IMAX 3D on YouTube

I just tagged this one yt3d: swap=true
Anyway, I’m switching to the "over/under" layout from now on because honestly, the "side by side" look just looks cheap and doesn't give me the quality I'm after.
Gear-wise, I’m rocking two cameras: a Sony HDR-TD10, which is my 3D rig shooting at 50i, and a 2D HDR 550V running at 30i.
If anyone actually cares to know the technical nitty-gritty on why I'm obsessed with the "over/under" format, just ask. I can go on about it.

Check out this promo video I put together for a buddy of mine over in Miami, Thailand. Guess we can finally find some decent burgers even when we're hanging out in Thailand, haha.

Moose, New Guesthouse in Miami, Thailand

http://www.youtube.com/watch?v=PCJMjW_cdcQ

Ronald Chavez7 said:If anyone is curious about how I landed on this "Over / Under" format, I would be more than happy to sit down and walk you through my reasoning in much greater detail. Does anyone want me to dive deeper into why I chose this specific approach?

Well, if you ask me... what’s the actual difference here, other than the resolution being cut in half vertically instead of horizontally?

I suppose I’m also quite curious about this, since common sense would lead one to believe that the image with a higher line count is inherently the superior one.

Look, here’s the deal. Full HD is 1920x1080 pixels. But if you want a true 3D image, you’re looking at 3840x1080 pixels.
Since YouTube won't support anything higher than that resolution, we’re stuck trying to "cram" two separate video channels into a standard Full HD frame.
When you go with the Side-by-Side (Swap L) format, you end up with a video that's basically 980x540 per channel. Now, why isn't it 980x1080?
It's because when you're watching 3D, the screen uses every other horizontal line for the left and right channels. It's how the tech works.
If I remember correctly, YouTube only supports passive 3D. So, when you shoot in an Over / Under format, what you actually get is
1920x540 per channel. And even though it's 3D and uses those alternating lines, you still end up with a format that looks exactly like what you'd see on a passive monitor.
In case anyone was wondering, that's essentially the 1080i format. 🙂That’s honestly one of the main reasons I’m so obsessed with my camera—it shoots in 3D 1080i.

Well, if we look at the side-by-side format—where each frame sits at a resolution of 960x1080—most people watching on YouTube tend to use the anaglyph method anyway, so I suppose that might actually result in a superior version. With the Over / Under layout, I feel like the aliasing on the horizontal lines becomes much more pronounced and distracting.
It seems we are circling back to my initial point; I guess the perceived image quality is determined more by the total number of lines rather than just the pixel count per line.

Sure, maybe there are more pixels and the image looks sharper, but honestly? After just a few minutes, I start getting a massive headache. Even fifteen minutes after I stop watching via the anaglyph method, my color perception is still totally wonky. And let’s be real—the colors aren't even right to begin with. Plus, it feels super unhealthy, like it could actually mess with your vision. I wouldn't recommend this to anyone unless they were absolutely backed into a corner. For me, I’m sticking with the Over / Under format. It doesn't even make sense for me to switch since my camera shoots in 1080i anyway. Right now, passive monitors are the way to go if you want to balance decent picture quality with not destroying your health. I'm just waiting for the tech to catch up. If someone could finally make passive monitors with 1920x2160 resolution, then we'd finally be talking. 🙂

I suppose you have a point regarding the anaglyph effect and the color accuracy issues, but unfortunately, that remains the only way most people can actually watch these YouTube videos, since 3D hardware isn't exactly widespread here in the States yet. I am not suggesting that Over / Under is a bad option in general; in fact, I guess it might even be a better choice for those viewing 3D video on passive glasses systems.

Ronald Chavez7 said:It works like this. Full HD is 1920x1080 pixels. To get a true 3D image, you'd actually need 3840x1080 pixels.
Since YouTube doesn't support that higher resolution, we're stuck trying to "squeeze" two video channels into that Full HD space.
When you use the Swap L format, you end up with a video that is 980x540 per channel. Why isn't it 980x1080?
Well, because when watching in 3D, every other horizontal line belongs to either the left or the right channel.
If I recall correctly, YouTube only supports passive 3D. So, when you record in an Over / Under format, what you get is
1920x540 per channel. And regardless of the fact that it's 3D and uses alternating lines for each channel, you still end up with the format used on a passive monitor.
For those who didn't know, that's essentially the 1080i format. 🙂 This is one of the reasons why I'm so fond of my camera that records in 3D 1080i.

Your camera doesn't capture 3D by simply putting the left channel in one half-frame and the right in another, because if it did, you would only be getting 25 (half) frames per second instead of the necessary 50... instead, the camera captures the left and right channels for every single half-frame using MVC format. A passive system isn't interleaved in terms of interlaced video; even though every alternating line on the screen shows a different eye, they are displayed SIMULTANEOUSLY, unlike an interlaced video signal where the half-frames are shown one after the other... which is also quite different from the "passive" systems you see in movie theaters, which are actually 1) active on the projector side and 2) alternate between showing the left and right eye.

Following that logic, I truly believe that side-by-side is the better way to go, especially if the recording is 1080i. By doing it the other way, you aren't just losing an entire half-frame when converting to progressive for YouTube, but you're also losing a quarter of the total image when you switch to Over / Under... which is exactly why 3D HD 1080i is broadcast via satellite in side-by-side.

Even for passive 3D screens that use alternating polarization, side-by-side seems superior to me... besides, isn't the human eye more sensitive to a drop in vertical resolution than it is to a drop in width?

I mean, I’m just following logic here. You never stop learning, right? 🙂
I get how this whole 3D MVC format thing works. It records the left channel and then just saves the difference for the right one, though in reality, it doesn't actually save that much space. Or maybe it does? I read that somewhere once, but honestly, my memory is shot.
I’ve been using Vegas Pro, and it basically handles all the heavy lifting for me. I just toggle the Over / Under option and set it to Blend fields for the video.
Why wouldn't it just convert all those half-frames into progressive? Because, frankly, the results I'm getting look awesome.
Once I finally get off work today, I'm gonna mess around with it some more and see what else happens.

Ronald Chavez7 said:I was just following my own logic, and isn't that how we learn throughout life? 🙂
I have a general understanding of how the 3D MVC format works—it captures the left channel and then records the difference for the right, though in practice, it doesn't actually save much storage space. (I read that somewhere once, though I can't recall exactly where.)
I usually work in Vegas Pro, and it handles all the settings automatically for me; I simply select the Over / Under option and use Blend fields for the video.
Wouldn't it be better if it just converted all those half-frames into progressive video? Honestly, I'm quite happy with the results I've been getting.
Once I get home from work later today, I think I'll try experimenting with a few more things.

"Blend fields" works okay if you're dealing with slow movement, but when things start moving quickly, you end up with this messy, ghostly artifacting... it's less about vertical resolution and more about temporal issues, yet visually, it looks "sharper" in terms of vertical resolution than just discarding half-frames entirely. Ideally, you'd want to create a 50p file without using "field blend," but I'm not sure if YouTube supports that, or if they would just drop the frame rate down to 25 fps if they did.🤷

Since MVC only requires a slightly higher bitrate (around 28 mbps compared to 24 mbps for a single view) for stereo recording, there really isn't any need to resort to recording stereo via half-frames and sacrificing your frame rate in the process...

Look, YouTube is absolute garbage when you actually care about image quality. That 1080p bit rate is barely hitting 6,000 kbps. It’s pathetic.
That’s why you have to jump through all these hoops, and even then, the platform just forces everything into 25 fps regardless of whether you shot it at 15, 24, 30, or 50 fps. It doesn't matter what you do.
For now, I'm just sticking to 1080p at 25 fps with a 7000 kbps bitrate and hoping YouTube doesn't butcher it by reprocessing it down to something even worse.
Sure, there are some 4K videos floating around, but they're encoded at roughly 20,000,000 kbps.
Oh, and one more thing. When I try to open an MVC file in Vegas, it tells me there are three different formats inside.
video 1: 25 fps, interlaced 1920x1080x12 AVC
video 2: 25 fps, interlaced 1920x1080x12 MVC
Audio: 5.1, 48000 hz, dolby ac-3
So that 25 fps is actually effectively 50, right? Or am I totally tripping here?

Edit: My bad, gotta correct myself—YouTube does support different frame rates, but only up to 30.

..think about it this way: if you're looking at 25 (30) interlaced video, that actually means the playback is running at 50 (60) half-frames per second, which implies you've got 50 (60) temporal samples happening every single second... however, if it's 25 (30) progressive, then you're truly just getting 25 (30) frames per second, meaning there are only 25 (30) temporal samples per second...

isn't it funny how YouTube can be a bit limited since it only accounts for up to 30 fps, which really isn't quite enough to capture that perfectly smooth motion we all want... but I suppose that's just a common oversight shared by pretty much all these streaming sites...

Look, give it to me straight. If I shot this stuff in 50i, am I actually being smart if I switch it over to 30p?

Ronald Chavez7 said:Be honest with me here. If the footage was shot at 50i, does it actually make sense to convert it over to 30p?

Not really... you’d want to go with 24p instead. Isn't it true that there's much less stuttering when a monitor or a graphics card displays 24p on a standard 60Hz refresh rate compared to when you try to force that conversion inside Vegas?

Check this out—one of the absolute best 3D clips on YouTube


http://www.youtube.com/watch?v=jJjJeYfAwQw

Check out this killer 3D video I found taken from the top of the tallest skyscraper in New York City.

Beautiful panoramic 360 degree View day / Night from Empire State Building, 3D video

http://www.youtube.com/watch?v=XgdQnuutkAU

Which kind of glasses do you all typically use for watching 3D content on YouTube?

I suppose you should just stick with the ones you already own. On YouTube 3D, you can simply select the viewing mode that actually matches your specific glasses.

Here’s my very first attempt at an interactive 3D film. Go ahead, pick your weapon and decide how you want to take me down.🙂
It would be awesome if you guys could share the video around, thanks. Sending love from sunny Miami!

http://www.youtube.com/watch?v=_uYfvLBJm00

It seems to me that I am experiencing issues with my left and right eyes once again.