alt curve vs. linear curve

[kronik]

hey all,
which of these has produced better results for u,
linear curve (0% and 0%) or the alternative curve?
what settings do u use for alternative curve?
thanks for u'r input
u'r time is greatly appreciated :D

KroniK

[Teegedeck]

I think most of us agree that linear cc gives the best results. (I've been a fan of that concept for quite some time, setting CC=0 back in my Nandub days. ;) And it seemed to work well in GKnot, too, though that's a little different.)

[iago]

I've been a fan of that concept for quite some time, setting CC=0 back in my Nandub days ;).

@Teegedeck,

Same here my friend; Linear CC, absolutely! ;)

regards,
iago

[AllTimeSToneD]

I go with Linear CC today :cool: like i did back in the days with Nandub and GKnot :D

[Gannjunior]

Linear CC, absolutely
Could I know your quality measurement method?Have you done any test about altc cc vs linear cc?Do you use some objective instruments to help your eyes? For example Xvid quality,PSNR or direct comparison between screenshots of DVD vs Xvid?Have you done any comparison in the same DVD film between DVD vs Xvid+alt cc and DVD vs Xvid+linear,to say the second choice is better than first choice?
'Cause it could be very interesting an objective comparison between the two methods to be sure which is the best.

Ciao:)

[Shayne]

Originally posted by Gannjunior
'Cause it could be very interesting an objective comparison between the two methods to be sure which is the best.


Sounds great when can we expect your results

[MfA]

When your eyes disagree with the objective quality measure the objective quality measure is wrong :)

[sam_b]

Linear Scaling. Every time.

Oh, unless it's a really high-action movie, in which case I can get a big improvement in the odd static scene for very little impact on the numerous high-motion scenes.

Oh, out of sheer morbid fascination, does anyone know what CC divx5 uses?

[SirDavidGuy]

Originally posted by sam_b
Linear Scaling. Every time.

Oh, unless it's a really high-action movie, in which case I can get a big improvement in the odd static scene for very little impact on the numerous high-motion scenes.

Oh, out of sheer morbid fascination, does anyone know what CC divx5 uses?

Morbid fascination indeed... Dual-Pass Leaky bucket... A technique so evil I hardly dare to tpe it's name...

[Snakeisthestuff]

since my first settings for xvid which i had become from Teegedeck ;o), i used Xvid at nearly all encodes i done since then, and always used linear-CC, and it works fine.

[NoLogo]

As far as I'm concerned, always AltCC, never been deceived with the results. I've tried linear scaling once, but, compared to AltCC, for the considered movie, the quality was lower (to my eyes, and with XviD analyzer).

[kronik]

hmm 1 for alt curve eh?! well...i'm going to use the exact same settings but with the linear curve and maybe i'll post some results :)
advice is nice but tests are the best ;)
thanks guys
KroniK

[Sirber]

Hi

The default settings for Alt. CC gives good results for most movies. For the matrix the picture was less blocky.

500/90/30 ==> good
175/80/30 ==> more blocks

[Shayne]

@ NoLogo

So can you please tell me at what precent above high distance you are steeling bits from. If only 0 percent are above 500 this is not too many frames and visually i do not know how you catch these.

[serbersan]

Well My results about linear scalling had been better than alt-cc.

I have obtained, less blocks in action scenes rather than using alt-cc, like koepi said, I've obtained more constant quality using linear scaling.

[Didée]

Scaling curves once again, yahoo!

I have obtained, less blocks in action scenes rather than using alt-cc
I'm not exactly surprised: That's the expected behaviour.

Alt-CC: Steal from the rich (HiMo), give to the poor (LoMo)
Linear-CC (0,0): Don't steal at all, but distribute equally


@ Shayne

In above example,
500 is not the point at which stealing begins.
500 is the point where stealing reaches its biggest relative amount.


Off Topic

That reminds me, I still have to finish playing Thief II.
Let's go hiding and stealing ...
;)

[kronik]

my tests showed equal quality during low motion but in higher motion i found more blocks in alt curve with a 300-100-30 setting. i also compared with 2 different settings with sbc and found xvid to have a bit more detail within blocks but seems to spread "mosquito noise" to the edges of the blocks where sbc stopped very close to the edges of the objects.
looks like i'll be getting to know some filters :D

[ixpi2001]

hy,
has anyone tried to set the alt curve settings to the values a first pass delivered? described in: http://www.animemusicvideos.org/guides/avtech/xvid.html

[Angrychair]

Alt curve for me, but I only encode anime, so my results might be a little different than what others get with live action films.

Alt curve set to medium or high gives much better look to the low motion scenes than linear, and the high motion scenes show little difference. Chroma ME, of course, is also a must. :)

[Sigmatador]

on anime: alt curve with low agression and max high & low distance calculated from de stats file.

on movie: alt curve with normal agression is better with fast motion than linear curve, which is a bit better with low motion scene.

for a movie i think it's a choice, but the both are good. for anime it's altcc every time :D .

ps: i confirm, altcc on matrix is beautiful ;)

[iago]

linear cc is good! Use it! :D

[Sigmatador]

2pass int ou 2pass ext ? :p :D