View Full Version : Am I cropping correctly?
Jeremy Duncan
14th March 2008, 23:22
I found this website: Here (http://www.doom9.org/index.html?/mpg/dgindex-frameserving.htm)
And it mentioned cropping like this:
Crop(0, 60, 0, -60)
I am using ntsc dvd's and they are a 720x480 resolution.
After I crop it using the crop code above, and resizing it multipliying it by 2.668.
The resolution becomes 1920x960.
All I want to do is get rid of the black bars on my 16:9 screen when watching this 720x720 movie.
I'll upload a sample.
http://rapidshare.com/files/99582802/Sample.zip.html
Besides wanting to know how to get rid of the black bars.
I want to know how to crop so I get the proper aspect ration for the sample I uploaded.
It looks a lot more anamorphic in the sample with wmp11 than it does in ffdshow and mpc.
I want some sample crop code I can try, please. :)
smok3
14th March 2008, 23:48
1920 example, if you want to change PAR to square;
http://somestuff.org/resizeCalc.php?ssmw=720&sar=1.215&sar2=&ssmh=480&CT=60&CL=&CR=&CB=62&trw=1920&dar=1&dar2=&mod=&padw=&padh=&css=&doit=true
Jeremy Duncan
15th March 2008, 01:07
smok3,
I read: Here. (http://forum.doom9.org/showthread.php?p=1089603#post1089603)
That you said reducing dar error is important.
I think this may be more important than getting rid of the black bars.
Crop(0,58,-0,-58) # this gets rid of most of the black bars.
DAR error = -2.14
Crop(0,8,-0,-8)
DAR error = 1.019
Crop(0,10,-0,-10)
DAR error = 0.159
I know this is a stupid question, but I would like to know which it less of a dar error: DAR error = 0.159 or DAR error = 1.019.
And could you please explain what Dar error means, please.
[P]ako
15th March 2008, 03:51
Well, the picture itself is 2.31 (panoramic). So, if you cropped the top and the bottom and stretched the video to a 16:9 resolution, then the picture will look distorted. Which I think that's what you wanted to do in the first place. Of course, when converting crop the black bars, but set the aspect ratio properly.
I personally will live with the black bars.
driftr
15th March 2008, 04:20
Pako's pretty much got it right.
What you need to understand is your TV is 16x9 which is optimal for HDTV and most movies. Other movies are filmed in other aspect ratios, the most common equaling 2.35/2.4 (scope) So when you play these or other semi common aspect ratios on your tv (1.86 is another) you will have black bars on your tv.
So crop the bars out of the video you're encoding, keep the original aspect ratio, and if it's not 16x9 you will have black bars on playback. Plain and simple.
45tripp
15th March 2008, 04:27
All I want to do is get rid of the black bars on my 16:9 screen when watching this 720x720 movie.
so what you're saying is that you're willing to sacrifice the sides of the picture?
if you want 2.35/1 to 16/9, you have to cut into the sides.
Crop(90,56,-90,-62)
Lanczos4Resize(1920,1080)
so I get the proper aspect ration for the sample I uploaded
then remove the bars,
but you'll still have borders on playback (2.35:1):
Crop(0,56,-0,-62)
lanczos4Resize(1920,816)
That you said reducing dar error is important.
I think this may be more important than getting rid of the black bars.
it's not!
Crop(0,58,-0,-58) # this gets rid of most of the black bars.
DAR error = -2.14
Crop(0,8,-0,-8)
DAR error = 1.019
Crop(0,10,-0,-10)
DAR error = 0.159
dead wrong aproach
you either want to crop the black bars or you don't.
if you wish to reduce the dar error, there are suggestions offered
by the calculator, analogous to your initial goal. (i.e. crop an extra pixel here or there)
I know this is a stupid question, but I would like to know which it less of a dar error: DAR error = 0.159 or DAR error = 1.019.
it's what you'd expect.
And could you please explain what Dar error means, please.
it's the deviation of output dar from input dar in a percentage.
say input has a dar of 1.77 and output has a dar of 1.75.
a 1% deviation. try and notice it...
why the deviation? to get mod-16 resizes.
important for compression, but with just playback your concern,
you have no need to stick to mod-16.
also, smok3's default pars are set for 704 widths.
gl
smok3
15th March 2008, 12:14
also, smok3's deafault pars are set for 704 widths.
as opposed to? - there is no reference in the math to any fixed resolution, everything is based on PAR, so numbers please? :).
Jeremy, you can get rid of black borders, get mod16 resolution and fix the DAR a bit with some additional croping, importance is on case by case or user by user basis.
45tripp
15th March 2008, 14:18
PAL 12:11 , PAL W 16:11 , NTSC 10:11 , NTSC W 40:33
(* the calc actually has pal values slightly off the above, i.e 12.05:11)
as opposed to:
16:15 , 64:45 , 8:9 , 32:27
SeeMoreDigital
15th March 2008, 15:31
Lets start with determining the "movies" aspect ratio by playing your "Blade.demuxed.m2v" sample in VLC player and taking a "SnapShot". Which will give us an image (including black mattes) of 853x480 pixels: -
http://i28.tinypic.com/29gim45.png
When you "hard crop" the black mattes away you are left with an image of 850x363 pixels: -
http://i28.tinypic.com/9sxnas.png
850 divided by 363, gives you a movie aspect ratio (MAR) of 2.341597796:1.
If it's your intension to re-size your source to 1920 pixels wide, then 1920 divided by 2.341597796, gives you a pixel height of 819.9529412. Which rounds-up pretty nicely to 820 pixels.
So you need to re-size to 1920x820 pixels: -
http://i26.tinypic.com/2dbjpki.png
Cheers
Jeremy Duncan
18th March 2008, 22:50
I like this code:
Crop(90,56,-90,-62)
Cheers
smok3
19th March 2008, 21:40
Immersion; i have just reread some of the theory, there is bunch of possibilities with NTSC sampling matrix size, however PAR is the same for the most common ones, so to begin with, user would have a least two choices:
a. it will have to know what is picture sampling matrix width and height, and what exactly sampling matrix is in the 1st place (as opposed to actual active picture)
or
b. have a nice life with some less knowhow and just blindly use my calc ;) (the calcs won't be a lot off even without that knowhow).
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.