Scintilla wrote:Pankyy wrote:Yeah, what do you think? It's better to upload an AMV of 7MB with the WMA format so people can download it faster than an AVI or MPG one, whichs video quality is too much superior?
What about a <I>7 MB AVI or MPEG</i> whose video quality is clearly NOT superior? :?
You have to remember that <i>container format does not have any bearing on visual quality.</i> There are huge, gorgeous WMVs and RM files, and there are tiny, crappy-looking AVIs and MPEG-2s.
In this vein, I'd like to point out that
filesize really means just about nothing with regards to quality. It's like saying that one microprocessor architecture running at 3.4 GHz must be faster than the other one running at 2 GHz just because the number is bigger: you're ignoring way too many factors.
I'd like to use two of my own videos as examples, because I think they illustrate my point really well.
Let's take this one. Its size is 8.53 MiB. What do you think it looks like? What resolution do you think it is?
You probably won't be able to tell without downloading the thing, and I doubt you really want to, so I'll just say what the resolution is and tell you its current "visual" score. If you want to look at it for yourself...well, download it.
The file is 704x480, with sample / display aspect ratios set so that the thing appears correctly on square-pixel displays. Its visual score is 9.50.
The file was encoded with x264 r501, is roughly a minute long, and really isn't all that complex. All of those factors (and others) contribute to its small size.
But it still doesn't look that bad.
Here's another one. The copy on local is 73.3 MiB. Now, if we're going by filesize alone, this thing has to possess godlike clarity.
It doesn't. Heck, it probably looks about the same as "loli-pop". People seem to agree: its visual score is 9.25.
However, it's around three times as long, was encoded with a different encoder, and is way more complex. (In particular, it involves lots of static and noise, which are two things that tend to not compress very well.) The audio is also 256 kbps MP3, which contributes a bit.
===
These examples don't describe everything that can affect filesize but end up meaning very little in terms of quality. However, I hope that this makes the point.
To close out, I'd like to use a third example. This time, I have two files that were generated from the same encode.
I have two encodes of Reflections of Style 3 (a recently-completed multi-editor project, duration 17:44, lots of fades and lots of action) that have the following filesizes:
Code: Select all
nevrast:~/Movies trythil$ ls -l Reflections\ of\ Style\ 3.*
-rw-r--r-- 1 trythil trythil 292866048 Oct 10 01:49 Reflections of Style 3.avi
-rw-r--r-- 1 trythil trythil 195288770 Oct 9 21:16 Reflections of Style 3.mp4
Now, which one do you think looks better?
You'll find out soon.