Vapoursynth

[Julek]

Originally Posted by lansing

I have a problem with the R59 installer. I have both Python 3.8 and 3.9 installed with 3.9 being the default. But the installer is only able to detect 3.8. I tested R57 and it is able to see both python versions.

Since R58 VS supports only 3.8 OR 3.10.
3.8 is still kept because it is the latest version available for windows 7.

More info: https://github.com/vapoursynth/vapoursynth/issues/858

[lansing]

Originally Posted by Julek

Since R58 VS supports only 3.8 OR 3.10.
3.8 is still kept because it is the latest version available for windows 7.

More info: https://github.com/vapoursynth/vapoursynth/issues/858

The documentation should be updated, as it is still stating 3.9 to be the supported version.

http://www.vapoursynth.com/doc/insta...s-installation

[Julek]

Originally Posted by lansing

The documentation should be updated, as it is still stating 3.9 to be the supported version.

http://www.vapoursynth.com/doc/insta...s-installation

Yes, the site usually takes longer to update, even though it is already updated on GitHub.

[Myrsloik]

R60 RC1

r60:
fixed blankclip crashing when fpsnum or fpsden is 0 despite it being allowed
fixed invalid format ids sometimes being treated as api3 values when passed to resize 
fixed error message frame generation in vfw
fixed writing beyond end out output buffer for audio in vfw
blankaudio now takes an array of channel constants since this better matches how most other functions work
fixed splitchannels crash on certain channel configurations (YomikoR)
added makefulldiff and mergefulldiff, these are versions of makediff and mergediff that don't clamp the difference to half range and instead use a higher precision diff clip
better error messages when filters get unsupported input formats or combinations thereof
freezeframes now accepts empty arrays and simply passes through the source clip
you can directly assign to frame-props
removed all deprecated (and replaced by better versions) functions in python: get_frame_async_raw, get_plugins, get_functions, list_functions, get_format, register_format
fixed convolution output for 9-15 bit material

Note that MakeFullDiff and MergeFullDiff are experimental functions that may be killed off if nobody does fun things with them. That's all.

[Greenhorn]

What's the difference between Make/MergeDiff and Make/MergeFullDiff when using float format? Was that also clamped?

[Myrsloik]

Originally Posted by Greenhorn

What's the difference between Make/MergeDiff and Make/MergeFullDiff when using float format? Was that also clamped?

It's the same for float.

[mastrboy]

Originally Posted by vapoursynth doc

ColorYUV std.Lut/std.Expr Do the adjustment yourself
https://github.com/amichaelt/vapours...isynthcomp.rst

Honestly, this is just too complicated to figure out for me, is there any helper functions available for this?

I tried to understand the code here: https://github.com/AviSynth/AviSynth...ters/color.cpp

But can not figure out how the Lut/expression for off_v is built from those functions...

I'm just trying to do the avisynth equivalent to "coloryuv(off_v=0.5)"

[Myrsloik]

Originally Posted by mastrboy

Honestly, this is just too complicated to figure out for me, is there any helper functions available for this?

I tried to understand the code here: https://github.com/AviSynth/AviSynth...ters/color.cpp

But can not figure out how the Lut/expression for off_v is built from those functions...

I'm just trying to do the avisynth equivalent to "coloryuv(off_v=0.5)"

Add .5 to the third plane of a float format?

Expr(clip, ["", "", "x 0.5 +"])

[mastrboy]

Originally Posted by Myrsloik

Add .5 to the third plane of a float format?

Expr(clip, ["", "", "x 0.5 +"])

Thanks, that works... I have a difficult time properly understanding lut/expr, is there any decent documentation freely available that is recommended for beginners that want to learn?

That only works for 8bit though, I guess ColorYUV scales input values internally... So what is the math to convert 8bit Expr to 16bit?

Edit: nevermind, found a helper function for scaling values between bitdepths: https://github.com/Irrational-Encodi...l/info.py#L129