get it right
How it gets there: the divx codec, by default, inserts it into it's decoder output. So people use divx with the logo enabled, and edit (decode, cut, re-encode) videos, and this way the logo gets into everything they create. While divx only shows the logo in the begin, if you cut pieces of video it shows in every piece again, so you see it during the entire AMV. If you use the registered version, you can disable the logo, otherwise you can't. So people editing videos either use the demo version of divx, or they simply don't think about disabling the option. To disable it: start menu -> programs -> divx -> divx codec -> decoder configutration utility.
If you have the demo version of divx, you have 2 choices. 1: Get the registered version, 2: Get xvid, a program which is free, and can decode and create divx compatible videos, without any demo restrictions. If you installed xvid, you can use this guide to configure it to produce videos which are compatible with both xvid and divx. Also disable "film effect" in the "playback" tab when decoding videos for editing. What it does is adding noise, which the subsequent encoding will waste bandwidth on.
If the video is edited correctly, you shouldn't have them here. How they get there: There are various common resolutions to store videos. 4:3 square pixel resolutions such as 320x240, 640x480, etc, are like you'd expect, exactly the "4:3 frame" - it is to show on a TV, or in a player, as a whole, with an aspect ratio of 4:3. Have you ever wondered why VCD video is 352 pixels wide, and DVD video is 720 pixels wide, slightly more than twice the VCD resolution? And why you sometimes see 704 pixels wide video? It is because if a video is 720 pixels wide, the 4:3 frame is only 704 pixels (exactly twice VCD), and it has 8 pixels of padding on the left and right of the scanline, which usually shows as black borders. Below is a DVD frame at native resolution. In an image editor you can measure the width of the picture and the borders.
For PAL, the widths, and editing procedures, are conveniently exactly the same as for NTSC, but the height of DVD video is 576 pixels. Note that a real/analog NTSC frame is 712x486, but everyone conveniently ignores that and uses 704x480 as the 4:3 frame for digital video.
What is the correct way to convert 720 pixels wide video to a resolution such as 640x480, or 320x240, or 352x240? First crop from the left and right until the width is 704. Usually crop 8 from both sides, but the borders may be asymmetric. Then resample/resize to your desired output size.
What is the correct way to convert 352x240, 320x240, 640x480, etc, video to 720 pixels wide for, for example, creating a video DVD? First resize to 704x480, then add borders to make it 720x480.
For 16:9 (anamorphic widescreen), the same applies: the 16:9 frame is 704 pixels wide, the rest is borders.
I'm not going into the problem of interlacing here, it is explained on another site: http://www.100fps.com/
I once encountered this horrible video editing botch job from p2p, a video of alizee - j'en ai marre, live, by M16 (is M16 the one who edited this?)
the TV studio sent the first few seconds as interlaced video (720x576 PAL), then switched to progressive, by discarding fields (288 effective scanlines, the other scanlines are linearly interpolated) Which operations have been done on this:
Note how the picture is sharp and all diagonal lines look correct. While the aspect ratio is obviously wrong here (this is normal for video material), a player could correctly show it as 4:3.
The text and action safe rectangle. Note how the subs and karaoke, added by the fansub group, stay within the action safe rectangle here.
a subtitle added by the original creators of the anime, exactly within the text safe rectangle.
This is from a commercial video which has been on TV. The subtitles are outside the text safe rectangle, but well within the action safe rectangle. Everything in the second screen is exactly within the text safe rectangle.
Fansubs, falling within the action safe rectangle.
Given the definition, and some good examples, the problem: most fansub groups don't care about this, and make their subs almost touch the bottom of the screen, using as argument "We don't want to watch it on a TV, or we adjust our TV's. So you should do the same". Some screenshots follow:
Fansubs by AonE (hard subs), falling well outside the action safe rectangle. Notice how the scrolling text i got in there by coincidence is exactly within the text safe reactangle. Fansubs by "No subs" (soft subs), falling well outside the action safe rectangle.