Page 1 of 1

Quality Test

Posted: Sun Mar 15, 2015 10:36 am
by Necromancer
I decided to compare youtube rendering quality.
I rendered the same clip once in 1080p and once upscaled from 1080p to 4k.
uploaded both versions to youtube and bestplays.
Results
phpBB [video]

gif

Youtube 1080p vs Youtube 4K
Spoiler: show
phpBB [video]

gif
Youtube 1080p vs bestplays 1080p
Spoiler: show
phpBB [video]

gif
Youtube 4k vs bestplays 4K
Spoiler: show
phpBB [video]

gif
bestplays 1080p vs bestplays 4K
Spoiler: show
phpBB [video]

gif
youtube 1080p vs source
Spoiler: show
phpBB [video]

gif
youtube 4K vs source
Spoiler: show
phpBB [video]

gif
bestplay 1080p vs source
Spoiler: show
phpBB [video]

gif
bestplay 4K vs source
Spoiler: show
phpBB [video]

gif
source vs origin
Spoiler: show
source - the recording after rendering (H.264 18MB/s)
origin - the original recording (lagarith lossless codec)
phpBB [video]

gif

source videos:
Youtube 1080p
youtube 4k
bestplays 1080p
bestplays 4k

Re: Quality Test

Posted: Sun Mar 15, 2015 3:47 pm
by Gwynzer
There's a lot of videos there, and I'm crap at noticing small changes in graphics stuff (Except hair tessellation, that's proper next gen crap.)

What's your conclusion wit the comparisons? What have you found the best method for quality assuring?

Re: Quality Test

Posted: Sun Mar 15, 2015 5:04 pm
by Necromancer
Is no video, its a gif comparison between the same video frame from the different host options. Espcially made this way to notice the differences.

My conclusion? Video hosting quality sucks in 2015.
Youtube 1080p is the worst.
Upscaling to 4k makes text a little bit sharper.
Bestplays quality is a bit better then youtube, but upscaling to 4k doesn't help at all.
So uploading to bestplays at 1080p it is.

If you have a 4k monitor and can record in 4k, that will make significant difference for people watching it in 1080p.

Re: Quality Test

Posted: Sun Mar 15, 2015 5:12 pm
by Jokerle
maybe worth to look at in this context
https://www.youtube.com/watch?v=ak3LWX- ... =1&list=WL

Re: Quality Test

Posted: Sun Mar 15, 2015 5:20 pm
by Necromancer
That video is why I did it. Jack doesn't put the frames one vs the other. In that video I can't see any difference between the options, thus i was suspecting placebo.
Plus he only compares youtube and not bestplays.
If you know any other video hosting services that may render at better quality please post.
Spoiler: show
Porn sites are not an option
.

Re: Quality Test

Posted: Sun Mar 15, 2015 6:46 pm
by RazY70
Necromancer wrote:Is no video, its a gif comparison between the same video frame from the different host options. Espcially made this way to notice the differences.
GIF is probably not the best option to compare quality. It's only 256 colors so probably a lot of dithering going on. PNG would be a better option for that (lossless 24-bit).

Re: Quality Test

Posted: Sun Mar 15, 2015 9:22 pm
by Necromancer
ok, i encoded it into .webm at highest settings and uploaded to gfycat, however the site re-encodes it at unknown settings.
either way, to me it seems the youtube/bestplay quality is so bad that the .gif quality hardly makes any difference.

Re: Quality Test

Posted: Mon Mar 16, 2015 9:28 pm
by nx-jojo
websites like youtube have to watch out for traffic. this is why they drop your bitrate of a 1080p60Hz vid to ~5MBit. higher bitrate would make a significant difference but you can't change that. tbh i don't know what VIMEO bitrates can be used on VIMEO. as far as i remember those videos always looked great so if you really want to look for high quility options check it out.

since bestplay / youtube aren't much of a difference i would recommend using youtube. one example: it might make it easier for someone who creates a playlist of battledays. there are other reasons but they aren't really significan't.

imo if you want the best product - use it. otherwise go for the product most people around you use.

Re: Quality Test

Posted: Tue Mar 17, 2015 4:33 am
by Necromancer
I did check vimeo and it did look great, but apperantly its only for original "artistic" content. Uploading game videos are likely to be deleted, and the free account does not allow 3GB uploads of 20 min rounds.

Re: Quality Test

Posted: Tue Mar 17, 2015 9:35 am
by Tea-Assault
I'm getting the best quality for my recordings by exporting at 1080p 60FPS, but I also render at maximum depth and set the bitrate to 50mbps..

I did a quality test a while back with these two samples:- one is standard 1080p, the other is 1080p 60FPS, and this is the difference

phpBB [video]

phpBB [video]

Re: Quality Test

Posted: Thu Mar 19, 2015 3:41 am
by knorren
That 60fps 1080p looks fantastic imo, 2nd video Tea posted!

Re: Quality Test

Posted: Thu Mar 19, 2015 5:38 am
by Necromancer
youtube codecs work great when the video changes slowly (when theres small changes between consecutive frames).
Tea's video the camera stands almost still, so there is very little change, its ideal conditions.

But in a BD you constantly move around, turning left and right very fast, thats when YT fails. In my comparison i was in the middle of high speed bike riding.

If you check BD footage 1080p@60fps its not nearly as good as Tea's video.
in cases the guy camps a doorway for >3 seconds you see the video becomes significantly sharper, but its not common.
I looked at Jojo and Funhit videos who record at 60FPS.