X264 video: Impact of resolution on quality

[sanosuke]

Hi everyone
I think we can all agree that the best way to reduce a video's size without damaging the quality is by lowering its resolution (and using x264 of course). But by lowering resolution, doesn't it blur or remove visual details like sweat on a man's face or the stripes on a shirt e.t.c. Well most people won't notice the change but people like us who keep experimenting or tweaking x264 settings for better video will surely notice the loss of details. I learned through my previous threads that we should not encode with the original full resolution or it will lead to patchy artifacts : https://forum.videohelp.com/threads/324470-Patchy-artifacts-in-re-encoded-video

But is there anything we can do to save the details? since the lower the resolution of the file, the more it will have to be enlarged on playback, and the greater the visibility of the damage you are doing.

[poisondeathray]

The larger the resolution, everything else the same, requires a higher bitrate for the same level of "quality"

Different types of content require different bitrates. You can use crf mode encoding if filesize doesn't matter or you don't need an exact filesize

You can preprocess with degraining, denoise filters to filter out noise to reduce bitrate requirements. Prefiltering is often the most important , even more than encoding settings. If your goal was ultra low bitrates, like those in the other thread, it is essentially mandatory . Denoising reduces wanted details along with unwanted grain , so you have to be very selective. At such low bitrates , you can't conserve any details anyway

[sanosuke]

Thanks poisondeathray. I knew it would be you.
One more thing. I heard that raising the psy-trellis strength will lead to artifacts like the one i experienced.
They say its best to keep psy-rd=1,trellis=0 (Although the actual trellis value should be 2). So what's the thing with trellis?

[poisondeathray]

psy-trellis and psy-rd both act to enhance detail, especially around edges. That requires more bitrate. If you starve the encode of bitrate, you can get similar artifacts.

You're grasping at straws here. The most critical factor is bitrate, then fitering, then encoding settings.

At the bitrates you intend on using, I would use 0:0 , and do heavy denoising to eliminate noise and detail. (it will look like plastic dolls and cartoony) , but that's the only way it will work. That bitrate is way way way too low.

To answer your question: bitrate saves details

Even the most tweaked compression settings might give you a few 3-4% reduction in filesize for similar quality but take 20x longer to encode. It's simpler just to use more bitrate.

If it hasn't hit home yet, use a higher bitrate

[sanosuke]

Ok i get the point. Thanks.

[deadrats]

Hi everyone
I think we can all agree that the best way to reduce a video's size without damaging the quality is by lowering its resolution (and using x264 of course). But by lowering resolution, doesn't it blur or remove visual details like sweat on a man's face or the stripes on a shirt e.t.c. Well most people won't notice the change but people like us who keep experimenting or tweaking x264 settings for better video will surely notice the loss of details. I learned through my previous threads that we should not encode with the original full resolution or it will lead to patchy artifacts : https://forum.videohelp.com/threads/324470-Patchy-artifacts-in-re-encoded-video

But is there anything we can do to save the details? since the lower the resolution of the file, the more it will have to be enlarged on playback, and the greater the visibility of the damage you are doing.

you can not reduce the file size of a video without hurting the quality, no matter what you do. by lowering the resolution, you have damaged the quality, as you now have fewer pixels to represent the scene, i.e. 1920x1080 has 2073600 pixels per frame, 1280x720 has 921600 pixels per frame, there's no way that a video with less than half the pixels per frame is going to have the same overall quality as the original.

if you use a superior compression method (i.e. going from mpeg-2 to avc) or the same compression method and reduce the resolution then by calculating the original video's per pixel bit rate you can produce a smaller file sized video that has the same per pixel quality.

[jagabo]

you can not reduce the file size of a video without hurting the quality, no matter what you do. by lowering the resolution, you have damaged the quality

I agree. Resolution is just as much a part of the quality as anything else. Otherwise we'd all just reduce the resolution to 1x1 pixels and save uncompressed.

[filmboss80]

I think we can all agree that the best way to reduce a video's size without damaging the quality is by lowering its resolution (and using x264 of course).

By this point, you should be realizing that NONE of us agree with that statement. I was just wondering where such an assumption came from.

Whatever the case, set your mind on these three things: bitrate, bitrate, and bitrate.

[jagabo]

I would simply say that reducing the resolution is one of the compromises to consider when you compress a video.

[deadrats]

I would simply say that reducing the resolution is one of the compromises to consider when you compress a video.

i can't understand why anyone is compressing video (or audio for that matter) these days. when you consider that you can buy an internal SATA 1.5 terabyte hdd for around $100, i see no valid reason to try and reduce the size of a video by re-encoding it to a lower resolution, reduce the bit rate, etc.

the only valid reason i can think of to re-encode is if the source file has some odd characteristic (like the weird 1440x1080 files i asked about some time ago) and even then you should choose the same resolution, bit rate, etc.

[poisondeathray]

i can't understand why anyone is compressing video (or audio for that matter) these days. when you consider that you can buy an internal SATA 1.5 terabyte hdd for around $100, i see no valid reason to try and reduce the size of a video by re-encoding it to a lower resolution, reduce the bit rate, etc.

the only valid reason i can think of to re-encode is if the source file has some odd characteristic (like the weird 1440x1080 files i asked about some time ago) and even then you should choose the same resolution, bit rate, etc.

Totally agree, and HDD's are getting cheaper every month. Or the other odd characteristic would be restoration efforts (like a badly mastered or noisy DVD for example)