AviSynth+ thread Vol.2

[LigH]

Another variation: If you don't explicitly use any return statement, then the last statement of a script is the implicit "return last" statement. All you have to do is assigning your active clip variable to the implicit "last" clip variable. You can do that with an explicit assignment using the equal sign, or implicitly by just mentioning the clip variable in a single line.

myclip = some_filter()

myclip
# means the same as:
# last = myclip

Prefetch(N)
# means the same as:
# last = last.Prefetch(N)

# implicit last statement:
# return last

Keep in mind: Wherever you don't explicitly assign a specific clip variable (using "=" or "."), the result of a filter is assigned to the implicit clip variable "last".

[DTL]

"the last statement of a script is the implicit "return last" statement."

If add

return last
Prefetch(N)


The MT also not working.

Prefetch(N) before return -

Prefetch(N)
return last

the MT is working. Putting Prefetch to the beginning of script returns error 'bad arguments for Prefetch'. So it looks Prefetch(N) need to be applied to the last (output) clip ?

Found in the documentation: http://avisynth.nl/index.php/SetFilterMTMode
Enabling MT
You enable MT by placing a single call to Prefetch(X) at the end of your script, where X is the number of threads to use. If there is a return statement in your script it must be placed after Prefetch().

So to return something from middle of script with MT:

myclip = some_filter()

myclip=Prefetch(myclip, N)

return myclip

?

[LigH]

You really misunderstood.

"Implicit" statements are those you do not write in the script but they are in fact executed because you did not write them.

"the last statement of a script is the implicit "return last" statement."

But only

If you don't explicitly use any return statement

Only if.

[Gavino]

If you don't explicitly use any return statement, then the last statement of a script is the implicit "return last" statement.
Being pedantic, it is not strictly true that "return last" is the default action when 'return' is omitted.
In practice, that is usually what happens, but there are some cases when 'last' is not returned, such as when the final expression is not a clip. See here (http://forum.doom9.org/showthread.php?p=1642079#post1642079).)

[guest]

Deleted. Wrong thought.

Try this :-

Go into "edit", and the delete option is in the bottom right corner of the message window.

[gispos]

Try this :-

Go into "edit", and the delete option is in the bottom right corner of the message window.

Thanks :)

[TomArrow]

Any clues on the "Device unmatch: ConvertToRGB48[CPU] does not support [CUDA] frame" problem?

Way to reproduce:

1. Get newest AVISynth+ release.

2. Grab CoronaSequence plugin from http://avisynth.nl/index.php/ImageSequence

3. Load image sequence with CoronaSequence command.

4. Add ConvertToRGB64() at the end.

[Emulgator]

If I only had enough time I would skim AviSynth source code for that error
"Device unmatch: ConvertToRGB48[CPU] does not support [CUDA] frame"

[VoodooFX]

Any clues on the "Device unmatch: ConvertToRGB48[CPU] does not support [CUDA] frame" problem?
No error on my side.

[Reel.Deel]

If I only had enough time I would skim AviSynth source code for that error
"Device unmatch: ConvertToRGB48[CPU] does not support [CUDA] frame"

https://github.com/AviSynth/AviSynthPlus/blob/master/avs_core/core/DeviceManager.cpp#L81

[TomArrow]

I should have added that I have an AMD 5700G APU, but someone else I know had this same issue and he has an RTX 3090 and an i9-10850k. Both our PCs run Win 10 Pro.

It happens with every AviSynth+ version above 3.7.0. 3.7.0 itself doesn't have this problem.

@Reel.Deel Yeah I found this code as well but I'm not sure what to do about it. I don't know how to debug AVISynth itself and I don't understand any of that code. And the plugin was written before any of this CPU/GPU stuff was added so it shouldn't really do anything in that regard.

Edit: Also for the record I wrote "ConvertToRGB48" in the error message and "ConvertToRGB64" in the reproduction steps. This was a mistake, however it applies to both really, it happens with both attempted conversions.

[flossy_cake]

Is there any interest in upgrading Avisynth+'s internal DirectShowSource filter to enable it to output 10-bit pixel formats like P010 or Y410 as these are the only 10-bit ones supported by LAV Video Decoder which means we can't get 10-bit output from DirectShowSource if using LAV.

I tried forcing pixel format to RGB48 which is the only >8-bit format supported by both LAV Video and DirectShowSource, however it fails to load LAV at all and I just get a blank screen.

edit: I noticed FFMpegSource2 and LWlibavVideoSource do support P010 but I'd still prefer to use DirectShow since it can use LAV Video Decoder which has a good feature set & no caching/indexing overhead.

edit: relevant log bit for DirectShowSource where it seem to be trying to use P010 but failing (I don't know why... LAV Video supports it)


00:00:00.270 001 0x00000000048542D0 0x00001154 *** Video: Subtype rejected - 'HVC1' {31435648-0000-0010-8000-00aa00389b71}
00:00:00.270 001 0x00000000048542D0 0x00001154 *** Video: Format type - {e06d80e3-db46-11cf-b4d1-00805f6cbbea}
00:00:00.270 020 0x00000000048542D0 0x00001154 GetSample::ReceiveConnection() ** VFW_E_TYPE_NOT_ACCEPTED **
00:00:00.300 020 0x00000000048542D0 0x00001154 GetSample::ConnectedTo() ** VFW_E_NOT_CONNECTED **
00:00:00.300 001 0x00000000048542D0 0x00001154 GetSample::QueryAccept(video) MEDIATYPE_Video
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Subtype rejected - 'P010' {30313050-0000-0010-8000-00aa00389b71}
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Format type - {f72a76a0-eb0a-11d0-ace4-0000c0cc16ba}
00:00:00.300 020 0x00000000048542D0 0x00001154 GetSample::ReceiveConnection() ** VFW_E_TYPE_NOT_ACCEPTED **
00:00:00.300 001 0x00000000048542D0 0x00001154 GetSample::QueryAccept(video) MEDIATYPE_Video
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Subtype rejected - 'P010' {30313050-0000-0010-8000-00aa00389b71}
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Format type - {05589f80-c356-11ce-bf01-00aa0055595a}
00:00:00.300 020 0x00000000048542D0 0x00001154 GetSample::ReceiveConnection() ** VFW_E_TYPE_NOT_ACCEPTED **
00:00:00.300 001 0x00000000048542D0 0x00001154 GetSample::QueryAccept(video) MEDIATYPE_Video
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Subtype rejected - 'P016' {36313050-0000-0010-8000-00aa00389b71}
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Format type - {f72a76a0-eb0a-11d0-ace4-0000c0cc16ba}
00:00:00.300 020 0x00000000048542D0 0x00001154 GetSample::ReceiveConnection() ** VFW_E_TYPE_NOT_ACCEPTED **
00:00:00.300 001 0x00000000048542D0 0x00001154 GetSample::QueryAccept(video) MEDIATYPE_Video
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Subtype rejected - 'P016' {36313050-0000-0010-8000-00aa00389b71}
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Format type - {05589f80-c356-11ce-bf01-00aa0055595a}
00:00:00.300 020 0x00000000048542D0 0x00001154 GetSample::ReceiveConnection() ** VFW_E_TYPE_NOT_ACCEPTED **
00:00:00.300 001 0x00000000048542D0 0x00001154 GetSample::QueryAccept(video) MEDIATYPE_Video
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Subtype denied - NV12
00:00:00.300 020 0x00000000048542D0 0x00001154 GetSample::ReceiveConnection() ** VFW_E_TYPE_NOT_ACCEPTED **
00:00:00.300 001 0x00000000048542D0 0x00001154 GetSample::QueryAccept(video) MEDIATYPE_Video
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Subtype denied - NV12
00:00:00.300 020 0x00000000048542D0 0x00001154 GetSample::ReceiveConnection() ** VFW_E_TYPE_NOT_ACCEPTED **
00:00:00.300 001 0x00000000048542D0 0x00001154 GetSample::QueryAccept(video) MEDIATYPE_Video
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Subtype accepted - 'YV12' {32315659-0000-0010-8000-00aa00389b71}
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: Format type accepted - {f72a76a0-eb0a-11d0-ace4-0000c0cc16ba}
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: format accepted: 1920x1080, pixel_type YV12, avg_time_per_frame 417083x100ns
00:00:00.300 001 0x00000000048542D0 0x00001154 *** Video: bFixedSizeSamples=1, bTemporalCompression=0, lSampleSize=3110400
00:00:00.512 002 0x00000000048542D0 0x0000135C GetSample::NewSegment(0, 63048817080, 1.000000) (video)
00:00:00.602 001 0x00000000048542D0 0x00001368 GetSample::QueryAccept(video) MEDIATYPE_Video
00:00:00.602 001 0x00000000048542D0 0x00001368 *** Video: Subtype accepted - 'YV12' {32315659-0000-0010-8000-00aa00389b71}
00:00:00.602 001 0x00000000048542D0 0x00001368 *** Video: Format type accepted - {f72a76a0-eb0a-11d0-ace4-0000c0cc16ba}
00:00:00.602 001 0x00000000048542D0 0x00001368 *** Video: format accepted: 1920x1080, pixel_type YV12, avg_time_per_frame 417083x100ns
00:00:00.602 001 0x00000000048542D0 0x00001368 *** Video: bFixedSizeSamples=1, bTemporalCompression=0, lSampleSize=3110400
00:00:00.602 020 0x00000000048542D0 0x00001368 GetSample::ReceiveConnection() ** VFW_E_NOT_STOPPED **
00:00:00.612 002 0x00000000048542D0 0x00001368 Receive: video sample time span x100ns 0 to 417083 (417083)
00:00:00.632 005 0x00000000048542B0 0x00001154 Directshow duration 63048817080, frame_count 0.
00:00:00.632 005 0x00000000048542B0 0x00001154 New Video: 1920x1080, frame_count=151167, pixel type=YV12.
00:00:00.632 fff 0x00000000 Close Create_DirectShowSource log 3.


The clip is 10-bit 4:2:0.

[tormento]

Is KTGMC supported by AVS+? I have read somewhere that the plugin should be rewritten to make it work.

[Reel.Deel]

Is KTGMC supported by AVS+? I have read somewhere that the plugin should be rewritten to make it work.

No.

1). Official AviSynth+ releases are not built with CUDA support enabled.

2). All of the AvisynthCUDAFilters need to be recompiled to work with the current AviSynth+.

See the issues page: https://github.com/AviSynth/AviSynthPlus/issues/296

[tormento]

No.
That is a real pity.

I hope that Pinterf (or someboby else!) one day will port the most useful Nekopanda plugins.

[gispos]

That is a real pity.

I hope that Pinterf (or someboby else!) one day will port the most useful Nekopanda plugins.

+1, it is a real pity that the Cuda support was not followed up.

[FranceBB]

+1, it is a real pity that the Cuda support was not followed up.

+2
We have Avisynth Neo largely merged into Avisynth+ now, however we can't use the actual useful part of it which is the OnCUDA() and OnCPU() functions with the old AVS Neo plugins. It would be really nice if someone picked the Neo plugins up and recompiled them to target the modern AVS+ headers as it would speed things up a lot.

[poisondeathray]

avs+ r3682 x64
https://i.postimg.cc/TwQ1BXF7/yuy2-coloryuv-analyze-true.png


b=blankclip(width=360, height=160, colors=[0,128,128], pixel_type="YV16")
w=blankclip(width=360, height=160, colors=[255,128,128], pixel_type="YV16")

stackhorizontal(b,w)
coloryuv(analyze=true)
subtitle("YV16")
planar=last

b=blankclip(width=360, height=160, colors=[0,128,128], pixel_type="YUY2")
w=blankclip(width=360, height=160, colors=[255,128,128], pixel_type="YUY2")

stackhorizontal(b,w)
coloryuv(analyze=true)
subtitle("YUY2")
packed=last

stackvertical(planar,packed.ConvertToYV16())

[kedautinh12]

Are you try r3820?
https://gitlab.com/uvz/AviSynthPlus-Builds

[poisondeathray]

Are you try r3820?
https://gitlab.com/uvz/AviSynthPlus-Builds

same issue with r3820

[Reel.Deel]

same issue with r3820

Not much has changed since r3682. The much higher number is just because of 120+ documentation commits.

I created a new issue for this bug: https://github.com/AviSynth/AviSynthPlus/issues/304

[pinterf]

Avisynth 3.7.3 test 2
Avisynth+ 3.7.3 test 2 (20221114) (https://drive.google.com/uc?export=download&id=1fCPbb4n1-nuR_mAaLG_4o1Zxfciwr2A4)

20221114 3.7.3 WIP
------------------
- Fix (#304): ColorYUV analyze=true was displaying wrong min-max values for YUY2
- Fix: C API undefined behavior when upstream throw runtime error
- Mute compilation warnings in avisynth.h
- CMakeLists.txt: fix clang-cl/intel with ninja generator
- Fix (#293): "Text" to throw proper error message if the specified font name (e.g. Arial) is not found among internal bitmap fonts.
- Fix (#293): "Subtitle" and "Text" filter to respect the explicitely given coorditanes for y=-1 or x=-1,
instead of applying vertical/horizontal center alignment.
- Fix: C interface crash when using avs_new_video_frame_p(_a)
- Fix (#283): broken runtime functions min/max/minmaxdifference when threshold is not 0 (returned -1). Regression in 3.7.2
- New: add a sixth array element to PlaneMinMaxStats: average. Defines variable "PlaneStats_average" as well if setting variables is required.
- Fix (#282): ConvertToRGB
- do check for exact 8 or 16 bit input, because packed RGB formats exist only for 8 and 16 bits
- keep alpha for RGBA planar - convert RGBAP8/16 to RGB32/64, while RGBP8/16 is still RGB24/48

[VoodooFX]

Avisynth 3.7.3 test 2

Thanks. So far so good.

[Dogway]

Thanks a lot pinterf, does this inherit changes from r3820 or are they different branches?

[Reel.Deel]

Thanks a lot pinterf, does this inherit changes from r3820 or are they different branches?

Same branch, pinterf's build is r3825.

[FranceBB]

Thank you, Ferenc!
Hugely appreciated, as always! :D
This is mid-November and we're already at Beta 2, looks like it's gonna be another white Christmas with Avisynth 3.7.3 Stable! :D

[poisondeathray]

Thanks for the new test version


RGBAdjust clips to [0,1] for RGBPS ? Is it intended behaviour ?


blankclip(pixel_type="RGBPS", colors=[1.1, 1.1, 1.1])
#avspmod correctly reads 1.1,1.1,1.1 with color picker

#RGBAdjust(rb=-0.05) #avspmod and RGBAdjust(analyze=true) read 1,1,1

#RGBAdjust(rb=-0.2) #avspmod and RGBAdjust(analyze=true) read 0.9,1,1 ; implies clipping only occurs after the adjustment

#RGBAdjust(analyze=true) #avspmod reads 0 to 1, when the call is alone, or any instance prefaced with RGBAdjust

#expr("x 0.05 -", "x", "x") #works correctly, avspmod reads 1.05,1.1,1.1 ; but when followed by RGBAdjust(analyze=true) reads 1,1,1


There is no way to load native float content into avs+ yet (eg. .hdr, .exr, or .tiff float, maybe the vapoursynth IM plugin could be ported...ahem.. :D )

[StainlessS]

Thanx P.

[pinterf]

RGBAdjust clips to [0,1] for RGBPS ? Is it intended behaviour ?

Yes, clipping is done before applying gamma.

[pinterf]

Thank you, Ferenc!
Hugely appreciated, as always! :D
This is mid-November and we're already at Beta 2, looks like it's gonna be another white Christmas with Avisynth 3.7.3 Stable! :D
Positive aspect of slowness: I spared tens of hours per month by not visiting doom9 forums. Addiction is dangerous.

[pinterf]

Any clues on the "Device unmatch: ConvertToRGB48[CPU] does not support [CUDA] frame" problem?

Way to reproduce:

1. Get newest AVISynth+ release.

2. Grab CoronaSequence plugin from http://avisynth.nl/index.php/ImageSequence

3. Load image sequence with CoronaSequence command.

4. Add ConvertToRGB64() at the end.
Being a 64 bit plugin from 2010, when no 64 bit Avisynth+ existed, it is pure luck if it works or not. Plus: maybe it is a 2.5-style plugin which can be an additional pain. If so, the plugin must be first recompiled with a 2.6 / Avisynth+ header.

[Reel.Deel]

Being a 64 bit plugin from 2010, when no 64 bit Avisynth+ existed, it is pure luck if it works or not. Plus: maybe it is a 2.5-style plugin which can be an additional pain. If so, the plugin must be first recompiled with a 2.6 / Avisynth+ header.

It seems Tom is using the Avisynth+ header but with AvisynthPluginInit2: https://github.com/TomArrow/CoronaSequence_x64mod/blob/master/imagesequence.cpp#L446

[kedautinh12]

Latest ver
https://github.com/TomArrow/CoronaSequence_x64mod/releases

[TomArrow]

It seems Tom is using the Avisynth+ header but with AvisynthPluginInit2: https://github.com/TomArrow/CoronaSequence_x64mod/blob/master/imagesequence.cpp#L446

Yeah I think I use the modern SDK.

Is this wrong? Either way I think this CUDA/whatever problem is not intended behavior so a fix would probably be nice to have.

I guess it's possible that the problem is in some of the plugin code elsewhere, like accidentally overwriting some property due to overflow or bad pointer handling, but I wouldn't even know where to start looking since I don't understand the feature. I'm also not the original author of the plugin, I just made changes.

[Reel.Deel]

Yeah I think I use the modern SDK.

Is this wrong?

Probably best to use AvisynthPluginInit3 when using AviSynth 2.6/AviSynth+ headers.

See here: https://github.com/pinterf/TNLMeans/commit/d0fb74ae8d8ef71100b52fb3131556f25441502a

[kedautinh12]

Avisynth 3.7.3 test 2
Avisynth+ 3.7.3 test 2 (20221114) (https://drive.google.com/uc?export=download&id=1fCPbb4n1-nuR_mAaLG_4o1Zxfciwr2A4)

20221114 3.7.3 WIP
------------------
- Fix (#304): ColorYUV analyze=true was displaying wrong min-max values for YUY2
- Fix: C API undefined behavior when upstream throw runtime error
- Mute compilation warnings in avisynth.h
- CMakeLists.txt: fix clang-cl/intel with ninja generator
- Fix (#293): "Text" to throw proper error message if the specified font name (e.g. Arial) is not found among internal bitmap fonts.
- Fix (#293): "Subtitle" and "Text" filter to respect the explicitely given coorditanes for y=-1 or x=-1,
instead of applying vertical/horizontal center alignment.
- Fix: C interface crash when using avs_new_video_frame_p(_a)
- Fix (#283): broken runtime functions min/max/minmaxdifference when threshold is not 0 (returned -1). Regression in 3.7.2
- New: add a sixth array element to PlaneMinMaxStats: average. Defines variable "PlaneStats_average" as well if setting variables is required.
- Fix (#282): ConvertToRGB
- do check for exact 8 or 16 bit input, because packed RGB formats exist only for 8 and 16 bits
- keep alpha for RGBA planar - convert RGBAP8/16 to RGB32/64, while RGBP8/16 is still RGB24/48

Clang build
https://gitlab.com/uvz/AviSynthPlus-Builds

[gispos]

Since Avisynth 3.7.3 or the extra builds from GS (https://gitlab.com/uvz/AviSynthPlus-Builds) I get error messages every now and then that are not present in Avisynth 3.7.2

Either no clip can be created (without error message) or the error message 'Function not found' appears (v3.7.3).
And in both cases: If the same script is loaded again, no error occurs.

If I go back to 3.7.2 everything is fine again.

I load some dll's and scripts via an avsi, has anything changed in the plugin loading order?

[ryrynz]

Yeah I decided to test the test build as well and just got nothing out the other end, I don't think test builds should be that broken :D
Basically an alpha, so I wasn't concerned and just went back to 3.7.2 as well, figuring something so broken would be identified soon enough.
Clang build wasn't any better either.

[gispos]

...so I wasn't concerned and just went back to 3.7.2 as well, figuring something so broken would be identified soon enough.
Clang build wasn't any better either.
If no one reports it, it cannot be identified.
I would like to provide more detailed information, but it is completely random and also not so often.

[VoodooFX]

Report: I still had no issues with Avisynth+ 3.7.3 test 2 (20221114) (https://drive.google.com/uc?export=download&id=1fCPbb4n1-nuR_mAaLG_4o1Zxfciwr2A4)

[poisondeathray]

Report: I still had no issues with Avisynth+ 3.7.3 test 2 (20221114) (https://drive.google.com/uc?export=download&id=1fCPbb4n1-nuR_mAaLG_4o1Zxfciwr2A4)

No problems either;

For the people having issues, post your script(s) maybe some clues there

[gispos]

Since I changed my avsi file to load my filters and dll's, there have been no problems with Avisynth 3.73 so far.

Before everything was listed randomly, now I load all dll's first and only then the scripts.

Something must have changed because 3.72 had no problems with the mixed loading order.

As long as it works now all is good. :)

[pinterf]

Happy New Year!

Let's start the year with some cosmetic and other changes and fixes in "Text" filter:
Avisynth+ 3.7.3 test 3 (20230118) (https://drive.google.com/uc?export=download&id=1FEww_BfRMY63D2NCg4h860ofxqfflRhu)

- Fix: "Text" filter negative x or y coordinates (e.g. 0 instead of -1)
- Fix: "Text" filter would omit last character when x<0
- Fix: "Text" halo_color needs only MSB=$FF and not the exact $FF000000 constant for fade
- "Text" ``halo_color`` allows to have both halo and shaded background
- "Text" much nicer rendering of subsampled formats

[FranceBB]

Happy New Year to you too, Master Ferenc!
Thanks for the new build! I'll definitely test! :D

[Reel.Deel]

Also, 32-bit AviSynth+ now supports non-decorated export names in C-plugins. This is something that was missing since the beginning of avs+.

By the way, this is how the Text filter on YUV420 looks like now, and how it used to look.

New
https://i.ibb.co/Z1sS3cP/text-new.png

Old
https://i.ibb.co/bPWq0Mz/text-old.png

[Emulgator]

Ah, welcome back, Ferenc !
(Was trying uvx 3825 and both 32 and 64 bit failed instantly...)

[StainlessS]

Welcome home P, and Happy New Year to U 2.

[Dogway]

Where is the Text filter in wiki?
I get error trying to use "Arial" as the font.
SimpleText: internal font name Arial in size 18 not found.

[Reel.Deel]

Where is the Text filter in wiki?
I get error trying to use "Arial" as the font.

It's not documented in the wiki. That is why there's a note at the top of the page with the link to the readthedocs documentation which is up-to-date. See here: https://avisynthplus.readthedocs.io/en/latest/avisynthdoc/corefilters/subtitle.html#text

Text, is meant for bdf fonts (https://en.wikipedia.org/wiki/Glyph_Bitmap_Distribution_Format). It comes with "Terminus" (available in different sizes) and "Info.h" that is a fixed dimension.

---

PS, if someone wants to update the wiki, go for it. Updating the actual avs docs was a large task, and I don't feel like doing double work :).

[kedautinh12]

Happy New Year!

Let's start the year with some cosmetic and other changes and fixes in "Text" filter:
Avisynth+ 3.7.3 test 3 (20230118) (https://drive.google.com/uc?export=download&id=1FEww_BfRMY63D2NCg4h860ofxqfflRhu)

- Fix: "Text" filter negative x or y coordinates (e.g. 0 instead of -1)
- Fix: "Text" filter would omit last character when x<0
- Fix: "Text" halo_color needs only MSB=$FF and not the exact $FF000000 constant for fade
- "Text" ``halo_color`` allows to have both halo and shaded background
- "Text" much nicer rendering of subsampled formats

Some new 32-bit plugin don't support XP anymore. So next update can you add x86 for win 7 and up beside x86-xp??