This was inspired by the other thread where the poster asked about a lower quality 1080p or a higher quality 720, so I decided to do a test. Attached you will find 2 files, one of the files is x264 with the "slower" preset and tune film at 1920x858 and 4000kb/s, the other file is x265 with the "superfast" preset at 1280x572 with the bit rate a bit lower at 3800kb/s, mainly because x265 undershot the target bit rate by 200kb/s.
Please view them at full screen, there's no sound, the source was 3840x1714 73mb/s Tears of Steel and when deciding which one your prefer do it from the point of view of a consumer, in other words assume you had a membership to a paid content site, like Hulu or NetFlix and for a fee of $20 a month you were allowed to download 20gb of content each month, DRM free, which of the two would you feel gave you the better "bang for the buck".
I also did SSIM and PSNR tests on each against the source and after a sufficient number of people have shared their preference I will post what the tests say is the better quality file.
Thanks for participating.
+ Reply to Thread
Results 1 to 19 of 19
I don't even have to look at them. I prefer the x264 encoding because all my devices play h.264. Only the computers play h.265.
It's also not an apples to apples comparison. Worthless.
If a choice between 720p x265 or 1080p x264 at the same bitrate ever comes up you can be sure something has gone terribly wrong somewhere down the line.
Complete waste of time. The bitrates are similar but you used a different encoder so there's nothing to say the 720p encode must be lower quality. Plus I can't be bothered downloading 700MB worth of files when I already know if I had to choose between a lower resolution and more compression artefacts I'd go for a lower resolution every time.
I used a noisy video and an unnaturally low bitrate so the difference would be more noticeable, and none are especially pretty, but I wonder if anyone will claim the 1080p video looks better because it's a higher resolution? I ask because the "other thread" sophisticles mentioned inspired me to do so. Obviously the 1080p version has much more picture detail.
Why two 720p versions? The first used the same encoder settings as the the 1080p version but that's not completely fair because there's less video to encode and it therefore doesn't take as long. So I figured for the 720p version a second encode using a slower x264 speed preset would even that out a little.
And for fun I threw in one at 540p. Mainly because it helps show resolution isn't the only thing determining picture detail. Stop the 540p and 1080p versions on the last frame, run them fullscreen or maximised and switch between them.
Edit: I forgot to convert the colours for the 540p version, in case anyone wonders why it looks a tad different (green might look a bit brighter, depending on the player being used).
Last edited by hello_hello; 11th Feb 2015 at 06:01.
you designed this to be an academic discussion, pretty much no practical reasons or very little are present to investigate, for myself I would not encode for a computer only for now,..., and in 5 years live transcoding could be possible (not for UHD yet but anyway), so you guys want to be "progressive" but at the same time it is lost in trans... encoding
Last edited by _Al_; 11th Feb 2015 at 09:31.
Encoding speed is all I can think of...
How long did each take to make?
So many people miss the point of compression.
Lossy compression is unfortunately a necessary evil but obviously should be minimized as much as possible.
For many it is about compressing the maximum amount while maintaining a minimum video quality instead of compressing the minimum amount while still being able to play the video.
The mindset of some is incomprehensible.
Perhaps they should put a note on each video they compress:
"While the video quality is mediocre I take great pride in getting a marvelous compression rate!"
I use QTGMC in lossless placebo mode for pretty much everything interlaced, my x264 settings are slightly higher than "Very Slow" with CRF set at 16, the FLAC encoding settings I use with MakeMKV means it takes about 4 hours to rip each Blu Ray, I've spent countless hours trying to get chapters into my MKVs in EXACTLY the right spots and all my Blu Ray rips are stored on my RAID6 drive in their original codecs...
I'd wonder about the kind of people who'd find the subject of this thread useful.
I'd be willing to bet for many of us it's about compressing the maximum amount while maintaining maximum quality. Or at least very high quality. Where do we fit into you theory?
Perhaps I should put a note on each video I compress:
"While the video quality is very high I take great pride in knowing newpball will decide it looks bad because it doesn't fit some preconceived notion he has regarding bitrate and resolution and if I post a sample to prove otherwise he'll just ignore it."
Last edited by hello_hello; 12th Feb 2015 at 03:48. Reason: spelling
Lossy compression should always be done because there are no alternatives, and the amount should obviously always be minimized to as little as possible. Why anyone would disagree with this simple thing is beyond my comprehension.
If you burn a standard DVD you have to use a certain amount of lossy MPGEG-2 compression as the disk is limited in space, but obviously not more than necessary. If you make a DVD from a high quality source and you have 30% free space left over you really failed. If you have a very high resolution video and your hardware cannot keep up with reading it you have to compress, hopefully lossless but if that does not work you have to do lossy compression. If you have to stream, that is if there are no better alternatives, you would have to compress as well.
Lossy compression is necessary to overcome some constraint, it could be a maximum file size, the speed in which a storage medium can deliver, streaming bandwidth. It is done out of necessity not done out of some sort of a coolness.
If you do not have such constraints but some harebrained idea that 'well 1GB is just too large' you just reduced the quality of your video unnecessarily.
Ideally a video would be compressed lossless, the same as with images. But we are not there (yet) there are too many constraints. But what we should do, or at least those who are quality-conscious, is to minimize lossy compression.
Last edited by newpball; 11th Feb 2015 at 14:10.
Quality can have different dimensions.
Where have you been ten - fifteen years back, were you even born? You did not need to give somebody 8Mbit, 50min and 30% was empty? Trust me, they would not care even if it was 4Mbits. All they wanted to see was a message. So doing it even now, I'm failing miserably perhaps.
Focus on reality, or quality of message!, not delivery box, Youtube is prove how things work, I'm happy for a 360p tutorial how to change a stupid recoil spring in lawn mover, in 30 year model (made to last or designed to be fixed) not needing to buy new lawn mower. Perhaps you have not done so far too many deeds to realize things like that. Man that guy reading you perhaps is feeling guilty and stupid now.
And movies .... , I said it before, there is about 100 people encoding the same movie every day at the same moment, it is entirely up to them what they get out of it, whatever reason ... some decide to encode 16 ref frames, some 1080p with 2GB, it is up to them ... btw, those guys should downscale but reading you and similar posts they think they have to keep quality because they keep 1080p, HD wouldn't you say .... it is always, always better to downsize if there is not enough bitrate ...
I could offer examples where I've encoded video and the encoded version has looked better than the original due to filtering, even when I haven't used a particularly low CRF value, and doing it at a lower resolution, but I don't want to be the cause of your ears filling up with sand again. QTGMC for de-interlacing would be typical of that. I'm not just referring to improving quality with filtering when re-encoding DVD video though, but when re-encoding Bluray video as well. It's so obvious to me your arguments are irrelevant to my personal encoding I'm not sure I can find a way to explain it. It's like having to explain to someone that grass isn't pink.
You do realise that as a general rule in forums, the strength of a persons argument is indirectly proportional to the likelihood they'll try to emphasise it with a silly picture?
Last edited by hello_hello; 12th Feb 2015 at 09:39. Reason: spelling
Maybe at the loss of fine detail stage now, but overall probably still better than the original.