Newbie: Neccessary to resize video?

[Newbie123]

I followed the Divx 4 encoding guide on the website and I have a question regarding resizing of the video. Is it really neccessary to resize the video. What is the advantages of doing so? What is the ideal resolution to resize the video to?

[aleksander]

Read doom9's guide about Gknot and use it for resizing.
What is the advantage??? Simple - better visual quality.
aleksander

[pale]

Alexander wrote: What is the advantage??? Simple - better visual quality.

This is quite misleading. In fact, to achieve the best possible visual quality, not to resize is the (only) way to go. However, most of people making divx's are interested in fitting a movie into a certain size. Getting the best possible quality while meeting the target size may (or may not) necessiate resizing, depending of the type of the source, length of the movie and desired size.

It all comes down to question whether the benefits of higher bits/(pixel*frame) value outweight the negative impact resizing has. I just wrote a long answer about this under topic "NTSC / 29.XX / Interlaced source problem" http://forum.doom9.org/showthread.php?threadid=14936

[b0b0b0b]

I think it depends in part on your output device. If you're watching on a digital display and magnifying the image, you're going to have different needs than someone who feeds the video into a 13" tv or a 60" rear projection tv.

Basically, if you're watching on a traditional tv you should never leave your picture bigger than 640x480 or whatever the native resolution is. Well, after writing that sentence I'm left wondering whether that's true. After all, DVD's are 720x480 regardless of the data's aspect ratio 4:3/16:9. Maybe I'm full of baloney.

All I know is I had better results when ripping Kids when I left it at full resolution because in the opening scene, when downsized, the detail of the girl's hair was completely lost. The point is, when you downsize, you lose detail. I guess on the other hand, when you compress you lose detail too. So I think the tradeoff you need to ask yourself is if you want to uniformly lose detail on the whole image, or if you trust the codec to lose detail where it thinks you wouldn't have needed it anyway. It sounds like most people don't fully trust the codec. i.e., in the case of 4.xx, look at the people who are tweaking quantizers.

[pale]

Quantisizers are a very different thing from resizing, they affect the compression itself. Resizing, on the other hand, affects the source before it is compressed.

As in all lozzy compression a very simple rule applies: shit in, shit out. Meaning if your source material is crap, it is not possible to get good quality, no matter what codec or setting you use. Thus by keeping the source as good as possible (by not resizing it) you are actually giving the codec a better chance to show what it's worth.

Unfortunately, there's no implicitly right solution, source and target size will determine whether the advantages of not resizing will be great enough to compansate the need to settle with lower bits/(pixel*frame) (which means that that better looking source has to be compressed more heavily in order to achieve target size).

[b0b0b0b]

adjusting quantizers across zones tells the codec where to spend more bits or fewer bits.

resizing the picture says, before the codec even gets to it, use less bits, period.

That's what I was saying.

[pale]

Seems that I interpreted your earlier post wrongly, my mistake.

[b0b0b0b]

no big deal! :D