01-12-2003, 12:14 AM
|
Free Member
|
|
Join Date: Jul 2002
Posts: 1,224
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
@Kwag,
Here's an interesting link from Doom9 where SansGrip started a
post about Psychovisual:
http://forum.doom9.org/showthread.ph...t=psychovisual
-black prince
|
Someday, 12:01 PM
|
|
Site Staff / Ad Manager
|
|
Join Date: Dec 2002
Posts: 42
Thanks: ∞
Thanked 42 Times in 42 Posts
|
|
|
01-12-2003, 01:34 AM
|
Free Member
|
|
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Now we have to double check that the file prediction is still consistent with this GOP. I'm currently encoding "K-19" with GripFit ( ) at 528x480 with 1-24-4-1-24 + BETA-1a matrix, so I'll post tomorrow if the predicted/actual size are on target.
-kwag
|
01-12-2003, 01:44 AM
|
Invalid Email / Banned / Spammer
|
|
Join Date: Aug 2002
Location: Brasil - MG - third stone from the sun
Posts: 5,570
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Kwag,
and for (320 wrong)352x240 lbr?
about matrix,gop,cq or cq_vbr,high quality?
if my player plays svcds,than should play skvcds?
i never do skvcd!
thanks in advance!
editing:
is 352x240!
|
01-12-2003, 01:53 AM
|
Free Member
|
|
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by jorel
Kwag,
and for 320x240 lbr?
about matrix,gop,cq or cq_vbr,high quality?
|
You mean 352x240 right
The above GOP should apply too. Need more testing. Try it out Quote:
if my player plays svcds,than should play skvcds?
i never do skvcd!
thanks in advance!
|
It should play SKVCD too. The only difference from SKVCD to SVCD is the resolution, where SVCDs are 480x480 and SKVCD is 352x480 ( Half D-1 resolution, or CVD resolution )
-kwag
|
01-12-2003, 01:55 AM
|
Free Member
|
|
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
I'm currently encoding "K-19" with GripFit( ) at 528x480 and the test GOP 1-24-4-1-24 + BETA-1 "Notch" Matrix. This is what the movie will look like on one CD-R: http://www.kvcd.net/k-19-small-samp.mpg
-kwag
|
01-12-2003, 01:58 AM
|
Invalid Email / Banned / Spammer
|
|
Join Date: Aug 2002
Location: Brasil - MG - third stone from the sun
Posts: 5,570
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Kwag wrote:
"You mean 352x240 right "
yes,of course...... :tongue2:
thanks,i try!
|
01-12-2003, 02:02 AM
|
Free Member
|
|
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Hey, don't worry jorel, I do that all the time too
|
01-12-2003, 02:48 AM
|
Free Member
|
|
Join Date: Jun 2002
Posts: 290
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
From Kwag "I'm currently encoding "K-19" with GripFit( ) at 528x480 and the test GOP 1-24-4-1-24 + BETA-1 "Notch" Matrix. This is what the movie will look like on one CD-R: http://www.kvcd.net/k-19-small-samp.mpg"
do you mean BETA-1 or BETA-1a "Notch"?
|
01-12-2003, 02:58 AM
|
Invalid Email / Banned / Spammer
|
|
Join Date: Aug 2002
Location: Brasil - MG - third stone from the sun
Posts: 5,570
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by kwag
I'm currently encoding "K-19" with GripFit( ) at 528x480 and the test GOP 1-24-4-1-24 + BETA-1 "Notch" Matrix. This is what the movie will look like on one CD-R: http://www.kvcd.net/k-19-small-samp.mpg
-kwag
|
very good....clear!
"do you mean BETA-1 or BETA-1a "Notch"?"
yes, i ask like syk2c11.
|
01-12-2003, 09:02 AM
|
Free Member
|
|
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by jorel
"do you mean BETA-1 or BETA-1a "Notch"?"
yes, i ask like syk2c11.
|
Sorry, it's Beta-1a "Notch" Matrix.
Movie encoded. Final file size was ~4% lower because my sample size was also ~4% lower
So ratio is correct and prediction is valid
-kwag
|
01-12-2003, 11:41 AM
|
Free Member
|
|
Join Date: Nov 2002
Location: Ontario, Canada
Posts: 1,135
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Well, I just finished running samples from Death To Smoochy (DVD 16:9 1h49m). Here's what I did:
CQ original GOP original matrix
CQ new GOP original matrix
CQ new GOP beta-1a matrix
CQ experimental GOP beta-1a matrix
and the same combinations of GOP and matrix for CQ_VBR mode.
Result:
The experimental matrix caused a huge drop in CQ level, from 48 for new GOP beta-1a to 39. To be honest it looked like I had dropped the CQ that much, with significantly more Gibbs and blockiness. The artifacts at CQ 48 are barely acceptable (I'm going to do a two-disc encode as well, just in case), but at CQ 39 they're definitely not acceptable for a one-disc encode.
The experimental matrix increases CQ_VBR compression greatly, causing a jump in CQ_VBR level from 7 for new GOP beta-1a to 11. All CQ_VBR encodes looked far worse than the CQ versions.
Overall I prefer the new GOP beta-1a sample. It seems to be about the right balance of compression and artifacts, blocks and Gibbs.
|
01-12-2003, 11:50 AM
|
Free Member
|
|
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by SansGrip
Overall I prefer the new GOP beta-1a sample. It seems to be about the right balance of compression and artifacts, blocks and Gibbs.
|
Hi SansGrip,
You mean the 1-12-1-1-24 or 1-24-4-1-24
-kwag
|
01-12-2003, 11:55 AM
|
Free Member
|
|
Join Date: Nov 2002
Location: Ontario, Canada
Posts: 1,135
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by kwag
You mean the 1-12-1-1-24 or 1-24-4-1-24
|
Sorry... By "experimental" I meant 1-24-4-1-24. By "new" I meant 1-12-1-1-24. I should've been more clear .
|
01-12-2003, 12:08 PM
|
Free Member
|
|
Join Date: Jul 2002
Posts: 1,224
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
@Kwag,
Just finished " Signs" using GOP 1-24-4-1-24, Q-Matrix Notch Beta-1a,
and CQ=100. This was for a 2 CD encode + audio. The results were
GREAT!!! Flashing was very subtle this time and from 2+ feet
from TV not noticable at all. There was Gibbs, but no worst than using
CQ_VBR. File size was smaller than before (e.g. 1.4GB vs 1.5GB) with
old GOP and Q-Matrix. The picture quality was Excellent!!!
Gibbs was slightly more noticable than before, but viewing 2+ feet
away from screen it was hardly noticable unless you were looking
for it. I'm encoding another full video with the new GOP and Q-Matrix
to make sure this was not just a fluke. I'll let you know how it
turns out
-black prince
|
01-12-2003, 12:17 PM
|
Free Member
|
|
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by black prince
Gibbs was slightly more noticable than before, but viewing 2+ feet
away from screen it was hardly noticable unless you were looking
for it.
|
Yes , I just made a couple of screen shot comparisons with Vdub, and I can see the Gibbs. So back to the drawing board. I won't be satisfied until I find the correct GOP that doesn't "flash" (like 1-24-4-1-24) and no more Gibbs than the reference sample (like 1-12-1-1-24) or any reference sample made with 1-12-1-1-24.
So back to more tests here .
-kwag
|
01-12-2003, 12:21 PM
|
Free Member
|
|
Join Date: Sep 2002
Location: Lahti, Finland
Posts: 1,652
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Would experimenting with MPEG-2 do any good as SansGrip said the I-frames would be less blocky? I know it produces bigger files than MPEG-1 but as we've all seen here, it's only a matter of time before the filesize gets shrunk down
|
01-12-2003, 12:25 PM
|
Invalid Email / Banned / Spammer
|
|
Join Date: Aug 2002
Location: Brasil - MG - third stone from the sun
Posts: 5,570
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
i lost something?
now we are testing
the BETA-1a notch :
8 9 12
9 10 14
12 14 18
or not?
correct it please!
|
01-12-2003, 12:28 PM
|
Free Member
|
|
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by Boulder
Would experimenting with MPEG-2 do any good as SansGrip said the I-frames would be less blocky? I know it produces bigger files than MPEG-1 but as we've all seen here, it's only a matter of time before the filesize gets shrunk down
|
Yes, but the primary focus right now is MPEG-1, so we can target to most standalone DVD players. I assume that for MPEG-2, we're going to wind up with another GOP . As far as the matrix, it seems to do a good job as it is with MPEG-2
But as soon as we get the MPEG-1 stabilized, which I hope is soon , then we can tackle the MPEG-2 optimizations
-kwag
|
01-12-2003, 12:30 PM
|
Free Member
|
|
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by jorel
i lost something?
now we are testing
the BETA-1a notch :
8 9 12
9 10 14
12 14 18
or not?
correct it please!
|
Yes, and I think that's the best and final matrix ( for a while ).
Really, I hope we freeze that matrix now, and that will be the new KVCD v2 Q. Matrix. Hopefully it will stay that way for a long time.
-kwag
|
01-12-2003, 12:33 PM
|
Invalid Email / Banned / Spammer
|
|
Join Date: Aug 2002
Location: Brasil - MG - third stone from the sun
Posts: 5,570
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
really good Kwag, i do now some more tests with 352x240 lbr!
|
All times are GMT -5. The time now is 01:06 AM — vBulletin © Jelsoft Enterprises Ltd
|