Question regarding resolution setting

[Kilcher]

Hi everyone, I have a rather stupid question on the output resolution setting in regards to bits/(pixel*frame) and the compressibility test. I read the Gknot guide on Doom9, but am still confused on the ideal setting.

Which is more important, adjusting output resolution by following the approx. .20 guideline for bits/(pixel*frame) for a single cd movie, or adjusting the output to accomodate the >40% on the compressibility scale?

I did a compressibility test on an 87 minute NTSC interlaced movie. The .20 I originally set was way off, as this only yielded 26.4% for the compress. scale. I adjusted the resolution so the compress % was >40% (I ended up with 41.4% and .325 for the bits/pixel value. By doing this the ouptut resolution was relatively small compared to the original resolution (about a 40% reduction). Did I do something wrong, or are these numbers not so far-fetched?

Another related question that's been baffling me:

I ripped a movie (100 minutes NTSC interlaced) just before the example movie I gave. The thing that was strange was that even though the movie was about 13 minutes longer (100min vs. 87 minutes above), my final output resolution (as set with compress>40%) was around 20% larger than the 87min movie. This just seems counter-intuitive. I would assume that a larger movie (length-wise) would require a SMALLER resolution than a movie that was shorter. the only explanation i could come up with was that the compressibility for the 100 minute movie was a lot better than the 87min one. Am I on the right track?

Thanks everyone and sorry for the long post.

[diji1]

'owdy :),

question 1: - only use the bits/pixel*frame number to set the resolution before ur first compression test, after this ignore it. Run the compression test and use that figure only to set the final res u decide on ( for better quality go *at least* 65%, preferably higher depending on res ). I always run a compression test each time i change the res but its easy to get it right after 2-3 compression tests i find.

question 2: ( and 1 ) - no two movies are the same and i get values that are very high and very low as well as well as midrange, so it sounds like ur doing everything ok.

[Kilcher]

I think I understand now. Does making multiple compress test change the values? I thought that the compress test would be the same regardless of output resolution.

Just to make sure I got it down:

Set the bits/pixel to approx .20 (for a single cd) and leave it at that. Then do a compress test. Adjust the output res so that comp % is atleast 40%-65%, then do another compress test? What is the reasoning for doing another compress test if that's what you meant? Thanks for being so patient with a newbie such as myself!

kilcher

[diji1]

Lo again :)

Does making multiple compress test change the values? I thought that the compress test would be the same regardless of output resolution.

Sorry, i think i was a little unclear :( - i meant do another compression test if you change the resolution after the first compression test you run. If that first check yields the settings you want then just encode it, but if you change res then run another test as res does affect compressability

Theres no need to run another compression test if ur happy with the percentage it gave you.

Set the bits/pixel to approx .20 (for a single cd) and leave it at that. Then do a compress test. Adjust the output res so that comp % is atleast 40%- 65%, then do another compress test? What is the reasoning for doing another compress test if that's what you meant?

That's all a good method - i meant adjust the resolution after running at least one compression test to get ur desired percentage and then run another to confirm that the value is right. It's like guessing the resolution based on what the previous compression test gave you, but u must run another to confirm that the res you have set is in fact going to give you that percentage.

Sorry for being unclear - in brief this is what u should do:

1- set res using bits/pixel*frames once only.
2- run a compression check.
3- if ur percentage is how u want - start encoding, if not set the res so the percenatge is what you want, then go back to step 2.

btw, if u can't avoid it or ur happy it doesn't matter, but 40 - 60 % seems a little low for me. :)

[Kilcher]

Thanks for your answers diji1. I think I really got it this time!

This is what you said:

1- set res using bits/pixel*frames once only.
2- run a compression check.
3- if ur percentage is how u want - start encoding, if not set the res so the percenatge is what you want, then go back to step 2.

btw, if u can't avoid it or ur happy it doesn't matter, but 40 - 60 % seems a little low for me.

for 1. you said to adjust the res using bits/pixel once, but doesn't changing the rez affect that number or is the number meaningless after the first compress test?

you said that (for you) 40-60% on compress is a bit low. Can you tell me what sort of output rez you are getting with compress higher than 60%? For example, a 2 cd movie of mine of 93 minutes (NTSC FILM) is .443 for the bits/pixel and 59.3% for compress at a resolution of 576x320 (or approx 81%W-zoom/67%H-zoom. Does the ouptut res seem ball-parkish for you?

Last question! Sorry..

Would a higher bit/pixel and compress value and a small output rez be better than a lower bit/pixel and compress value but a larger output rez? Or are they the same, since you're compensating with smaller output rez for higher quality and larger output rez for lower quality.

To put it more concretely:

Would these be the same or is one better than the other:

Example 1:

Bits/pixel: .491 compress: 66.1% of .743 (btw what does the second number mean? is that important?) with output rez of 544x304

OR

Bits/pixel: .328 compress: 44.2% of .743 (btw what does the second number mean? is that important?) with output rez of 672x328

Will they be of the same quality if I play them on the monitor or tv?

The reason I'm asking is that I've heard of people encoding at nothing below a certain resolution, so I'm afraid that quality might be not as good if I encode at too low of a resolution, but on the other hand, the quality should be a lot better. I'M BAFFLED.

Thanks again for your time!

Kilcher1

[diji1]

Thanks for your answers diji1. I think I really got it this time!

No problem! :)

but doesn't changing the rez affect that number or is the number meaningless after the first compress test?

Meaningless after 1st test, yes, *only* use it to get the res setting *before the first* compression test, after that ignore it.

Can you tell me what sort of output rez you are getting with compress higher than 60%?

No i cannot as no two movies are the same - sometimes figures are out of the ballpark. For what its worth i consider having a higher percentage figure to be a lot more important than a higher res setting to the final quality - try it urself if u don't believe me. The movie ur doing seems like it's a hard compressor going by ur figures ( pr0n ? ) but nothing out of the ordinary. Just ignore the bits/pixel*frames number - remember, it only matters before the first test. Going on what i just said, i would drop the res ( or change resize filters ) more to get *at least* 65%, preferably 75% for ur percentage - that's me tho :), u may prefer the higher res.

Last question! Sorry..

oi, don't be! ;)

Would a higher bit/pixel and compress value and a small output rez be better than a lower bit/pixel and compress value but a larger output rez?

As i said above, personally i alway sacrifice res for quality - it looks better imho ( and me m8+s 2 ) - however, i dislike lo res ( anyone like it...? ) so i have a "rule" of not going below 512*xxx, i add another cd for the backup if thats the case ... course, im not too fussed about size of my rips :)

btw what does the second number mean? is that important?

good question - ahhh, wef ? anyone else ?

Will they be of the same quality if I play them on the monitor or tv?

u know, it's hard to say i think, because everyone has a different opinion of what's "good" quality and whats "bad". For example, i think divx looks better on the monitor whereas 3 other ppl i know prefer the TV - i prefer the extra detail, they prefer the blurriness of the teev to mask macroblocks and noise ... i think the best way to find out is to do it urself.

[pdontthink]

> Meaningless after 1st test, yes, *only* use it to get the
> res setting *before the first* compression test, after that
> ignore it.

wow, really? that sounds crazy, i mean, i don't doubt
you know what you're talking about, but surely the bits
per pixel figure is important all the time, since the
fewer bits you have to render each pixel, the lower the
quality of the image, no? why would you just throw that
number away just cuz you have a more accurate idea of
the movie's compressability? if I go back and reencode
a movie with much bigger rez but the bits/pixel goes into
the yellow, will it not look much worse???

[pdontthink]

my mistake - please ignore previous post...
i wasn't thinking clearly :rolleyes: and
i got it backwards; i tried to say that the
compressability load percentage (is that figure
as simple as I'm thinking... 100% - %compression??)
goes UP when you increase rez, but it does in
fact go DOWN, as you'd expect.

while I'm at it, can anyone please expand upon
the WHYs of doing a new compression check every
time you change the rez? thanks!

[pdontthink]

OK, diji, I am gunna have to disagree with you about
re-running the compression check every time you change
the rez. See theReal's explanation at the end of the
thread at:

http://forum.doom9.org/showthread.php?s=&postid=96900

If theReal is right, then there is no such need. And
since I didn't read any reasons for doing so in your
posts, and since when I re-run comp checks and don't
see any difference (tho I hadn't been paying close
attention), I am gunna believe in the 'run once'
theory. Prove it wrong! :)

[diji1]

i think theReal actually agreed with me in the other thread - like it's no hard and fast rule that u *must* run a compression check *every time* u change res, it's just more accurate. the extrapolation that gknot makes to get the percentage isn't always accurate - i'm guessing it would be impossible to write an app that could get it 100% accurate. thats what i noticed and i was paying close attention btw ;).

certain situations i don't always run the test again - say i'm aiming for 90% compression check and im on 95 then im pretty sure i can drop one res setting without the test.

[pdontthink]

yer absolutely right. :D


(i've one more question, see the other thread, linked in my post above)