AviSynth+ thread Vol.2

[qyot27]

Presumably, the check exists the way it does because you would be allocating PLANAR_Y and PLANAR_G at the beginning of the sequence, since they're both the first plane in the the sequence order. Planar RGB is ordered GBR, as opposed to endian-specific packed RGB (which is either BGR in little or RGB in big).

For comparison, see how FFmpeg initiates the format in the AviSynth demuxer,
http://git.videolan.org/?p=ffmpeg.gi...40abeba6f#l127

 127 static const int avs_planes_rgb[3]    = { AVS_PLANAR_G, AVS_PLANAR_B,
 128                                           AVS_PLANAR_R };
 129 static const int avs_planes_yuva[4]   = { AVS_PLANAR_Y, AVS_PLANAR_U,
 130                                           AVS_PLANAR_V, AVS_PLANAR_A };
 131 static const int avs_planes_rgba[4]   = { AVS_PLANAR_G, AVS_PLANAR_B,
 132                                           AVS_PLANAR_R, AVS_PLANAR_A };

[ENunn]

Hey there. I don't know if this is the place for bug reports or not but I found a bug. Whenever I use the normalize function, the audio gets super loud. Even if I have it set to 0.0001 it's really loud. I downgraded to 3.7.3 final and it's fixed. Don't know if I'm the only one with this issue but I wanted to give you a heads up.

[tebasuna51]

Originally Posted by ENunn

Hey there. I don't know if this is the place for bug reports or not but I found a bug. Whenever I use the normalize function, the audio gets super loud. Even if I have it set to 0.0001 it's really loud. I downgraded to 3.7.3 final and it's fixed. Don't know if I'm the only one with this issue but I wanted to give you a heads up.

I can't reproduce the problem with last test 14 (r4066), what is your installed version?
If it is the same version please upload the .avs used and the source (or a sample)

[pwnsweet]

I'm also having this issue. I'd love to enjoy 3.7.3 but haven't been able to resolve this so I'm stuck on 3.7.2

Originally Posted by real.finder

edit: anything newer than Avisynth+ 3.7.3 test 7 (20230223) will give same error

[kedautinh12]

Originally Posted by pwnsweet

I'm also having this issue. I'd love to enjoy 3.7.3 but haven't been able to resolve this so I'm stuck on 3.7.2

Cause you don't update to 3.7.3 beta version. You just use the 3.7.3 stable version

[Dogway]

What's currently the recommended audio loading filter in avisynth? I generally use ffms2 but apparently bestaudiosource is better/less bugs? The thing is bestaudiosource hasn't been updated in 3 years.

[Emulgator]

LWLibavAudioSource here, occasionally FFAudioSource

[ravewulf]

Be careful with LWLibavAudioSource and the Dolby codecs as LWLibavAudioSource applies DRC by default. drc_scale needs to be manually set to 0 to avoid it

https://github.com/HomeOfAviSynthPlu...Works/issues/5

[Julek]

Originally Posted by Dogway

The thing is bestaudiosource hasn't been updated in 3 years.

A big update is coming.

And the current repo is this one:
https://github.com/vapoursynth/bestsource

[Kurt.noise]

Hi,

I would like to test the channelmask internal functions from the latest stable release (3.7.3) using this script :

LoadPlugin("C:\Users\LionelDUCHATEAU\Downloads\megui_git\megui\bin\x64\Debug\tools\lsmash\LSMASHSource.dll")
LWLibavAudioSource("C:\Users\LionelDUCHATEAU\Downloads\AC3 Samples\8w3D.eac3", drc_scale=0)
AudioDubEx(BlankClip(length=Int(1000*AudioLengthF(last)/Audiorate(last)), width=720, height=720, fps=25), last)
m=IsChannelMaskKnown()
g=GetChannelMask()
s=SetChannelMask(true,g)

Info()
Subtitle(
\   "\nAudioLength  = "  + String(AudioLength)
\ + "\nAudioLengthS = '" + AudioLengthS + "'"
\ + "\nAudioLengthF = " + String(AudioLengthF)
\ + "\nAudioLengthLo= " + String(AudioLengthLo)
\ + "\nAudioLengthHi= " + String(AudioLengthHi)
\ + "\nIsChannelMaskKnown= " + String(m)
\ + "\nGetChannelMask= " + String(g)
\ + "\nSetChannelMask= " + String(s)
\ , font="courier", text_color=$ffffff, size=32, align=4, lsp=0)

but String(s) doesn't return something whereas I get information from Info(). Did I miss something here ?

[rgr]

Originally Posted by FranceBB

I'm pretty sure it's the "frozen as ice" issue with avstp.dll I faced a long time ago and that Ferenc fixed.

See here: https://github.com/pinterf/mvtools/issues/46

Either update to the new avstp.dll https://github.com/pinterf/AVSTP/releases or get rid of the one you have in the plugins folder.
I'm pretty sure that it will solve the issue

I updated a few things in the last 2 months (including AviSynth and AVSTP) and haven't had any freezes since. So something helped

[FranceBB]

Originally Posted by rgr

I updated a few things in the last 2 months (including AviSynth and AVSTP) and haven't had any freezes since. So something helped

Yep, Ferenc fixed it in avstp.
I mean, of course he did, he's amazing.

[Emulgator]

Be careful with LWLibavAudioSource and the Dolby codecs as LWLibavAudioSource applies DRC by default. drc_scale needs to be manually set to 0 to avoid it.

Thanks ravewulf. Comes in handy, just developing a snippet where this can be used:
https://forum.doom9.org/showthread.php?t=167435&page=80

[pinterf]

Originally Posted by Kurt.noise

Hi,

I would like to test the channelmask internal functions from the latest stable release (3.7.3) using this script :

LoadPlugin("C:\Users\LionelDUCHATEAU\Downloads\megui_git\megui\bin\x64\Debug\tools\lsmash\LSMASHSource.dll")
LWLibavAudioSource("C:\Users\LionelDUCHATEAU\Downloads\AC3 Samples\8w3D.eac3", drc_scale=0)
AudioDubEx(BlankClip(length=Int(1000*AudioLengthF(last)/Audiorate(last)), width=720, height=720, fps=25), last)
m=IsChannelMaskKnown()
g=GetChannelMask()
s=SetChannelMask(true,g)

Info()
Subtitle(
\   "\nAudioLength  = "  + String(AudioLength)
\ + "\nAudioLengthS = '" + AudioLengthS + "'"
\ + "\nAudioLengthF = " + String(AudioLengthF)
\ + "\nAudioLengthLo= " + String(AudioLengthLo)
\ + "\nAudioLengthHi= " + String(AudioLengthHi)
\ + "\nIsChannelMaskKnown= " + String(m)
\ + "\nGetChannelMask= " + String(g)
\ + "\nSetChannelMask= " + String(s)
\ , font="courier", text_color=$ffffff, size=32, align=4, lsp=0)

but String(s) doesn't return something whereas I get information from Info(). Did I miss something here ?

SetChannelMask returns the clip itself, you cannot stringify it.

If you'd like to get the friendly name of the channel mask constant combination, it's not possible. It's used by Info, so it is only an internal function.

Docs.
https://avisynthplus.readthedocs.io/...annelmask.html

[pinterf]

Originally Posted by FranceBB

Yep, Ferenc fixed it in avstp.
I mean, of course he did, he's amazing.

Your welcome, yes, the culprit was possibly avstp.

[pinterf]

What's happening in the background:
There are some difficult, though very specific phenomenons under investigation, such as
- why memory consumption grows when someone manually singlesteps forward 40 then backward 10 frames in Avspmod
- is it theoretically possible to eliminate the ever-growing behavior of Avisynth's string heap, which can be vry agressive if a runtime function (ScriptClip) contains a lot of string operation
They are both annoying ones but still, interesting challenges to solve.

Then there are some ideas from qyot27 about supporting/emulating frame property inheritance / setting for old plugins and filters which possibly would never get rebuilt.

- implemented 'continue' and 'break' for control loops, after Asd-g's request. (no build from me yet)
More info at
https://avisynthplus.readthedocs.io/...tinue-in-loops

[FranceBB]

Originally Posted by pinterf

- implemented 'continue' and 'break' for control loops

uuuuuuuuuuh very very nice!

[wonkey_monkey]

- why memory consumption grows when someone manually singlesteps forward 40 then backward 10 frames in Avspmod

Just Avspmod, not VirtualDub? I have my own viewer I can test against if there's a particular clip that has this behaviour.

[pinterf]

Originally Posted by wonkey_monkey

Just Avspmod, not VirtualDub? I have my own viewer I can test against if there's a particular clip that has this behaviour.

Probably the situation is the same with VirtualDub and with any tool that can request arbitrary frames.

[pinterf]

Originally Posted by pinterf

Probably the situation is the same with VirtualDub and with any tool that can request arbitrary frames.

Not necessarily. I don't know what AvsPmod is doing.

I put several debug lines to see what the real frame sequence that must be served by Avisynth (avs_get_frame C interface call). In order to see, which specific frame sequence fools Avisynth prefetcher prediction logic. (I supposed that when I do a backward single-step in Avspmod, Avisynth prefetcher will recognize the new - now negative - pattern, and the Prefetcher will lock into the new delta=-1 prefetching schema.)

And it seems that AvsPMod is requesting the same frame numbers multiple times during its display, even in arbitrary (not one-by-one) order

I see multiple clips, two Prefetch statements, which made the debugging a bit harder , maybe they are different clips, one before the resizing - the actual unaltered clip - and one after the display conversion?

E.g. sometimes, when I press the "back-reverse", it skips back by 2 frames on one of its script instance, which ends with Prefetch(1,1).
And only by 1 frames with another (?) script instance which ends with the Prefetch(4) - the last line in the script.

@gispos, can you help me with it? The actual issue on github is:
https://github.com/AviSynth/AviSynth...ent-2048852993
after a doom9 report:
https://forum.doom9.org/showthread.p...03#post1995403
Thank you