GripCrop bug?...maybe!
 

source:
Peter Pan(Disney)
4:3
29.970 fps
film
progressive

using this script from MovieStacker(overscan 2),
i got perfect aspect radio:

Undot()
Asharp(1,4)
BicubicResize(448, 448, 0, 0.6, 0, 0, 720, 480)
Deen()
DCTFilter(1,1,1,1,1,1,0.5,0)
AddBorders(16, 16, 16, 16)

but using this script with GripCrop,
:arrow: i got wrong aspect radio with big black borders on top and botton
8O 8O 8O

Undot()
Asharp(1,4)
GripCrop(480,480,overscan=2,source_anamorphic=fals e,dest_anamorphic=false)
GripSize(resizer="BiCubicResize")
Deen()
DCTFilter(1,1,1,1,1,1,0.5,0)
GripBorders()

and without overscan i got the same borders on the top and botton too.
:?

any help please?

Always preview your movie, and if the results are wrong with GripFit, use AviSynth's resize ;)
If you want to be 100% correct every time, then use AviSynth's resize :!:

-kwag

ok yes Kwag,
i did little samples not the full movie!
:wink:

is the first time that i see wrong aspect radio with GripCrop!
...very strange.

thanks for the hints !
:D

i re did the scripts and encode using the
scripts again with and without overscan!

:(
just the same...
using GripCrop is anamorphic and give big black borders in the top and botton.

it's a bug, problem or "don't know what"?
:?

Other thing : if you want to use bicubic precise (b=.6, c=0) as you in your example, you must use avisynth resizing. Gripcrop always uses bicubic standard (b=c=1/3).

lust?
this is not a girl Phil!
:lol:

i got the same problem with the dvd2svcd script!

i did another sample with bicubic standard too:

Undot()
Asharp(1,4)
BicubicResize(448, 448, 1/3, 1/3, 0, 0, 720, 480)
Deen()
DCTFilter(1,1,1,1,1,1,0.5,0)
AddBorders(16, 16, 16, 16)

the result is ok,the same like bicubic precision!
any recomendations Phil :?:
:)

Lol. I'd edited my post (but too late ;-)).

No they are not. Bicubic precise is a lot sharper than normal. You can see that with the file length of a sample using same CQ for both : the one with bicubic precise will be bigger.

Note: Lanczos is often compared to a bicubic with b=0.75 or even b=1 !

too late...lol!

oh,ok,ok...but
differents things and problems:

using (any) bicubic the aspect radio is perfect,
using GripCrop the aspect radio is wrong with black borders,
with lanczos the aspect radio is perfect but give
:arrow: "strange artefacts in the edges"...
the edges seems "moving ants" with extreme bright.
:(

more recomendations and hints please Phil!

:)

That's exactly why I always use Bicubic Precise ;)

-kwag

Taken alone, that's true, Lanczos does too much sharpening.

But the problem is not the same with other filters. Light blur + Lanzos (my Divx script for instance) is in my eye a lot better than asharp + bicubic (MA script).

A resizer sharpen only the edges. Asharp sharpens everything, included possible remaining noise. That's a part of the difference.

In fact I have to say that I never use any sharpening filter ;-)

let me test your idea Phil.
<EDITED>
i did using:

Undot()
LanczosResize(448,448,0,0,720,480)
Deen()
DCTFilter(1,1,1,1,1,1,0.5,0)
AddBorders(16, 16, 16, 16)

and compare the quality with the others scripts:

with lanczos without asharp is really good too but
i see just a (little) few more details with bicubic/asharp...
seems more "dimentional" too...can't explain better.
deserve more tests cos i did with Peter Pan (Disney),
like all samples with scripts used in this tests.
:wink:

A classic cartoon is not the perfect choice for a sample. I do all mine with a short (5 minutes) making-off of Star Trek Insurrection : explosion, battle action, computer generated scenes... That's a really good test sample :-)

thanks Phil,
i will test now with "real" movies from dvds, not cartoons!

and like i wrote, the differences are tiny(really tiny),
Lanczos without "sharpens" is very good
:arrow: and give short final size too :!:

:wink: