What is the movie quality means....

[BxWrapper]

This might be unpopular subject. After all, what does quality movie mean when you transcoding it to DivX ?

Let see how the quality movie is CONSISTENTLY down graded during the process.

1. Movie scene are recoded in high quality celluloid. It means that you record a snapshot of unlimited resolution real image down to something like 5000 x 5000 grain/pixel resolution of film.

2. Then you have to to scan the film images to make digital movie master. You talkking about down grade the image again to 3000 x 3000 resolution.

3. You need to transcode the master digital movie to MPEG2 format (DVD) which has resolution only 720 x 576 (PAL).

4. Then probably you need to resize the movie to somewhere 640 X ... or ever heard some one just happy to resize it to 416 x ... ??


Think about it. Without any compression, the image quality has been lostso much for each step / transcoding. Feel free to add macro-blocking / blurriness factor to your nice movie due to image compression.

After all, what is the movie quality means when we are talking about DivX.

Cheers..

[DJ Bobo]

As long there is no real visual loss, you can't talk about real quality loss.
You must differ between theory and practice.

[manono]

And since we're working off the DVD, that's what we measure our results against. I don't see how the original film image figures into the equation at all. There's nothing we can do about that.

[BxWrapper]

That's exactly the problem....:)

When you referring to perceived quality, it is become subjective.
If you are talking about facts/data, they speak for themself, aren't they ?

So, tell me how much subjective is when somebody saying:

1. My rip's size is 690 MB. I feel that I will have a better movie if I can make it a full 703 MB rip.

2. I always use 64-bit floating point IDCT to get a better quality and 32-bit integer is sucked.

3. Always use bicubic resize because it is more accurate...well..I don't care if most of the video card / software do bilinear resizing for full screen display..:-)


Ignoring that we are working on already distorted source.

No offence..OK..just brainstorming.....;-)

Cheers

[DJ Bobo]

Ever heard about the "relativity theory"? ;)

[manono]

1. Heck, I redo them if they come out to 695MB. I freely admit that my eye can't tell the difference, but I just feel better knowing that I've gained another 1% in quality.

2. I just take Doom9's word for it that there's no perceivable difference.

3. Is that true? If so, I didn't know that. Even newer vid cards use BiLinear?

No offense taken.:) We're just an argumentative bunch around here.

[-h]

2. I always use 64-bit floating point IDCT to get a better quality and 32-bit integer is sucked.

That depends on what authored the DVD - if the particular iDCT implementation of the authoring application was integer-based (this is quite likely), you would get the best results by using the same integer iDCT when transcoding. There's no way to know unfortunately.

If it is a quality production house, they will stick to IEEE-spec iDCT accurary however, in which case you'd want the floating point variety (or an integer implementation which is just as accurate).

3. Always use bicubic resize because it is more accurate...well..I don't care if most of the video card / software do bilinear resizing for full screen display..:-)

I suppose, however this will cause higher quantizers to be used, which will result in lower subjective quality. That depends on what the viewer prefers though - an blurred but accurate video, or sharp but blocky.

-h