as 960 x 720 would be the correct 4:3 resolution

Nope, not for NTSC "IMX" DV format, which the OP's AJA card capture appears to be.

486x(720x0.9091) does not equal 4:3 aspect. The true aspect of that format is 1.3468, not 1.333_ (the exact 4:3 ratio is true only for NTSC non-DV, 1.0 PAR source, from the 1990s). That's where the extra 12 pixels in width come from.

On the other hand, the true aspect for standard NTSC DV is 480x(720x0.9091), or 1.3637

There are 2 checkboxes in Vegas to fudge the output to avoid thin black lines in the project and render, but I'm a wannabe perfectionist ( who still makes mistakes!)

As far as social media, it's almost exclusively 16:9 ratio, but that would make his 4:3 source look really fat.

@mark-y Thanks! In the upscale plugin, do I simply put in the 1.48 factor, or are there any other settings I need to bother with?

That should do it, but be sure to put it at the Media Level, or at the very least before Event Pan/Crop.

I don't remember if upscale in 20 had the newer AI models that 21 has... Smooth will be lowest quality but will render the fastest, sharp will render very slowly, but will provide better quality (test this first with a short render, as your mileage will vary. I often find the AI upscale effect does absolutely nothing while still costing hours in render time).

@Pea-Hicks Let me know if I've made any mistakes, I very rarely work with IMX footage, which is as scarce as hens' teeth.

@mark-y Honestly, I'm not sure how I ended up with IMX? My VHS deck is going through a DPS 475 TBC via S-Video, then SDI out to a Blackmagic Decklink Studio card to BlackMagic Media Express, 10-bit YUV. Guess I need to look into whether I've set something wrong?

I don't remember if upscale in 20 had the newer AI models that 21 has... Smooth will be lowest quality but will render the fastest, sharp will render very slowly, but will provide better quality (test this first with a short render, as your mileage will vary. I often find the AI upscale effect does absolutely nothing while still costing hours in render time).

I don't use the Sharp models, I place a very light Unsharp Mask before the Upscale plugin in the Media fx chain, it's not only faster, it looks more natural to me.

It's happening in your capture card, and it's not "very" wrong. My only real concern is the accuracy of the deinterlacing.

Another unimportant observation; unlike movie film to digital transfers, capturing 10 bit 422 is overkill, but will do no harm.

Your native NTSC video tape is limited to 4:1:1 (at best), 8 bits per channel, 480 lines, 59.94 fields per second. Capturing at 422 10 bit, your net usable bits are still 8 bits in a bigger bucket, 4:1:0 chroma subsampling (by LCF), 29.97 frames per second.

An added "trick" for Youtube that I didn't mention for the sake of clarity, is "bobbing" the separated fields to smoother 59.94 fps, the one disadvantage being increased shadow noise, which can further impair underexposed video tape footage.

@mark-y Yeah, I know my capture settings are overkill, but I guess I'm going with the "oversampling" idea- capture more than is actually there on the tape, do whatever processing/editing I need to do, then encode with a more sensible setting. In most cases, I delete the original capture file anyway, so it's not a big deal for me to have a huge file temporarily taking up hard drive space.

@mark-y Another question... can you explain the benefit of adding the upscale plugin, vs not adding it and rendering at your recommended settings anyway? it dramatically increases the render time, even using the "smooth" setting, and I'm not sure I see much quality difference (at least not with this particular project).

@Pea-Hicks this is turning into a good discussion, thanks for keeping an open mind.

With DV and VHS source, which tends to be sketchy, You may not notice much of a difference. You may not wish to bother with the extra encoding time.

In tests with commercial 480i DVD source using the Smooth Model and ultra-light Unsharp Mask in Vegas, the results at 720p when compared with conventional Bicubic upscaling, can be dramatic:

@mark-y Yeah, that's a huge difference. With my EP mode VHS source, I'm not seeing anything like that kind of difference.

Time-shifting the 1996 Olympics on a tape that my wife used to record her soap opera every day produced a lot crap on tapes that I don't have the will to overcome.

It is refreshing that one could buy a car for less than $40,000 though.

Edit: ... and to see some people who have likely retired by now.

Vegas does quite a nice job of upsampling 29.97 Interlaced to 59.94 Progressive, aka "bobbing," but even better. https://youtube.com/shorts/MUqhxNTX53k

@mark-y Yeah, that's a huge difference. With my EP mode VHS source, I'm not seeing anything like that kind of difference.

For a free upscale option (trial limited to 5mins per video) there is VideoProc AI

It's by no means perfect and from my limited use meant for medium to high quality footage, your VHS may fall on the low quality size.

Example, original on left, 2x upscale on right

This is giving it something it likes, but still I see problems with too much denoising leading to less detail and it's changing the color of the sky. It should improve as they add more models making it compatible with more source material. They need to give you control over the denoising though, look at the guys jacket at the end, the upres has less detail.

After deinterlacing, I would upscale it to 720 lines so that YouTube treats it as 'HD'. It can still be 4:3 aspect ratio (or near enough - I've done numerous PAL-DV-derived web uploads at 988x720). No need for pillarboxing, at least on YouTube. Other platforms may prefer it.

Last time I did this stuff, I was still deinterlacing with QTGMC in an Avisynth script. Other methods may have caught up. Keep the "double" framerate after deinterlacing if you want a smooth look (i.e. not cinematic 24p).

And I was then upscaling with Lanczos 4. Other methods have definitely caught up or surpassed that, with Topaz in the lead, but not free.

I can expand on these methods tomorrow if you want.

@NickHope This is something I'm confused about, because when I set the project to progressive double frame rate and de-interlace, all i get is literally each frame doubled. i don't see any difference between the two doubled frames. so i'm confused as to how this results in a smoother look, since no actual visual information is being added?

@Pea-Hicks

For really nice 59.94 Progressive frame rate from your IMX DV source, this is how to set your project properties.

Render with a 60p template altered to match your project resolution of 972x720, which matches your source aspect and vertical scan lines perfectly for deinterlacting, and is "essentially" 4:3 HD.

I "do" recommend using the Upscale plugin at the Media level, with the aforementioned magnification ratio of 1.48x.

Good luck!

@mark-y Thanks, though actually i always do have the de-interlace and resample mode set that way, and I still don't notice any benefit to rendering at 60p. Maybe I'm just dumb, but isn't having each frame doubled (with each doubled frame being indistinguishable from the other), the same as having a single copy of each frame running at 30p?

Then something is not correct in your project or render settings.

The method I described uses optical flow interpolation to synthesize an "in-between" frame; there is no frame doubling going on, each progressive frame is completely unique unto itself.

Perhaps the example I posted wasn't formatted clearly: here is the 60p demonstration again in a better format on Youtube.

I have also uploaded the rendered 30p and Bobbed 60p clips to Drive for you to examine at your leisure

https://drive.google.com/file/d/18xmaCI9qWWrJIfzJzDnTGyXcY6MHBfs4/view?usp=sharing

https://drive.google.com/file/d/18us32Ztw0ww46YxBJo-pBLbp7zHWvhHK/view?usp=sharing

I'm involved in a heavy book restoration project right now, so I'll check in next week to see how you're doing. On the chance we're missing something regarding your video source, I suggest uploading an original sample to Drive or Dropbox (not Youtube or the forum!) so that I can examine it further if needed. Best of success!

@mark-y Thanks, though actually i always do have the de-interlace and resample mode set that way, and I still don't notice any benefit to rendering at 60p. Maybe I'm just dumb, but isn't having each frame doubled (with each doubled frame being indistinguishable from the other), the same as having a single copy of each frame running at 30p?

Allow me to give you a test file, https://we.tl/t-lLYXh4Hgda

There are 2 dots next to the frame counter, you want to see the top dot appear in every 2nd frame in your final encode, if it's not there you are losing 50% of your field data. This is how it should look after you render but for this site I've had to slow down the playback to 30fps(no 60fps) so it is 2x slow motion.

Also had another try of the free upscaler..... well it's looking like more of a disaster than previously thought, right side is upscaled. This is the most performant upscaler that exists currently, it will use all your GPU, they just got to fix these very obvious problems.