Movie resolution, aspect ratio, quality, size, and compression relationship

[mario595]

I am not familiar with video editing so I have a few questions. I have read a few good articles about resolutions/ratios:

http://en.wikipedia.org/wiki/Display_resolutions
http://events.hometheaterforum.com/home/wsfaq.html

If I rip my copy of The Dark Knight to the hard drive and look at it in tsMuxer. The resolution shows as 1920:1080p. It is a 34GB file. I believe most movies ripped from blu-ray are 25-35GB. I have a bunch of other (recent) movies that exist on my hard drive that were taken from Blu-Ray or HD/DVD sources that range in size from 6GB - 12GB. I did not manipulate the files myself so I do not know what was done to them. When I look a them in tsMuxer I see the following resolutions:

1280:528p 1920:800p
1280:532p 1920:818p
1280:536p 1916:1038p
1280:544p 1920:1080p
1280:688p
1280:720p

Are these lower resolutions a result of compressing the file to make it smaller for bandwith reasons? Does the vertical resolution dicate image quality? If I compress the Dark Knight down to say 6GB will my resolution change?

Or is it that vertical resolution is dictated by the aspect ratio of the film:

1280:528p (2.42) 1920:800p (2.4)
1280:532p (2.40) 1920:818p (2.35)
1280:536p (2.37) 1916:1038p (1.85)
1280:544p (2.35) 1920:1080p (1.77)
1280:688p (1.86)
1280:720p (1.77)

That would lead me to believe that the 1280 movies are 720p and the 1920 movies are 1080p. But aren't all Blu-Ray and HD-DVD movies 1080p on disk? Do you compress a 1080p movie down to 720p to reduce the size? Or can you have two copies of the same movie; one ripped from the disk and a copy that has been compressed to a smaller file size and have them both show the same resolution in tsMuxer? How does this work?

Thanks in advance.

[edDV]

Most BluRay movies are 1920x1080p square pixel 16x9 although other resolutions are allowed. Since most movies are other than 16x9 (1.77 to 1) aspect ratio, they are encoded letterbox.

The "1920" samples you list above were probably cropped to eliminate letterbox. Since letterbox is flat black, compression would be little affected.

The "1280" samples were probably downsized or were broadcast 1280x720 samples.

Resolution is independent of bit rate. If you compress a 1920x1080p sample from 34GB to 6 GB you will see considerable compression artifacting.

[jagabo]

And smaller frame sizes require less bitrate to keep image quality. So if you shrink a 1920x1080 video down to 1280x720 it will require less bitrate (hence a smaller file size). You are exchanging resolution for fewer compression artifacts.

[mario595]

So if you start out with a 1920x1080 movie, is it true that you can compress it down a bit and still have a 1920x1080 resolution? Can you then choose to compress it to 1280x720 to get a big reduction in the file size? Is that how this works? If you start out with a 30GB 1920x1080 movie how small (roughly) does the file get when you make the jump to 1280x720? How small (roughly) could you make it and maintain 1920x1080?

From your experience; if you have say a DLP rear projection TV at what point would you start to notice a difference in quality as you compress the file?

[manono]

So if you start out with a 1920x1080 movie, is it true that you can compress it down a bit and still have a 1920x1080 resolution?

You can compress it as much as you want - make the file size as small as you want - and still keep it at 1920x1080. Beginning at a certain point, though, it may not look so good.

Can you then choose to compress it to 1280x720 to get a big reduction in the file size?

Well, you wouldn't first shrink the size keeping the resolution and then secondly shrink the resolution and the size some more. You'd do it all at one go. Encode as few times as possible.

If you start out with a 30GB 1920x1080 movie how small (roughly) does the file get when you make the jump to 1280x720?

1920x1080=2073600 pixels
1280x720=921600 pixels. The lower resolution one has just under half the pixel count. So, removing the audio tracks from the equation, you could shrink the video size just under 50% and retain similar quality at the cost of lower resolution.

How small (roughly) could you make it and maintain 1920x1080?

There's no way to tell. It depends on how much it was originally compressed from the source and your tolerance for lower quality. You could do quality-based encodes, rather than bitrate-based ones, and decide for yourself the quality you want to retain.

From your experience; if you have say a DLP rear projection TV at what point would you start to notice a difference in quality as you compress the file?

Again, there's no way to tell.

[jagabo]

If you have a 1920x1080p HDTV with pixel-for-pixel mapping, a very sharp and clean source, look closely, and know what to look for, any resizing or recompression will be noticeable. It's up to you to decide how much degradation you will accept to get smaller files. It will also vary depending on the source.

[edDV]

...Can you then choose to compress it to 1280x720 to get a big reduction in the file size? Is that how this works? If you start out with a 30GB 1920x1080 movie how small (roughly) does the file get when you make the jump to 1280x720? ...

A simple change of resolution would not change file size. File size is a function of bit rate, not resolution.

You operate on the file separately for resolution (downsize) and recode bit rate. The two are separately optimized for your perception of quality on your display equipment. It would appear different on other displays and by another viewer.

For your 1024x768 display, a resolution reduction to 1280x720p and a 50% bit rate reduction may show little quality loss. Additional bit rate compression will degrade quality and file size to your taste.

Keep in mind that any recode will cause some amount of quality loss.

Also expect this recoded video to look much worse on higher quality displays.

The above assumes progressive 23.976 fps film source. TV film source captures have more processing steps (frame decimate or inverse telecine). Interlace TV captures are a different story.

[mario595]

A simple change of resolution would not change file size. File size is a function of bit rate, not resolution.

I take it that "bitrate" is the amount of bits per second that is being used to create the image on the screen. The higher the bitrate the greater the quality and/or resolution.

If you convert a video from 1920x1080 to 1280x720 without compressing then you will reduce bitrate simply because you are dealing with less than half the pixels right?

Or are you saying that I can have a movie that is 30GB at 1080 and make it 720 with the same amount of bits?

[jagabo]

I take it that "bitrate" is the amount of bits per second that is being used to create the image on the screen. The higher the bitrate the greater the quality and/or resolution.

The higher the bitrate the higher the quality. The greater the resolution the more bitrate a video will require to keep from degenerating into macroblock artifacts.

If you convert a video from 1920x1080 to 1280x720 without compressing then you will reduce bitrate simply because you are dealing with less than half the pixels right?

No. If you don't compress again you will get a gigantic video file. So you compress again to get a file of manageable size. You pick the bitrate when you compress. That determines the file size. Or you use a constant quality mode where you pick the quality and the bitrate (and hence file size) falls where it may.

Or are you saying that I can have a movie that is 30GB at 1080 and make it 720 with the same amount of bits?

Again, you pick the bitrate when you encode the video. A smaller frame size won't require as high a bitrate to maintain quality. But by downsizing the frame you will lose some quality. Then by re-encoding with a lossy codec you will lose more quality.

[edDV]

Bit rate and file size are directly related. Resolution is an independent variable.

[edDV]

A simple change of resolution would not change file size. File size is a function of bit rate, not resolution.

I take it that "bitrate" is the amount of bits per second that is being used to create the image on the screen. The higher the bitrate the greater the quality and/or resolution.

So resolution being independent of file size or bit rate for compressed video, you can say that the higher the bit rate the greater the quality for a given resolution.

I think you are cofusing compressed video with uncompressed. That 35GB Blu-Ray file will decompress to somewhat less than a TB at 4:2:0. Once decompressed resolution and bit rate are directly related.

[mario595]

This is great info thanks for the clarifications...

If you take a movie and compress it to a bitrate of x at 1080p and then do it again at the same bitrate but at a resolution of 720p the file sizes will be roughly the same.

I assume then that the better quality file (on a 16x9 monitor) would have to do with the value of x. You don't want too low of a bitrate with a 1080p resolution and you eventually reach a point of diminishing returns as you increase the bitrate of a 720p. Is this correct?

[edDV]

This is great info thanks for the clarifications...

If you take a movie and compress it to a bitrate of x at 1080p and then do it again at the same bitrate but at a resolution of 720p the file sizes will be roughly the same.

I assume then that the better quality file (on a 16x9 monitor) would have to do with the value of x. You don't want too low of a bitrate with a 1080p resolution and you eventually reach a point of diminishing returns as you increase the bitrate of a 720p. Is this correct?

Lets assume you start with uncompressed 23.976 fps progressive video.

For a 1080p source and 1080p display, you will see a decline in image quality as you reduce bit rate. Many of the compression artifacts only affect moving objects or pan/zoom*.

If you downsize from 1920x1080p to 1280x720 on a 1920x1080p display, you will see an immediate resolution hit. That said, you will find that other compression artifact issues will be about the same at half the bit rate vs. those seen at 1080p.

If you start with a compressed file, there will be additional recode losses.


* Normally variable bit rate is used to allocate more to high motion scenes and less to low motion scenes.

[mario595]

Ed,

I lost you at that last part, what exactly are the two objects you say will have roughly the same compression artifact issues?

[edDV]

Ed,

I lost you at that last part, what exactly are the two objects you say will have roughly the same compression artifact issues?

Starting from uncompressed 1080p/23.976 source, 1080p at say 16Mb/s will have similar compression artifacting vs. 1280x720p at 8Mb/s assuming the same codec is used.

These bit rates are similar to what is used for broadcast MPeg2 video. H.264 is good for another 2x compression to about 8 Mb/s for 1080p/23.976 and 4Mb/s for 720p/23.976.

Any of these bit rates will resemble broadcast video quality, not Blu-Ray which uses much higher bit rates (up to 35 Mb/s).

[jagabo]

In my experience, half the pixels requires a little more than half the bitrate. Maybe 60 percent rather than 50. It will also depend on what resizing method you use. For example, a precise bilinear resize requires less bitrate than a precise bicubic resize. The sharper the resizing filter the more bitrate you'll need.

[edDV]

In my experience, half the pixels requires more than half the bitrate. It will also depend on what resizing method you use. For example, a precise bilinear resize requires less bitrate than a precise bicubic resize.

I agree. Same with VBR issues. I was just keeping it simple until he catches on to the basics.

Broadcast HD video (film based) contains repeat fields/frames that also affect bit rate.

1080i (telecine) repeats two fields out of ten (20% pad).

720p/59.94 repeats 3 frames out of 5 (60% pad).