. My upload speed is very limited, reason why I also started using Handbrake in the first place.

I see though. Thanks for your time thus far.

I see, that explains it! Sorry if that sounded silly, I really couldn't find any info on whether I should use one or the other. Glad it's settled. Changed settings back to 60.000 fps.

The 2nd link in the link I posted above points to this: https://www.vegascreativesoftware.info/us/forum/faq-why-does-my-rendered-video-look-bad-troubleshooting-quality--103361/ The first section is all about matching your project properties to your media properties, which is very important.

I'd like to mention something; I noticed that the two MediaInfo reports for the two files (direct Overwatch "raw export", Vegas Pro 14 render) have different properties; namely, GOP. It's not there in the Vegas render's info, and it's m=1 n=30 in the raw vid's info.

How do I set GOP in Vegas' "Render As" menu? Is it even possible via Vegas? YouTube recommends a "closed" GOP.

GOP stands for "group of pictures". It's a termed used for inter-frame compression. The first frame of a GOP is fully described and known as an I frame. The remaining frames in the GOP describe the only differences from that I frame, so that file size can be saved. The converse of this is intra-frame compression (sometimes known as "all intra") whereby each frame is fully described. That's a very over-simplified explanation which doesn't touch on P frames, B frames etc.. If you're not doing "all intra" compression then I suggest not interfering with GOP settings and letting the encoder work out the best settings for you. Just concentrate on the quality value (usually termed "CRF" or just "RF" for x264 encoding) and encoder preset value, or the bitrate if you choose to take that approach.

I was also reading into using Vegas2Handbrake, since I normally use Handbrake to re-encode for smaller filesize. Is that still a good practice, or are there better ways / programs? I should have no problems installing it, just asking if it's still relevant or there are better ways.

There are some alternative x264 encoders listed in the 2nd paragraph of section 4 of that link. The advantage of Vegas2Handbrake is that it's quicker/simpler to execute a render directly from the timeline, and it deals with the glitch at the end of audio streams introduced by Frameserver. So at this time it's still a good solution, but a bit fiddly to set up. If you want to render directly from media, rather than the timeline, for example by rendering out a lossless intermediate file, then you could use Handbrake directly or try one of those other front ends for x264 instead.

I'm a little confused with the rest of your post about exactly what videos you'd like us to compare. Anyway I suggest you give Vegas2Handbrake a try. Set the "Encoder Preset" slider to "slower" and RF to 18 and see how that looks. You can of course go less lossy than that but you're into diminishing returns and it won't really help the final result on YouTube. Read my link and the other posts it links to again carefully so you don't drop a cod somewhere in your workflow.

Also you could try the 2 "tricks" explained in Walsh's comment here to try and force VP9 renders of your video by YouTube, which allegedly look better than H.264 renders. I haven't tested those tricks myself and they may be red herrings, but you never know. Lots of discussion about that on Reddit etc. if you go fishing (example).

Thanks for the very detailed reply Nick!

I went through the links you posted again, and made sure all my settings comply to the guide(s).

These are my current preset / render settings: http://i.imgur.com/0vvTath.png (render quality on best, audio is 384kbps 48khz AAC, couldn't bother screenshotting - most info is there either way)

Are those fine, in your opinion? The output video on my PC looks fine. The issue is YouTube...

Now let me reply to what I'd like you to compare.

I uploaded this video: which is a file I produced via the game Overwatch; it has a function that lets you create a video file from the equivalent of a demo file (which have been in other games for ages, like CS:GO. Very new to Blizzard's Overwatch though). The output produced is 1080p60, bitrate is monstrously high, and the video itself is basically perfect to the eye, although lossy by definition (AVC), it's probably impossible for the human eye to distinguish it from a lossless. Except... YouTube screws it up.

Firstly I was wondering if re-encoding the video changing some of its settings (with any method that you would deem appropriate, like we both mentioned Vegas2Handbrake) would be beneficial to its final YouTube quality, since in its current state it is very, very far from the file played on my PC. Which is this MEGA link by the way: https://mega.nz/#!hkYFRBjS!Jqb7lsj7om4cXxeXZ8MKP5uvyCLXM9sLNcqiw9R8YEI (200MB approx.)

The difference is honestly very large. This is what I would love for you to check. It's crazy for me to think about that such quality cannot even come close to be matched on YouTube, while other people have managed to do it (Fitz for example, variety YouTuber; another Overwatch example similar to the one I posted above). There must be a way, I'm here trying to figure out how. Your help would surely make this search quicker, you seem to know a lot more than I could hope to understand. I have mild experience with IT, recording on the other hand... I feel like a fish out of water, mostly.

From my limited knowledge, the source quality of the media on my PC, trying lots of different combinations of render settings, didn't change all that much; the problem isn't the file on my PC, it's what YouTube does to it, I'm pretty sure.

You mentioned VP9, and who would've thought, one of my most widely viewed videos did get re-encoded (is that the right word? I've been using it in my head that way) in VP9. It apparently has something to do with popularity, because only a few select videos of mine got that treatment, mostly highly viewed / commented ones, or others picked at random (apparently. Haven't noticed a pattern if it's there). I also noticed that the video(s) from other youtubers which I used for comparison are in fact encoded in VP9! That would explain lots of things. I checked out the links about VP9, but I don't have the slider in the enhancements tab, the video just repeats over and over. Do you have it?

Should I consider rendering in 1440p my 1080p file? People seem to have gotten the dreaded VP9 that way.

I will try the Google method later.

I'll go through the installation process of Vegas2Handbrake, reporting back later on as well.

That's all from me for a bit, I need to rest.

Thanks again for your time and your very detailed answers! As a curious person and learner I love reading those.

NickHope wrote on 8/20/2017, 3:34 AM

I downloaded your 1080-60p Overwatch file and the highest quality file that YouTubeByClick would give me, which was 1080-30p. Actually its true frame rate seems to be not exactly half of the original 60p file, as you can see by the frames not lining up when you put both files on separate tracks the Vegas timeline. The quality has certainly taken a big hit. Lots of detail lost. I don't know of a current way to download YouTube's 1080-60p file for an A/B comparison on the Vegas timeline. The methods I used to use for downloading YouTube videos no longer work.

I suspect that this VP9 business might be key, but I'm afraid I don't even know how to find out if a video is being delivered by YouTube as H.264 or VP9. Would be interested to learn if anyone knows.

ItsClint wrote on 8/20/2017, 3:54 AM

I downloaded your 1080-60p Overwatch file and the highest quality file that YouTubeByClick would give me, which was 1080-30p. Actually its true frame rate seems to be not exactly half of the original 60p file, as you can see by the frames not lining up when you put both files on separate tracks the Vegas timeline. The quality has certainly taken a big hit. Lots of detail lost. I don't know of a current way to download YouTube's 1080-60p file for an A/B comparison on the Vegas timeline. The methods I used to use for downloading YouTube videos no longer work.

I suspect that this VP9 business might be key, but I'm afraid I don't even know how to find out if a video is being delivered by YouTube as H.264 or VP9. Would be interested to learn if anyone knows.

Ahh, thank you, always spot on.

I'm glad I'm not the only one that thinks quality is very different. Thanks for taking time to check it out actually :)

There are a number of ways to download videos from YouTube in highest quality, I remember trying the Vlc method listed here and it worked. https://www.reddit.com/r/youtube/comments/3k13cb/the_best_way_to_download_youtube_videos/

Some also recommend jdownloader (use this link! https://board.jdownloader.org/showthread.php?t=54725 The official site - or, well, the first result on google, which I assume would be the official - has been reported as infected by malware), I have never used it for downloading YouTube videos, but I bet it would work just fine (someone on Reddit, in that thread above, mentioned that the program let him download even higher resolution / fps combo than the ones available by viewing the video itself - can't confirm nor deny, but I think it would be possible, at least; unlikely, yeah, but possible).

I'm not currently at my PC, I will download my own video and upload that one too, if that works better for you :)

Edit: found this other one too, looks very promising. https://www.reddit.com/r/DataHoarder/comments/5wddlb/download_youtube_videos_with_jdownloader/

Edit2: forgot to mention; you can check the codec by right clicking on the YouTube video player and selecting "stats for nerds" or whatever it says in English (I'm Italian, not sure about the word by word translation in English). Codec is one of the entries there, along with bitrate and other... Nerd stuff. Ahah.

Edit3: Here's the link for the 1080p60 video straight from the youtube stream. I used jdownloader2 (from the link I provided above) which automatically installed ffmpeg for me (which I hadn't reinstalled on this ssd) and let me choose the quality. Sure enough, 1080p60 was there. Mega link: https://mega.nz/#!ggIVSaRZ!ra5olvN6yaTJxkOKZy8jIid9yMEzRihoDFT6wNDt8U0

Edit4: I found this, which is an Adobe Premiere plugin for exporting videos in the VP9 codec directly! If only it was for Vegas... I might just have to ask my friend who has a copy of Premiere to try and render this out for me. It looks very well made and full of functionality, although, admittedly, there are many, many arguments that I would have to learn to truly maximize quality/compression/encoding speed etc.

Link for the plug-in: http://www.fnordware.com/WebM/

Now that I think about it, the WebM team is behind it, so I shouldn't have doubted the outcome, ahah.

Edit5: I think I found VP9/WebM-compliant settings for YouTube, or at least inner workings of their encoding/re-encoding whatever they do. It's out of my comprehension for now, but I'm guessing you could probably make something out of this: https://developers.google.com/youtube/v3/live/guides/encoding-with-dash

It talks about encoding with dash, but it also covers, among other codecs, VP9, with some very advanced information that's out of my grasp. Could be rubbish for all I know, but I figured I'd link it.

Edit6: I cheated the system. I rendered a 1080p60 file in 1440p60 (2k60), YouTube encoded it as VP9 as soon as 1440p60 became available. Quality appears to be much better. This is a very dirty and cheap workaround, but it seems to work. What's your take, Nick? This was from my friend's house using Premiere, but I reckon I could do the same with Vegas, just matching the settings (which I noted in the description, so I won't forget).

Musicvid wrote on 8/21/2017, 7:51 PM

Golleez, this YouTube thing sure got complicated recently...

;?)

ItsClint wrote on 8/21/2017, 8:01 PM

Golleez, this YouTube thing sure got complicated recently...

;?)

It did, but I have a temporary solution, so I guess I'm good for a while. After all, Vegas was not the problem, nor my settings really. That's a relief. Thank you and Nick both for your useful insight.

I think this could be closed.

NickHope wrote on 8/22/2017, 5:35 AM
...you can check the codec by right clicking on the YouTube video player and selecting "stats for nerds" or whatever it says in English...

Thanks. In English it also says "Stats for nerds". Then for Mime Type it shows either video/webm; codecs="vp9" or video/mp4; codecs="avc1.64...something...". I might become a little obsessed with this feature!

Edit4: I found this, which is an Adobe Premiere plugin for exporting videos in the VP9 codec directly!...

I doubt the format that YouTube encode is affected by what format you send them, but I suppose it might be. That could be tested of course with a few test uploads.

I looked at my own YouTube videos. I always upload AVC/AAC MP4 encoded with an x264 encoder. On my more popular channel ( https://www.youtube.com/bubblevision ) I see that everything over the last few years is getting delivered as VP9. On my less popular channel ( https://www.youtube.com/nickhope ) the last couple are VP9 but before that they are mostly AVC. Maybe that channel has now reached some sort of "popularity threshold". That's a little evidence that perhaps YouTube discriminate a little depending on how popular they think a video will be, but far from conclusive. There could be all sorts of other variables involved.

Edit6: I cheated the system. I rendered a 1080p60 file in 1440p60 (2k60), YouTube encoded it as VP9 as soon as 1440p60 became available. Quality appears to be much better. This is a very dirty and cheap workaround, but it seems to work. What's your take, Nick?

I installed JDownloader from your link but it only offers me 1440-30p for this video, so I can't do an A/B comparison on the Vegas timeline. However the 1080p60 VP9 stream on YouTube is pretty clean and definitely looks WAY better than the 1080p60 AVC stream on the previous version. This trick of upscaling to 1440 really is a horrible, inelegant botch but worth it in this sort of case. If you can capture and edit in 1440 then that should be better still.

I took a quick look at the "Enhancements slider" trick from Walsh's post. Doesn't feel right to me, and I worry about even more quality loss if I press "Save". The Google Drive trick might be more worth trying.

I have my own YouTube account manager, although they're in the process of changing me from one to another. I will ask them if they can find out why some videos get VP9 and others don't.

Musicvid wrote on 8/26/2017, 12:59 PM

Heard a great quote from a local newscaster yesterday, in reference to Internet myth and legend in general:

"The plural of 'anecdote' is 'trend.'"

winrockpost wrote on 8/26/2017, 5:22 PM

really making this way more complicated than need be, simply render a small section to every way you think may make a difference on you tube ...upload and let your eyes decide

ItsClint wrote on 8/26/2017, 8:57 PM
...you can check the codec by right clicking on the YouTube video player and selecting "stats for nerds" or whatever it says in English...

Thanks. In English it also says "Stats for nerds". Then for Mime Type it shows either video/webm; codecs="vp9" or video/mp4; codecs="avc1.64...something...". I might become a little obsessed with this feature!

Edit4: I found this, which is an Adobe Premiere plugin for exporting videos in the VP9 codec directly!...

I doubt the format that YouTube encode is affected by what format you send them, but I suppose it might be. That could be tested of course with a few test uploads.

I looked at my own YouTube videos. I always upload AVC/AAC MP4 encoded with an x264 encoder. On my more popular channel ( https://www.youtube.com/bubblevision ) I see that everything over the last few years is getting delivered as VP9. On my less popular channel ( https://www.youtube.com/nickhope ) the last couple are VP9 but before that they are mostly AVC. Maybe that channel has now reached some sort of "popularity threshold". That's a little evidence that perhaps YouTube discriminate a little depending on how popular they think a video will be, but far from conclusive. There could be all sorts of other variables involved.

Edit6: I cheated the system. I rendered a 1080p60 file in 1440p60 (2k60), YouTube encoded it as VP9 as soon as 1440p60 became available. Quality appears to be much better. This is a very dirty and cheap workaround, but it seems to work. What's your take, Nick?

I installed JDownloader from your link but it only offers me 1440-30p for this video, so I can't do an A/B comparison on the Vegas timeline. However the 1080p60 VP9 stream on YouTube is pretty clean and definitely looks WAY better than the 1080p60 AVC stream on the previous version. This trick of upscaling to 1440 really is a horrible, inelegant botch but worth it in this sort of case. If you can capture and edit in 1440 then that should be better still.

I took a quick look at the "Enhancements slider" trick from Walsh's post. Doesn't feel right to me, and I worry about even more quality loss if I press "Save". The Google Drive trick might be more worth trying.

I have my own YouTube account manager, although they're in the process of changing me from one to another. I will ask them if they can find out why some videos get VP9 and others don't.

I wanted to quote you but I'm on my phone and I can't make a clean looking one (I tried, messing the quote up), it's going to be a bit long, whoops.

Thank you very much for your response and testings, Nick! By the way, there is a little arrow on jdownloader, you can select any video or audio quality from there, just FYI for future needs :)

I also feel that way about VP9. It must be because of popularity, if not of the channel, of the singular videos instead.

About enhancements tab, I see. I just hope I could try that myself on a dummy video to see the results, it could be my extensions on the browser. About that, which browser are you using? Because I tried Chrome and Firefox and neither allowed me to use that slider, it basically auto-plays the video (like a webm embedded file) and there is no slider. Just the classic webm player (not YouTube's). It may be due to the new design of YouTube (which is currently under way of rolling out, apparently), I will test that out too, I guess, ahah.

I see! That would be a great question for them for sure.

Heard a great quote from a local newscaster yesterday, in reference to Internet myth and legend in general:

"The plural of 'anecdote' is 'trend.'"

I really do not see the meaning behind this; what are you implying? I might just be oblivious or tired, but what I gather from this response is "VP9 and YouTube is a myth"...? I find that hard to understand. Also because some of these VP9 tricks are very much real. The 1440p one is very inelegant, but it does indeed work, so...

really making this way more complicated than need be, simply render a small section to every way you think may make a difference on you tube ...upload and let your eyes decide

I did that already! It's just I am a perfectionist; since I saw little to no change between every rendering options I chose, but every one of those turned out to be much, MUCH worse than many videos of the same content I tried to render-test, a question naturally came to mind: how? How does that look so much better? You might think it would be the source footage quality, but that was the equivalent of CQP/CRF of 1 quality, essentially lossless to the eye, although lossy by definition, as already mentioned. It turned out to be because of the way YouTube delivers videos that are either 1440p+ or popular enough: the VP9 codec, no more, no less.

It offers better quality at the same bitrate, and since YouTube tries to use the combination of smallest power/compression combination, they only compress 1440p+ (or highly popular) videos because VP9 takes longer than AVC to encode. Doing that for every video would slow them down, so they don't. The result is less slightly less appealing videos to the eye, for much speed gain. Difference is quite big in high motion videos though, noticeably high in 60 fps ones.

Now if there was a way to get that VP9 codec asking for it specifically... That wouldn't be something that your average uploader would know (or specifically want, for that matter), so no big loss there, but it would make the day of people that want that extra bit of quality, like myself.

Especially when the local video quality is impeccable and the YouTube video quality looks no better than a 480p upscale when there is high motion. That bothered me.

I hope I answered you all! I had to take a small break, sorry for the hiatus!

ItsClint wrote on 8/26/2017, 10:00 PM

Again, I've not seen quantified tests demonstrating various YouTube "tricks." I do believe you, but without some direct evidence (not "this one looks better"), it's just another good story. A perfectionist should understand that straightaway.

Claims that forcing VP9 on YT overcomes dither and quantization noise from bicubic upscaling sufficiently to realize a measurable advantage in the delivery is something I find skeptical until confronted with the numbers to back up that perception.

Which brings me back to that quote...

When I was first writing this response your edit wasn't there. Let me rewrite this then.

How would I go about getting "numbers" from such a comparison? Heck, how would I compare the two videos? Nick talked about doing an A/B comparison in Vegas, how does one do that? Google gave me this Wikipedia page talking about A/B in general, what about Vegas specifically? https://en.m.wikipedia.org/wiki/A/B_testing

And even then, what should I be comparing or showing with either screenshots or videos to prove what I am claiming? I am guessing it could very well be something out of my qualities to demonstrate, I kindly ask you to use your preferred method of downloading YouTube videos (or use links I provided if you trust 'em - or me, heh - enough) and comparing them yourself, because I don't know what I should be comparing. All I see is one looks less "blockish" when there are fast moving sections; I could screenshot the same frame in both videos and show that, I guess.

 

Just give me some time, I will post back here from my machine :) and please do let me know about how to do such a comparison, or if you're willing to test that yourself. I would gladly do a proper comparison or see your results, whichever you choose.

Thanks for your response, I will document myself about dithering and quantization.

Musicvid wrote on 8/26/2017, 10:11 PM

Again, I've not seen quantified tests demonstrating various YouTube "tricks." I do believe you, but without some direct evidence (not "this one looks better"), it's just another good story. A perfectionist should understand that straightaway.

All else being equal, the notion that VP9 retains better quality than x264 at the shamefully low bitrates YouTube serves has been tested and confirmed by others. But all else is not equal, is it?

Claims that forcing VP9 on YT overcomes dither and quantization noise from bicubic upscaling sufficiently to realize a measurable advantage in the delivery is something I remain skeptical about, until confronted with the data to back up those claims.

Which brings me back to that quote...