AnimeSuki Forums

Noct · 2010-09-06, 23:16

Did you by any chance Copy&Paste your own code into the avsi?
You have to use the code quoted in the first post.

If you did that i have no idea what's wrong.
Maybe you have 2 different versions of avisynth installed?

Edit:
Didn't see this before, but try deleting the " in the insertsign lines.

Kristen · 2010-09-06, 23:21

DirectShowSource("C:\Users\Aimbotter33\Desktop\[FFFpeeps]_Kaichou_wa_Maid-sama_23_[720p][B40216DC].mkv")
sign1 = ("C:\Users\Aimbotter33\Desktop\[FFFpeeps]_Kaichou_wa_Maid-sama_23_[720p][B40216DC].mkv")
logo = imagesource("C:\Users\Aimbotter33\Desktop\images.p ng", pixel_type="RGB32", end=100)

insertsign=("last, logo, 0")
insertsign=("last, sign1, 563, 900")

The bolded "'s are probably what's causing the problem. Try removing them.

In addition, you never specified what kind of a source ("C:\Users\Aimbotter33\Desktop\[FFFpeeps]_Kaichou_wa_Maid-sama_23_[720p][B40216DC].mkv") was for sign1. Even with that, I'm not sure if you can overlay it with insertsign unless it's an AVI.

Aimbotter33 · 2010-09-06, 23:33

wheres the bolded, im a failure >.<

jfs · 2010-09-07, 03:32

If you're going to be use Avisynth I suggest you learn at least the basics, such as understanding function usage, quoted string literals and parameters to functions.

This line:
insertsign=("last, logo, 0")

Two primary problems.

The first is that you're putting an equals sign after the name of a function you're trying to use.
What Avisynth will read here is: There's an equals sign, so the operation is assignment; the left-hand side of the equals sign is "insertsign" so that means we will be storing something into that name; the right-hand side is a parenthesized string literal, so store that into the variable.
What you meant to do: Call the "insertsign" function with three parameters. Function calling is done by putting the name of the function directly before a parenthesis pair containing the parameters to pass to the function.
First thing to learn: No equals sign when you're calling functions. So remove the equals sign.

The new line is:
insertsign("last, logo, 0")

This line still has a problems: The InsertSign function wants 3 parameters, but you're passing it just one, a single quoted string literal.
When you put something inside a pair of " " double quotes, it's a string literal, which is a single value.
What you mean to do is pass the function three values, none of which must be string literals.
Remove those quotes and the meaning changes a lot:
insertsign(last, logo, 0)
Now you're calling the InsertSign function with three parameters. The first parameter is last, which is a special variable that is the result of the last function call that isn't assigned to something else, in other words you're inserting the sign into the last clip produced. Also, the last variable is reassigned from using the InsertSign function here, because there's no assignment of the result otherwise.
The second parameter is logo which is a clip produced by the earlier ImageSource function call. The last parameter is the number zero, which in this case is the frame number to insert the sign at. (That's what the InsertSign function is defined to do.)

Aimbotter33 · 2010-09-07, 04:26

Well when i put this insertsign(last, logo, 0) inside it will say Invaid command

jfs · 2010-09-07, 10:59

Most likely, you didn't follow instructions and never created the .avsi file with the InsertSign function, or you placed it in the wrong folder, or you gave it a wrong name.

Aimbotter33 · 2010-09-07, 11:10

Well I have Worked It out And it success From another code

gorgonite · 2010-11-08, 07:54

forgive my shitty engrish
can the fgo patch posted in this thread : http://forum.doom9.org/showthread.php?t=137117 be helpful to make this look like non shit ? (I want to KEEP that original grain)
even crf=3 don't keep the grain with --psy-rd 0.60:0.10

this is what I got with: (best result that I managed to get)

Quote:

--crf=17 --deblock 1:0:-2 --subme 10 --partitions all --trellis 2 --psy-rd 0.65:0.10 --aq-strength 1.2

also, is there a way to reduce that color change effect ? I played with --colormatrix and --colorprim and I got no change
colormatrix() was of no help either. GrainFactory3 helped a little improving it at the cost of huge extra bitrate.

--
can someone do me a favour and upload the fgo build coz I just can't get it from that page :§

mastrboy · 2010-11-08, 08:42

Quote:

Originally Posted by gorgonite

(I want to KEEP that original grain no matter how much size it'll take)
even crf=3 don't keep the grain with --psy-rd 0.60:0.10

You could try the following x264 options:
--deblock -2:-2 --aq-strength 1.2 --psy-rd 1.10:0.5 --no-dct-decimate

There is also the option qcomp=1.0 though i rarely adjust it over 0.7

But if you are denoising before x264 you might be creating that banding you mention, which can somewhat be resolved with Gradfun2dbmod: http://forum.doom9.org/showthread.php?t=144537

gorgonite · 2010-11-08, 17:38

Quote:

Originally Posted by mastrboy

You could try the following x264 options:
--deblock -2:-2 --aq-strength 1.2 --psy-rd 1.10:0.5 --no-dct-decimate

There is also the option qcomp=1.0 though i rarely adjust it over 0.7

But if you are denoising before x264 you might be creating that banding you mention, which can somewhat be resolved with Gradfun2dbmod: http://forum.doom9.org/showthread.php?t=144537

No, I'm not denoising; if it was the case it would be very easy. I want to make some grain rentition, instead.
I was able to "somehow" enhance the x264 output by:

Quote:

--preset slow --tune animation --crf 16.0 --deblock 1:-3:-2 --scenecut 50 --ref 8 --bframes 5 --qpmin 15 --qcomp 0.5 --rc-lookahead 80 --aq-strength 0.8 --merange 24 --subme 10 --partitions all --trellis 2 --no-dct-decimate --no-fast-pskip --fps 24000/1001

and I got this.
It seems that alpha debloking is the one that caused the "unusual color change". But, still, I'm not yet happy with the results so I, at the same settings, tried lowering psy_rd to 0.5, raising it to 0.7, raising AQ-strenght to 1.0 and lowering it to 0.6 but thing get from bad to worse every time (grain rentetionnaly speaking). Also, raising psy-trellis was of no help too. As for the avisynth script it contains the only AVCSource line with the trim.

I don't really get what I'm messing here. Usually I can do better with the grainy TS sources which are really shitty compared to this Blu-ray.
Any suggestions?

Desbreko · 2010-11-08, 19:10

That patch is really old. I think it's been replaced by the current psy-rd.

The grain and overall amount of detail in BDs tends to be finer than in broadcast sources, and thus harder to retain without using a really high bitrate. This is why you can use crf 18-20 on most .ts's and get perceptual transparency, while on BDs you might need to go down to 15-17 or so to maintain transparency.

So first, don't use tune animation. This tuning is meant mainly for animation with only flat color and low detail textures. If you use it on grainy animation, it will blur the crap out of the grain because it sets aq-strength and psy-rd really low. You'll probably want aq-strength somewhere around 1.0 to 1.2, and psy-rd around 0.8:0.1 to 1.0:0.2.

Second, don't lower qcomp. This strengthens the effect of mbtree, which will hurt grain retention. Sometimes it can help to raise it a bit, like maybe 0.65 or 0.7, but don't go too high or it will become less efficient than simply lowering the crf value.

Third, don't raise qpmin. Doing so cripples the quality of the I and P frames, which tend to get encoded with a lower qp (i.e. higher quality) because other frames will be referencing from them a lot. If you really want to keep your file size from getting too huge at the cost of some quality outside of the grain, it's better to lower ipratio and pbratio.

And with a bunch of the other options you've got set, you might as well just use preset veryslow and not clutter the command line so much.

TheFluff · 2010-11-09, 04:20

FGO is still relevant, even for anime. It's even advantageous to use a low level of FGO (~5) on non-grainy anime sources that have been gradfunkmirror()'d, it helps a lot with preserving gradients. It can be used together with low (<0.5) psyrd without any ill effects, but I haven't tested it together with psytrellis or higher psyrd myself.

Desbreko · 2010-11-09, 08:35

Hm, guess I'll have to try it next time I encode something grainy, then.

And it looks like JEEB's x264 builds have the FGO patch included, so you could grab one from there to try it, gorgonite.

gorgonite · 2010-11-12, 18:50

Thank you Desbreko,
That worked like a charm. I even got many settings choices

--psy-rd 1:0.15 gave me a pretty much better results than fgo 14

I got now a new issue with the motion estimation, I think.
here is a sample of it:http://www.mediafire.com/?w6eti4maq9qvgjq

I tweaked ipratio , bpratio, scencut and rc-lookahead without getting any improvement.

Desbreko · 2010-11-12, 20:34

Is that sample from your encoded version or the source? I don't get any ghosting when I play it in MPC-HC or view it in AvsP with ffms2, and re-encoding it with --crf 16 --preset veryslow --deblock -2:-2 --psy-rd 1.0:0.15 also came out fine.

gorgonite · 2010-11-13, 17:37

It's an encoded testing sample, and that image is from mpc-hc. I don't get that ghosting with AvsP. Furthermore, the ghosting exist even when I play the raw on mpc-hc.
It seems that it's a decoder problem? I uinstalled CCCP and reinstalled it again and nothing changed.
I don't have a "1080p resolution capable screen", instead I have a 15.6" LCD screen 1366x786.

Aside from that, now I want to convert the AC3 5.1 TrueHD sound track to FLAC 5.1 with eac3to. so I tried these commands (I don't have any of the commercial decoders/encoders):

eac3to "D:\the movie.ac3" "D:\the movie.flac"
eac3to "D:\the movie.m2ts" 2: "D:\the movie.flac"
eac3to "D:\the movie.ac3" "D:\the movie.flac" -blu-ray

and I also tried a bunch of other options that I found on the eac3to guide.
And everytime I don't hear the sound (I can hear a very very low noisy one if I set the volume to maximum and every time, the flac file get just 808 mb!)

However, when I convert it to AC3 640 it works fine.

here are the sound track infos:

Spoiler for info:

Since it's my first Blu-ray experience I don't even know what the heck is that : "Duration : 6h 20mn"
here is a one-minute-11MB sample from the original audio : http://www.mediafire.com/?49ya9p13ynrw3l4

Desbreko · 2010-11-13, 21:06

Quote:

Originally Posted by gorgonite

It's an encoded testing sample, and that image is from mpc-hc. I don't get that ghosting with AvsP. Furthermore, the ghosting exist even when I play the raw on mpc-hc.
It seems that it's a decoder problem? I uinstalled CCCP and reinstalled it again and nothing changed.
I don't have a "1080p resolution capable screen", instead I have a 15.6" LCD screen 1366x786.

Did you reboot before reinstalling the CCCP? Also, make sure you don't have any post-processing filters enabled in ffdshow or MPC-HC.

Quote:

Aside from that, now I want to convert the AC3 5.1 TrueHD sound track to FLAC 5.1 with eac3to.

This seemed to work fine for me:
eac3to "One minute sample PID 1100 3_2ch 48KHz 640Kbps DELAY 0ms.ac3" "test.flac"

Daiz · 2010-11-13, 22:12

I think you need a more capable TrueHD decoder to extract the full lossless track out of it, not just the 640kbps core. In case you don't have a decoder like that, you shouldn't re-encode the 640kbps core AC3 to FLAC, as that's just converting lossy to lossless and thus a waste of space.

Desbreko · 2010-11-14, 00:05

I thought libavcodec could decode TrueHD. Or does it just use the AC3 core?

sneaker · 2010-11-14, 00:21

Quote:

Originally Posted by Desbreko

I thought libavcodec could decode TrueHD.

You are correct, it can and eac3to can use it.

2010-09-06, 23:21	Link #502
Kristen Senior Member Author Join Date: Jul 2007 Location: Virginia Tech	DirectShowSource("C:\Users\Aimbotter33\Desktop\[FFFpeeps]_Kaichou_wa_Maid-sama_23_[720p][B40216DC].mkv") sign1 = ("C:\Users\Aimbotter33\Desktop\[FFFpeeps]_Kaichou_wa_Maid-sama_23_[720p][B40216DC].mkv") logo = imagesource("C:\Users\Aimbotter33\Desktop\images.p ng", pixel_type="RGB32", end=100) insertsign=("last, logo, 0") insertsign=("last, sign1, 563, 900") The bolded "'s are probably what's causing the problem. Try removing them. In addition, you never specified what kind of a source ("C:\Users\Aimbotter33\Desktop\[FFFpeeps]_Kaichou_wa_Maid-sama_23_[720p][B40216DC].mkv") was for sign1. Even with that, I'm not sure if you can overlay it with insertsign unless it's an AVI. __________________

2010-09-07, 03:32	Link #504
jfs Aegisub dev Join Date: Sep 2004 Location: Stockholm, Sweden Age: 39	If you're going to be use Avisynth I suggest you learn at least the basics, such as understanding function usage, quoted string literals and parameters to functions. This line: insertsign=("last, logo, 0") Two primary problems. The first is that you're putting an equals sign after the name of a function you're trying to use. What Avisynth will read here is: There's an equals sign, so the operation is assignment; the left-hand side of the equals sign is "insertsign" so that means we will be storing something into that name; the right-hand side is a parenthesized string literal, so store that into the variable. What you meant to do: Call the "insertsign" function with three parameters. Function calling is done by putting the name of the function directly before a parenthesis pair containing the parameters to pass to the function. First thing to learn: No equals sign when you're calling functions. So remove the equals sign. The new line is: insertsign("last, logo, 0") This line still has a problems: The InsertSign function wants 3 parameters, but you're passing it just one, a single quoted string literal. When you put something inside a pair of " " double quotes, it's a string literal, which is a single value. What you mean to do is pass the function three values, none of which must be string literals. Remove those quotes and the meaning changes a lot: insertsign(last, logo, 0) Now you're calling the InsertSign function with three parameters. The first parameter is last, which is a special variable that is the result of the last function call that isn't assigned to something else, in other words you're inserting the sign into the last clip produced. Also, the last variable is reassigned from using the InsertSign function here, because there's no assignment of the result otherwise. The second parameter is logo which is a clip produced by the earlier ImageSource function call. The last parameter is the number zero, which in this case is the frame number to insert the sign at. (That's what the InsertSign function is defined to do.) __________________ Aegisub developer [ Forum \| Manual \| Feature requests \| Bug reports \| IRC ] Don't ask for: More VSFilter changes (I won't), karaoke effects, help in PM's

2010-09-07, 10:59	Link #506
jfs Aegisub dev Join Date: Sep 2004 Location: Stockholm, Sweden Age: 39	Most likely, you didn't follow instructions and never created the .avsi file with the InsertSign function, or you placed it in the wrong folder, or you gave it a wrong name. __________________ Aegisub developer [ Forum \| Manual \| Feature requests \| Bug reports \| IRC ] Don't ask for: More VSFilter changes (I won't), karaoke effects, help in PM's

2010-11-09, 04:20	Link #512
TheFluff Excessively jovial fellow Join Date: Dec 2005 Location: ISDB-T Age: 37	FGO is still relevant, even for anime. It's even advantageous to use a low level of FGO (~5) on non-grainy anime sources that have been gradfunkmirror()'d, it helps a lot with preserving gradients. It can be used together with low (<0.5) psyrd without any ill effects, but I haven't tested it together with psytrellis or higher psyrd myself. __________________ \| ffmpegsource 17:43:13 <~deculture> Also, TheFluff, you are so fucking slowpoke.jpg that people think we dropped the DVD's. 17:43:16 <~deculture> nice job, fag! 01:04:41 < Plorkyeran> it was annoying to typeset so it should be annoying to read

2010-11-13, 17:37	Link #516
gorgonite C Join Date: Oct 2010	It's an encoded testing sample, and that image is from mpc-hc. I don't get that ghosting with AvsP. Furthermore, the ghosting exist even when I play the raw on mpc-hc. It seems that it's a decoder problem? I uinstalled CCCP and reinstalled it again and nothing changed. I don't have a "1080p resolution capable screen", instead I have a 15.6" LCD screen 1366x786. Aside from that, now I want to convert the AC3 5.1 TrueHD sound track to FLAC 5.1 with eac3to. so I tried these commands (I don't have any of the commercial decoders/encoders): eac3to "D:\the movie.ac3" "D:\the movie.flac" eac3to "D:\the movie.m2ts" 2: "D:\the movie.flac" eac3to "D:\the movie.ac3" "D:\the movie.flac" -blu-ray and I also tried a bunch of other options that I found on the eac3to guide. And everytime I don't hear the sound (I can hear a very very low noisy one if I set the volume to maximum and every time, the flac file get just 808 mb!) However, when I convert it to AC3 640 it works fine. here are the sound track infos: Spoiler for info: Format : AC-3 Format/Info : Audio Coding 3 Format profile : TrueHD / Core File size : 1.70 GiB Duration : 6h 20mn Overall bit rate : 640 Kbps Audio Format : AC-3 Format/Info : Audio Coding 3 Format profile : TrueHD / Core Mode extension : CM (complete main) Muxing mode : After core data Duration : 6h 20mn Bit rate mode : Variable / Constant Bit rate : 640 Kbps Maximum bit rate : 3 333 Kbps Channel(s) : 6 channels Channel positions : Front: L C R, Side: L R, LFE Sampling rate : 48.0 KHz Bit depth : 16 bits Stream size : 1.70 GiB (100%) Since it's my first Blu-ray experience I don't even know what the heck is that : "Duration : 6h 20mn" here is a one-minute-11MB sample from the original audio : http://www.mediafire.com/?49ya9p13ynrw3l4 Last edited by gorgonite; 2010-11-13 at 17:40. Reason: adding sample

2010-09-06, 23:16	Link #501
Noct Junior Member Join Date: Mar 2009	Did you by any chance Copy&Paste your own code into the avsi? You have to use the code quoted in the first post. If you did that i have no idea what's wrong. Maybe you have 2 different versions of avisynth installed? Edit: Didn't see this before, but try deleting the " in the insertsign lines.

2010-09-06, 23:33	Link #503
Aimbotter33 Junior Member Join Date: Sep 2010	wheres the bolded, im a failure >.<

2010-09-07, 04:26	Link #505
Aimbotter33 Junior Member Join Date: Sep 2010	Well when i put this insertsign(last, logo, 0) inside it will say Invaid command

2010-09-07, 11:10	Link #507
Aimbotter33 Junior Member Join Date: Sep 2010	Well I have Worked It out And it success From another code

2010-11-08, 19:10	Link #511
Desbreko Senior Member Fansubber Join Date: Apr 2009 Age: 36	That patch is really old. I think it's been replaced by the current psy-rd. The grain and overall amount of detail in BDs tends to be finer than in broadcast sources, and thus harder to retain without using a really high bitrate. This is why you can use crf 18-20 on most .ts's and get perceptual transparency, while on BDs you might need to go down to 15-17 or so to maintain transparency. So first, don't use tune animation. This tuning is meant mainly for animation with only flat color and low detail textures. If you use it on grainy animation, it will blur the crap out of the grain because it sets aq-strength and psy-rd really low. You'll probably want aq-strength somewhere around 1.0 to 1.2, and psy-rd around 0.8:0.1 to 1.0:0.2. Second, don't lower qcomp. This strengthens the effect of mbtree, which will hurt grain retention. Sometimes it can help to raise it a bit, like maybe 0.65 or 0.7, but don't go too high or it will become less efficient than simply lowering the crf value. Third, don't raise qpmin. Doing so cripples the quality of the I and P frames, which tend to get encoded with a lower qp (i.e. higher quality) because other frames will be referencing from them a lot. If you really want to keep your file size from getting too huge at the cost of some quality outside of the grain, it's better to lower ipratio and pbratio. And with a bunch of the other options you've got set, you might as well just use preset veryslow and not clutter the command line so much.

2010-11-09, 08:35	Link #513
Desbreko Senior Member Fansubber Join Date: Apr 2009 Age: 36	Hm, guess I'll have to try it next time I encode something grainy, then. And it looks like JEEB's x264 builds have the FGO patch included, so you could grab one from there to try it, gorgonite.

2010-11-12, 18:50	Link #514
gorgonite C Join Date: Oct 2010	Thank you Desbreko, That worked like a charm. I even got many settings choices --psy-rd 1:0.15 gave me a pretty much better results than fgo 14 I got now a new issue with the motion estimation, I think. here is a sample of it:http://www.mediafire.com/?w6eti4maq9qvgjq I tweaked ipratio , bpratio, scencut and rc-lookahead without getting any improvement.

2010-11-12, 20:34	Link #515
Desbreko Senior Member Fansubber Join Date: Apr 2009 Age: 36	Is that sample from your encoded version or the source? I don't get any ghosting when I play it in MPC-HC or view it in AvsP with ffms2, and re-encoding it with --crf 16 --preset veryslow --deblock -2:-2 --psy-rd 1.0:0.15 also came out fine.

2010-11-13, 22:12	Link #518
Daiz Pioneer in Fansub 2.0 Join Date: Aug 2007	I think you need a more capable TrueHD decoder to extract the full lossless track out of it, not just the 640kbps core. In case you don't have a decoder like that, you shouldn't re-encode the 640kbps core AC3 to FLAC, as that's just converting lossy to lossless and thus a waste of space. __________________ "A good user is remembered from his posts, not from his 160px tall animated 'pink flying unicorns' signature picture." --- The Guide for best H.264 playback

2010-11-14, 00:05	Link #519
Desbreko Senior Member Fansubber Join Date: Apr 2009 Age: 36	I thought libavcodec could decode TrueHD. Or does it just use the AC3 core?