6K resolution is only possible with fish-eye. Beware of fake 8K
New 6K videos are coming to SLR in the highest resolution possible thanks to SLR projection meshes based on the camera lens profiles for the most popular cameras. Now, videos are released in the original ZCam K2 Pro sensor resolution, producing a 5800x2770px 200º FOV stereo image that is marketed as 6K. For equirectangular projection the original sensor image is cut down to 5400x2700px in 180º FOV.
The red circle in the image shows how much extra visual footage is gained with fish-eye compared to the traditional equirectangular format. This picture can also be watched in VR, simply make sure to keep ‘_MKX200’ in the filename. That 10º extra from each view adds an incredible amount of immersion as it prevents the viewer from going outside of the visible area due to the natural head movements we do all the time. Another advantage of fish-eye is that fewer renderings are required in the workflow process, meaning the image quality is closer to the original raw files.
You can see this by comparing Lauren Phillips in the “Hard Sell” movie, from fish-eye footage to equirectangular. Check SLR for all the fish-eye videos. Every producer can also get their videos in 6K fish-eye by following this Mistika Boutique tutorial.
It is worth noting that some producers create ‘6000x3000’ (“6K”), ‘7200x3600’ (“7K”) or even ‘8000x4000’ (“8K”) equirectangular videos from the same 5400x2700 actual sensor image. These are merely upscaled images boosted with junk pixels that have no value for the viewer. Instead “6K”/“7K” and "8K" is barely useful as a selling point. Things get even worse, as the 5K version of the video, that would otherwise be the proper resolution, is of lower quality because it’s downscaled from the upscaled 8K files. See for yourself: upscaled 8K video vs original quality 5K footage.
It's thought that fake 7K/8K emerged with the release of the ZCam K2 Pro camera, which offered enhanced image quality. Having the same sensor resolution as its predecessor the K1 Pro, the K2 Pro is capable of much better image quality due to better sensor specs. Also, the K2 Pro offers 10-bit color vs the K1 Pro’s 8-bit color, has better performance in low light, better HDR, native h265 support and more. Some producers were tempted to overstate their camera upgrade with false advertising.
It will be even more notable now that actual 8K cameras are coming to the market, showing significantly improved quality compared to upscales.
On the SLR website we tag 5400px equirectangular videos as “5K”. Videos submitted by producers as “6K”/”7K” and “8K” are tagged accordingly. We’re unsure of the best way to deal with this issue. Ideally, we would like every VR video producer to render their videos at the original sensor resolution and not upscale.
Commentators have mentioned the useful impact of AI machine learning (ML) upscaling suggesting many examples on Youtube. However, none of the “6K”/“7K”/”8K” producers suggested the use of ML. Video upscaling is a basic feature of any video editing software and does not enhance image quality. You end up requiring more bandwidth, longer download times and your headset or desktop computer heating up for no reason.
While ML requires huge processing power and time to effectively upscale flat videos, it still can’t be used for high resolution stereoscopic videos because of convergence plane requirements. Simply put, ML would have to also respect the stereoscopic disparity between the left and right eye and the tech is simply not there yet. And once this tech does emerge it would be right and proper to tag it as “ML or AI upscaled”.
Please let us know your thoughts on this issue.
Update. Around the time of this blog post, CzechVR released an 8K video, believed to be natively recorded in 30FPS and later interpolated to 60FPS. It's the only known exception to this post.