Cineversity

Out of curiosity, is there a reason why the output settings are set to 16:9 ratio and not 2:1 equirectangular?

I have worked with spherical photography for 10 years and 360 is always shot at 2:1. I have been using Kolor products for creating 360 video and their settings are also 2:1 ratio.

Facebook also recommends outputting in a 2:1 ratio. If you look at their faq
https://www.facebook.com/facebookmedia/get-started/360-video

I see that YouTube is recommending 16:9 as this does not create black bars on their viewer
https://support.google.com/youtube/answer/6178631?hl=en-GB
This does not make much sense though, as the final image will be viewed in a spherical environment and not flat.

As I say, I am just curious as to why 16:9 was chosen over the standard 2:1 equirectangular format.

ps. I Love this plugin! Been looking for a way for ages to create 360 from C4D. Previously, I did 360 renders from 3ds max which was a doddle to do but C4D could never do these easily before now.

Regards,
Chani

Hi Chani,

My best guess is, (and my personal idea bout) that TV standards (UHD) and its processing, as well storage option will be a reason.

Since over ten years we see that computer and TV-sets merges more and more. This huge shift will take time, based on the hardware installed in hundred of million households. Typically not the best solution wins (Beta vs VHS for example), it might win what reaches into the most households, or what is the most cost effective. Sadly, not often is what technically possible as well the final decision for the end user. Production wise, that is a different story, there we have to keep it hi-end. Perhaps, this time we can see a change, since smart-phones and tablets seemed to have a typical one year cycle/lifetime these days, and represent a mix of TV and computer. ;o)

I shoot since decades panoramas, and in case of 360ºx180º my typical storage is the equirectangular projection. Is it the best option (?), not in general, but for mainly horizontal oriented users, it might be. The Zenith and Nadir areas are heavily distorted, as even a quick look at it unveils easily.

If we talk about “stereo” or 3D-VR, then the format would be, based on eqiurectangular, better 4:1 or even 1:1. Or oscillate the channels…?

Going back to my research about VR in the early ‘90s, the main vision was, besides HMD [head mounted displays] “The Cube”, a six sided box with large 1:1 screens. The “Cross” available as conversion in C4D, would be the better solution, given less distortion, and a better workability during production, fixing patches, etc. This format is 4:3 (or 3:4) and would have been better suited for SD back in the days. Things changes. Perhaps we will see a six panel view packed closer than the cross for this, which would be 3:2,  close to 16:9—again.

For now, my best idea is, to produce always above the needs, and sample down. I do that with my RED in 6K (funnily enough a 2:1 sensor, well: 6144 (h) x 3160 (v)) or with my photography, for 3D in 40K+ 32bit/channel/float or for my art in 16K+ at least 16bit/channel/integer equirectangulars. Not in sRGB or REC 709, since we move to UHD and REC 2020, with HDR options.

Bottom line for me, I would go with YouTube and 16:9, but if the target market is FB, well, then we know already where it goes, given the bad image quality they allow now. Besides that, Google likes to produce a representation of the world, FB does not so far [except for twenty cities, but many years behind Google]. It is more likely that Google owning YouTube will push this very hard. Yes, FB declared this to be a tool, but between want to have it as tool or that the mainstream use it is a long way. In all fairness, google “glasses” didn’t make it.
My prediction is, as long as the GUI is not changed from desktop to “office space”, VR will grow slow, except for games.

As a vivid “Making Of” fan and collector of such, I can already see that this will be a feature on Blu-ray™ discs especially with the upcoming UHD discs: Behind the scenes or behind the stage documentaries. Which would certainly work nice if available from the start in UHD, which is: 16:9. But well, things have often played out differently, based on strong none technical influences.

Again, this is my personal view, I guess the developers of the plug-in have other sources to share or opinions.

Hi Chani,

Thanks for the feedback!

I had a discussion just yesterday with some folks about this - they had the same question. When I created that first series, I was pretty focused on YouTube VR - halfway through the series I discovered Facebook 360 and decided to cram it in there as well. Of course a month in VR is like 10 years in most fields it seems, and I’ve since been working with other players and delivery mechanisms. I’m definitely going to have to update the original VR tutorial as things continue to change.

I think the ideal resolution is probably 2:1 (4096x2048) for mono, and 2:2 (4096x4096) for stereo. But we always have to shift the ideal resolution based on the delivery medium.

The YouTube help says to output at 3840x2160. I think that’s probably because that’s the biggest res YouTube can currently take, and YouTube is completely geared towards 16:9 aspect footage. Outputting 3840x2160 for YouTube allows you to use every pixel both in the native resolution and in the 16:9 aspect resolutions they automatically generate as quality options.

When outputting video for Cardboard, iOS or GearVR you have to look at the limitations of the various mobile devices. Most mobile devices can’t decode a 4096 wide MP4, and even then won’t be able to decode 4096 tall as well. So we have to make compromises there too. There’s a great post from Purple Pill where they discuss their observations and encoding decisions.

So ultimately, I think it comes down to delivering a resolution based on what the client or delivery medium expects. Fortunately YouTube is forgiving if you upload a 2:1 and GearVR will happily play a UHD Over Under frame. And CV-VRCam will scale to whatever res you need to output - the default render settings just provide a starting point for YouTube VR.

Let me know if you come across any other info about this topic, if there’s any other features you’d like to see in CV-VRCam, or if there’s other VR topics you’d like us to cover in Cineversity tutorials.

Excellent, thanks for the concise replies!

I use Panotour (from Kolor) to build out 360 tours that also have interactive 360 videos (both live action and 3D renders). When I import a CV-VRCam output into Panotour, it does not recognise it as a true 360x180 panorama. This is obviously due to the aspect ratio being 1.778x1 instead of 2x1. Panotour thinks these are 320x180 but I can override that setting and tell it to be 360x180.

I appreciate that there are not hard and fast rules (yet) for 360 video as these are still relatively new and no real standards have been defined yet. I am just looking at it from a traditional spherical photo point of view.

Dr Sassi - you wrote “I shoot since decades panoramas, and in case of 360ºx180º my typical storage is the equirectangular projection. Is it the best option (?), not in general, but for mainly horizontal oriented users, it might be. The Zenith and Nadir areas are heavily distorted, as even a quick look at it unveils easily.”
What do you mean by ‘heavily distorted’? Are you talking about when you view these as flat, equirectangular images?

Great work though Rick in creating an excellent plugin/method for C4D users to generate full 360 output with ease. The best part? You can output true stereoscopic results. Fantastic!

Cool! I haven’t had a chance to play with Panotour yet but this sounds like a great workflow. Without looking into it specifically, you probably should just adjust the C4D render settings to 4096x2048 for use with Panotour.

Dr. Sassi mentions the distortion that’s inherent in equirectangular projection - because the poles of the sphere pinch at top and bottom, and spread out over a very large area of the equirectangular image that polar distortion is inevitable with a spherical projection. The next version of CV-VRCam will include Pole Merging, which will minimize the stereo effect at the poles to help minimize the distortion.

A lot of VR video is migrating to cubic mapping, because it doesn’t suffer from the distortion. The major downside to cubic mapping is that it doesn’t fit well in a 16:9 or even 2:1 frame, so technologies that are designed around traditional frame sizes (YouTube, mobile phones) have difficulties supporting the format. The next version of CV-VRCam will also offer cubic mapping as an option.

Hi Chani,

I think Rick cleared all the points. Thanks, Rick. :o)

Except for the Autopano Tour part. Would you mind to open a new thread in the “QA Forum”(as this is then no longer a plug in discussion, and please share an example file (I can provide an private upload link). I’m happy to investigate it. I guess you should have no problems if you use first Autopano Video Pro and export from there.

In the late ‘90s I used Apple’s QT-VR-Studio for that (images back then) and it is pretty much more an authoring program to connect things (simplified said).

Yes, this company is great in many areas, but there is too much “auto” in their products, with no option (directly) to overcome problems based on it. For my pano work I use, besides others, their products, and often one needs to explore it carefully, to find the way it fits into ones pipeline. I have no idea what kind of problem you hit, but we are all very interested in finding out how to connect things “headache free” :o)

Speaking of which, have you tried to change the suffix to “.mp4”, if you have that file as “.mov”? So far I can’t see any other problem, both [2:1 and 1.778/16:9] work here just fine. Check perhaps the file properties>projection.

Let me know if you like to dig deeper.

All the best

___________________
P.S.: some of my photographic work is here, so you can see that I use this stuff heavily:   https://plus.google.com/+DrSassiLA/posts