hy friends!

script load in vdubmod from dvd2avi project:

mpeg2source("D:\dvd teste\a1.D2V")
converttoyuy2() #more contrast and vivid colors

give more contrat and vivid colors than:

mpeg2source("D:\dvd teste\a1.D2V")
#converttoyuy2() #using the "#"in this line, the image seems washed

ok,it's slow but the image is better if use any filter or MA script!

any comments?
thanks.

The image LOOKS better, but it is not better.

Let me compare with audio as you seem to know that better : a lot of cheap audio receivers have tons of DSP that can transform the sound and let it seems better to a newbie hear. But.. a "purist" will never use these DSP as they completly tranforms the original sound.

You know what I mean ?

Ok. So "ConvertToYUY2" has the same effect on your eyes : perhaps it looks better, but the original colorspace of the DVD is YV12.
And for me it is really important, especially for THX labelised sources, to not screw the work of the editor by doing such stupid convertion.

Note: okay, all this is just theory because everything screw up the colors, including TMPGENC (we saw that in comparing CCE/ffvfw and TMPGENC some time ago). I'm using currently the new Deen insteed of Convolution3D and the colors aren't the same at all

" ...DVD is YV12."
oh yes,i knew!

why vmesquita use this line in the end of the script in his guide if it :
"... screw the work of the editor by doing such...etc,etc,etc!



Phil,
i'm reading your edition in the last post while i'm writing this!


new Deen?
...found:
http://ziquash.chez.tiscali.fr/
http://www.avisynth.org/warpenterprises/

BEcause old (2.50) version of CCE couldn't work in other thing than YUY2. But it is not the case anymore with 2.66 or 2.67.
I do not have such line in my script and I don't have any problem.

Yes. A new one that has near 1 year old but never had time for it before . See there :

http://www.kvcd.net/forum/viewtopic.php?t=9532

These are both the old one.
New is there :
http://atlas2.tgv.net/~media-video/f...pic.php?t=2808
(see the french word "Telecharger" and click on it).

the link from Télécharger,give a new version from 13/08/03 and
the warplink and the marc homepage give the version from 19/01/03.

thanks!


"...is not the case anymore with 2.66 or 2.67."

right, now i see cos i have 250 and 266 and 266 really don't need!


did you read that cce 266/267 is better for p4 but
cce250 is better for athlon and give more quality

Jorel, according to colorspaces this here is a very useful site:
http://www.mir.com/DMG/chroma.html

Now to the "more contrast" in case of YUY2 output via Avisynth:

Different colorspaces at the end of an avs script sometimes will be treated by different system decoders! And that depends on how you configured your system, maybe XVID-Codec for YV12 and MS-YUY2-Codec for YUY2, ... or even FFdshow.
And heres the point, where on YUY2 output there could be done a postprocessing --- maybe activated in your case?

Do a test and open an avs script leaved at YV12 in Vdub and watch under File Information ... you will see which decompressor will be used - in my case the XVID codec.
Now apply a ConverttoYUY2() at the end of your script and open it again using Vdub ... you will see which decompressor will be used in this case.

There are so many options in decompressor engines like deblocking, Filmlook, etc. et. etc not mention the postprocessing capabilities if ffdshow as decompressor is used.

And here is the question (espec. to Phil).
IF you get into newer CCE using YV12 this would mean directshow is used to convert that YV12 to YUY2 so CCE can handle it internally?? Would be interesting as I only trust in Avisynths colorconversations ... so maybe a converttorgb24() in case of TmpgEnc or a ConverttoYUY2() in case of CCE could be a better way??

Or even if directshow is not used to pre-convert internally, how do we know TmpgEnc or CCE do end up with their type of internal colorconversation compared to avisynth colorconversation routines.

right incredible!

using converttoyuy2() in vdubmod file information:
YUV 4:2:2 (YUY2) ---> really better!

and without the line show:
ATI YUV12 format codec (16) --->washed!


.....i have an old ati all in wonder,
should i reinstall xvid to turn it better?

thanks for the link too!

As I "could" remember, phil reported some issues according to that ATI YV12codec. Well in my case everything ends up very well by using XVID for decompressing YV12.

http://roeder.goe.net/~koepi/xvid.shtml

Be shure you don't have already an older version deeply installed on your system if yes, deinstall it first as Koepi uses a new installer from now on.
And for shure, ... ged rid of that ATI YV12 codec ..... IF not needed by other things and that I don't know in your case of existing ATI hardware .


PS: What do you mean by saying "washed colors"??
Do you mean that cause of resulted artifacts out of an real! interlaced d2v input?
http://www.kvcd.net/forum/viewtopic.php?t=9380
Watch there the YV12 "interlaced" Chroma issue.

I noticed this too (I'm also using an All-In-Wonder with the ATI YV12 codec). I put ConvertToYUY2() at the bottom of all my scripts since I figured that one conversion from YV12 to YUY2 wouldn't be as bad as several RGB<->YUYV-type conversions...

Actually, I would have thought a conversion from YV12 to YUY2 would be lossless, since you should only have to encode the chroma values twice (instead of once). Does anyone have the algorithm for YV12<->YUY2 conversion? I've been reading the code from AviSynth for a half-hour or so, but my assembly language skills aren't what they used to be...

__________________
--
Chris "Bob" Odorjan

I think the solution is not "how aviynth upsamples the color frequency" an issue is already given by an added colorspace conversion which could be avoided. Every Colorspace conversion means a loss in chroma uspampling even YV12 to YUY2 as it can be read in the link I posted above.
Maybe your avs results good by previewing it in Vdub but in worsed case it will be even resampled 2! times more (like if using TmpgEnc where first rgb24 and finally YV12 mpeg).

So find out IF that ATI codec is really needed (I don't know as I do not have ATI hardware onboard), if not, trash it and use another one as it seems that this codec is responsable for wrong data processing.

You remember very well ! With ATI I had to use a "level" command in my script to lower the gamma to 0.8

For your other question, I wonder also if it will not be better to convert manually at the end just to be sure to control everything from a to z in the convert chain. It would avoid problems like thie ATI YV12 codec.

that is something to test

WIth ATI YV12 installed, "Washed" means ... use the level feature of ffdshow, boost the gamma to 1.4 or 1.6... you will see what washed means !

incredible,
"washed" means.....washed!
just kidd.
means that the image have less color and contrast..like old photos(not so old)!

now i'm waiting the verdict (write this way?) from you all.
what codec i have to use?
the vobs after the authoring have better color and image(luma)
and 30% less size(with cce-Q10) than the original vob in powerdvd.
i got the same with tmpgenc, few differences in images.
i don't did tests in the tvs, only in pc (monitor 15)!

Xvid !

thanks Phil!

i'm back with mine results:

scripts:

1-
mpeg2source("D:\dvd teste\a1.D2V")
#converttoyuy2()
in vdub
with
Koepi's xvid 290204
or
TI YUV12 format codec (16)
nothing is different between this 2, still "washed" !

2-
mpeg2source("D:\dvd teste\a1.D2V")
converttoyuy2()
in vdub
with
Koepi's xvid 290204
or
YUV 4:2:2 (YUY2)
nothing is different between this 2,
still really better with converttoyuy2()
i see real details !



please,can someone test with and without converttoyuy2() in the end of the script?
thanks!

I did a lot ! And your problem is to find why your colors are washed in YV12, not why YUY2 is better ! In a normal situation colors in YV12 should be the same than in YUY2 !

Did you at least remove the ATI YV12 ? Did you check that this time this is Xvid that is used ?

I tried it. I have XviD CODEC installed.
Can't see any difference with or without ConvertToYUY2() at the end of the script.

-kwag

ok Kwag.....now choose:
me or my pc is mad?
maybe both!

more:
my friend Edson is here with me and see the pictures.
he see that with converttoyuy2() is really better easily!




edited>
if someone want to see,i can send the pictures with 1mb size each by mail!

Hi jorel,

It must be a CODEC issue
That's why you see it better with the line inserted

-kwag