Encoding specs from AVI to Youtube?
 

Thanks to so many on this forum I'm finally making progress and having fun converting 100+ NTSC VHS broadcast TV tapes (1980's-2000's) to digital. It's been 7 years since I first found this group and there have been a few stops and starts. This time I decided to prioritize the capture and do the restoration later (the tapes won't last forever but digital should). I'd like to share some of what I have on YouTube, not for money but to share with friends.

Here are topics I think I know about but could use some advice. Please feel free to add to the list. Maybe this could be helpful to others just starting out.

My AVI captures are mostly very good thanks to LS help years ago to acquire the right workflow.

Almost everything I have is telecine (and I know how to tell) and of course interlaced. IVTC and QTGMC, when and where to apply for uploading to YouTube?

I've read here that you want to upsize to get the best YouTube compression. Details on this topic would be appreciated!

Tools? I've actually managed to get AVIsynth working on some old sanlyn scripts (he was the king of forum scripting advice!). Curious what else others use and why.

Someone recently suggested Vimeo instead of YouTube. Others?

And finally, I've never noticed if anyone on this forum shares their YouTube links. Is that allowed? I love when I find video on YouTube that was obviously created by an expert. It would be useful to see what others are posting to get an idea of what "good" looks like!

Thanks!

A lot of people use Selur’s Hybrid now. LS uses it for his YouTube uploads and anything that doesn’t involve more complex scripting or where he doesn’t have to change the script order. The install of Hybrid is just click download where as with Avisynth the install, getting all of the dependencies, and just getting everything to work correctly is a challenge. Hybrid is a GUI for scripting like Vapoursynth and Avisynth. Vapoursynth is faster and more optimized for multi threading than Avisynth. Hybrid is a really good encoder.

So is Vaporsynth QTGMC surpassing Avisynth QTGMC nowadays? It is my understanding that the former is better optimized and is the only one that runs on Apple Silicon-based MacOS.
Granted, I do have the Windows 7 computer I use for audio, but the amount of time where I have access to it in usable form seems to be decreasing.

Don't think LS uploads anything to YouTube, he has however uploaded some restored clips here and there on some forum posts, most recent example I can find is this.

I myself use Hybrid for all my VHS rip uploads, the workflow details can be found in the description of each video.

For upscaling, use 2K (1920x1440) to enforce the VP09 codec, that codec has better compression compared to the avc1 codec used for 1080p resolutions and lower. You must also upscale up to 1080p and above so that a 59.94 deinterlaced capture is shown at it's proper frame rate. That and 480p videos on YouTube suck anyway.

Never used it, but I've heard the Odyssey doesn't re encode videos that are uploaded (unlike YouTube), so the quality of what you upload there might be better.

In terms of YouTube channels that show VHS captures from those who know their stuff, the two that come to mind are Capturing Memories (he goes by latreche34 here) and lollo

For myself, not really.
For others/clients, yes, I do.

Ah, my mistake. Checked your account and didn't see any posted videos. Guess you do have a collection of unlisted videos then if it's meant for clients and the like.

Videocaptureguide did a video recently showing that to get the best YouTube quality, you have to add some grain as well. This more or less tricks the encoder into thinking there's motion in areas that are mostly static, so you don't get macroblocking as much. Worth a watch: https://www.youtube.com/watch?v=tfzw5AoFGrE

With the US holiday and a day off from work I hope to make some progress today!

More newbie questions... once upscaled what filetype do you upload to YouTube? I'm working with Lagarith lossless AVI's which are about 25GB/hour. I would guess once de-interlaced they will be closer to 50GB/hour. The longest video I imagine I'd upload would be 4 hours so that's 200GB. YouTube says maximum 256 GB or 12 hours, whichever is less so I'm within the limits. Is it best to upload those large files or do something like X.264?

I myself go by Google's recommended spec sheet for YouTube

I use Selur's Hybrid to encode them as h.264 mp4s, with a bitrate of around 25 Mbps for 1440p videos.

Just in time and just what I was needing. I think I'm ready to try Hybrid thanks to the Video Capture Guide video. Lots of familiar terms, let's see if I can produce something half decent.

That's always the correct answer, too. :congrats:

At the studios, we just exported/encoded to the specs that iTunes/etc required, and then they'd further transcode down to their own requirements (actually multi-transcode for multicast). That actually used to irritate me, because the specs would be sometimes be insane. We'd sometimes have to ship HDDs, because upload would just saturate the studio 'net connection for days (back in the 00s into 10s).

In the absence of specs, we'd use my in-house specs. The most important part there was 4:2:2, interlace handling, and high bitrate to withstand transcoding.

Youtube is really the only "wild west" of specs, letting you do whatever you want.

In my exact case, I'm working with SD formats, so something like 1080p is ridiculous (*usually) for the offline archiving and local/DLNA watching. Any Youtube work tends to be so a specific person can see it, and a special 1080p encode is rarely warranted vs. just uploading the 720p. (*For me, the main exception is mixed-source work, generally documentary. Because historically I do work with a lot of indy filmmakers.)

If bandwidth is not an issue such as fiber, just upload the lossless de-interlaced and upscaled version directly to YT, the hours you spend on encoding can be used for uploading to YT. If you are on a DSL or cable internet then it's worth encoding to save on bandwidth.

My YouTube channel is Live! https://www.youtube.com/@video-airchex

Live...? well I have two versions of my most prized Miami Vice video (1/10/1986) to see the effects of upscaling. One 720x480, the other upscaled to 1920x1440. Both Lagarith losslessly compressed AVI. The only thing done to the video before uploading was to inverse telecine. Then I borrowed a few lines from a lallo script just for fun (one with upscale, one without) and ran that through Avisynth+ (with Vdub2). For now I want to use Avisynth+ to better understand how various tools work before I fall in love with Hybrid (or not). This was done on a fresh install of Windows 10 (man, I had problems with W11). Capture was on my XP machine, Vdub 1.9.11, ATI-600 USB, AG-1980, TBC3000.

In "stats" the 720x480 shows that (avc1 as expected). The upscaled video shows 960x720 which is half what I upscaled it to (vp09 as expected). While processing the upscaled said HD. I have to admit, I like the looks of the 720x480 on YouTube better although the upscaled one looks like it maintains the 4:3 ratio. Any ideas why this is?

Side note: for anyone struggling to get Avisynth filters and required filters working. Install Hybrid, make a backup copy of your plugins64+ folder, and then copy the plugins64+ folder from the Avisynth folder in the Hybrid install folder. I had to remove a few that caused errors but within maybe 15 minutes I had Avisynth and a ton of filters ready to go. Thank you Selur!

That TV show music sounds like Motley Crue. :)

To truly thank selur, use the Paypal link at the bottom of his homepage: https://www.selur.de/
Toss him a few bucks, he deserves it!

The music is Ted Nugent, he is the bad guy in that episode.

Selur, absolutely!

Ah, the Nuge! Right era, wrong band.

It's because 720x480 has an aspect ratio of 3:2, and needs to be corrected to 4:3 or 16:9 depending on the source.

640x480, 960x720 (720p), 1440x1080 (1080p), 1920x1440 (2K) and 2880x2160 (4K) are all 4:3 aspect ratio resolutions.

I like to use this aspect ratio calculator site

In the early and mid 80's he was my hero. This is my oldest VHS tape (besides the phenomenal February 1985 Grammy Awards VHS my wonderful wife saved) and of course reminds me of my youth a bit so it's memorable to me.

Technically what amazes me is how much film grunge there is and that NBC would air it like this. I love the first 2 frames of the show open (at 2:03) where the sun looks purple. I'd guess that's just how the projector light looked passing through the film?

For fun I bought this one episode on Amazon Video just to see what the original looks like. OMG! If I was to ever attempt a restoration of this episode or even just the show open I'd have to work pretty hard. There is a lot of color that was lost on this VHS tape. But that's what makes this fun, right?

VHS capture is a never ending stream of what I thought I knew going right out the window...

I thought capture at 720x480 was the right choice for 4:3 NTSC broadcast video. Is that not correct and if so what is correct?

I know about upscale and spline. Is that the right way to correct to 4:3?

- capture 720x480
- deinterlace then resize to 720x540 (vertical stretch)

I've read sanlyn's amazing post a million times - https://www.digitalfaq.com/forum/vid...broadcast.html - And every time I find something new. Or something I should have known by now.

I had asked "Why did you encode the final mp4 at 23.976 fps? It would seem to me if you did the 3:2 conversion you would leave it there at 29.97 fps."

sanlyn's (edited) reply " If you want to mount the video on the internet, it can't be telecined. So you will need a progressive 23.976 version for the 'net. Also, the last time I heard about YouTube submissions, video intended for 4:3 display aspect ratio has to be submitted in a certain frame size, or YouTube will reformat it for you using the quickest and dirtiest means possible. They prefer submissions in 720p format. 720x480 can't be used on YouTube. The 4:3 video has to be resized to 960x720, then 160 black border pixels are added to each side to make a 4:3 video enclosed in a 1280x720 frame."

In Avisynth you can do it this way:
Code:
AviSource("path to 23.976 720x480 video")
Spline36Resize(960,720) <-I assume this could also be 720,540?
AddBorders(160,0,160,0)"

If the footage is from a film (or any material shot at 24 FPS), then yes it needs to be inverse telecienced back to 23.976. Not kept at 29.97 nor deinterlaced to 59.94 as that results in duplicated frames. That's how the film footage was able to play by the rules of analog tape, by repeating frames.

If the footage was shot at 29.97 FPS (or if the 24 FPS movie footage is included with footage shot at 29.97i, so something like a documentary or TV broadcast with commercials) then there is no choice but to deinterlace when all the footage has different frame rates.

For YouTube, 720p and above preserves the new 59.94 frame rate (if the footage is deinterlaced, not inverse telecinced. 480p does not preserve any frame rate higher than 30.

But YouTube compression sucks butt for Standard Definition so it's best to upscale to 1920x1440 at most, I think so anyway.

YouTube will automatically de-interlace and conform to the proper pixel aspect ratio as long as it is correctly flagged in the uploaded file's metadata. I have uploaded many 4:3 and 16:9 interlaced videos at 720x480 and YouTube automatically de-interlaces them and scales them to 640x480 or 854x480, respectively. Obviously this is far from the best quality, especially since it won't give you the full 59.94fps frame rate, but it does work.

In fact, it will even preserve CEA-608 closed captions and display them properly, even including different colors. For example, on this clip, which was recorded from VHS using a Sony DVD recorder, ripped from the disc, and directly uploaded to YouTube:

https://www.youtube.com/watch?v=dlLXmW2PJUU

The default in Hybrid for x.264 is 1.5 Mbps.

x264 --preset fast --pass 1 --bitrate 1500

Is that where I should be setting it to 25 Mbps (25000)? I've done some basic side by side comparisons and I'm not sure I see a difference in quality but do get a much larger file once converted.

Hi, bringing this one up again as I'm making (slow but steady after 9 years!) progress.

Here's what I generally do with my VHS captures (AG1980 or HR-S9800U, DataVideo 3000 TBC, ATI600)
- Capture with VirtualDub 1.9.11 on XP, 720x480, losslessly compressed and saved.
- Moved to Windows 11 (monitor calibrated with datacolor SpyderX Pro).
- Check histograms, crop and add borders (vdub, avisynth)
- Adjust color, darks, lights with ColorMill (in vdub) and/or ColorYUV, Tweak, Levels (vdub, avisynth)
- denoising, degrain, sharpen - lallo's script (vdub, avisynth).
- Inverse Telecine or QTGMC depending on what I have (vdub, avisynth).

- If it's going to YouTube, upscale to 1920x1440 - lallo's script (vdub, avisynth).

- Hybrid to convert avi to x.264.

So, back to my year old question. Should the bitrate, which defaults to 1500 kbits/s, be set to 25000 kbits/s?
Obviously that makes the file much larger. So I want to be sure and am curious what others do here.

And yes, I plan to slowly convert to using Hybrid for many of the above steps. Just taking it one step at a time!

I would put the de-interlacing first in the chain after capture, unless you're absolutely certain that all of the post-processing you're doing is able to handle interlaced video with no adverse effects.

I thought the message about deinterlacing has always been to do it last, just before converting to internet or not at all since most modern players handle deinterlacing just fine?

I any case, any comments on the bitrate (1.5, 25, other Mb/s) to use in Hybrid when converting VHS to x.264 that has been upscaled to 1920x1440 for YouTube?

Deinterlace must be first due to some filters only working with progressive video.
Anyway bitrate would be 25mbps for 1440p at 60fps

I will double check the tools I use in avisynth. Thanks to both of you for pointing this out.