Vapoursynth

[Myrsloik]

R48 RC1

Go test it everywhere. Once again all known regressions have been fixed. Mask all the things!

r48:
it's now possible to select which optimized code path is used for internal filters (sekrit-twc)
avx2 optimization in many filters, previously the internal ones were mostly sse2 (sekrit-twc)
expr filter can now better optimize expressions (sekrit-twc)
the 7zip executable is now bundled with vsrepo
the portable version now includes the documentation as well
the portable version now includes all the plugins bundled with the normal installer again

[ChaosKing]

Now that python 3.8 is final, will R48 support it?

[Myrsloik]

Originally Posted by ChaosKing

Now that python 3.8 is final, will R48 support it?

Nope, there's no cython release with proper 3.8 support yet.

[Myrsloik]

R48-RC2

More bugs and minor issues fixed. Should be the final RC unless someone finds a serious bug within a week. Test the new add to PATH option a bit extra.

r48:
it's now possible to select which optimized code path is used for internal filters (sekrit-twc)
avx2 optimization in many filters, previously the internal ones were mostly sse2 (sekrit-twc)
expr filter can now better optimize expressions (sekrit-twc)
the 7zip executable is now bundled with vsrepo
the portable version now includes the documentation as well
the portable version now includes all the plugins bundled with the normal installer again
fixed deadlock when setMessageHandler is called a second time
added an option to add vspipe, avfs and vsrepo to path in the installer
added registry entries for the path to vspipe and vsrepo
imwri is now included in the installer
the overwrite argument in imwri now also disables the requirement for output filesnames to contain a number
fixed corrupt output from imwri when requesting alpha output but the read image doesn't have an alpha channel

[ChaosKing]

The PATH stuff works for me (vspipe, vsrepo, avfs). Green stripes are fixed too.

[l00t]

Unfortunately I still have the pink borders, when bbmod (from latest havsfunc) is used. See sample pics:

R47.2:
https://images2.imgbox.com/21/ee/LCxpNQn8_o.png

R48-RC2:
https://images2.imgbox.com/f0/df/Nlp16rcQ_o.png

Relevant part of the VS code:

clip = core.std.SetFieldBased(clip, 0)
clip = core.std.CropRel(clip=clip, left=0, right=0, top=20, bottom=20)
clip = core.fb.FillBorders(clip, left=0, right=0, top=1, bottom=1, mode="fillmargins") 
clip = havsfunc.bbmod(clip, cTop = 2, cBottom = 2, cLeft = 0, cRight = 0, thresh = 128, blur = 999)
clip = core.resize.Spline36(clip, width=clip.width-0-0, height=clip.height-1-1, src_left=0, src_top=1, src_width=clip.width-0-0, src_height=clip.height-1-1)

I used the portable version of R47.2 and R48-RC2 (both x64) with VSEditor r19 and Python 3.7.5 (embedded). Video is loaded with dgdecodenv.DGSource, most recent version.

[Myrsloik]

Originally Posted by l00t

Unfortunately I still have the pink borders, when bbmod (from latest havsfunc) is used. See sample pics:

R47.2:
https://images2.imgbox.com/21/ee/LCxpNQn8_o.png

R48-RC2:
https://images2.imgbox.com/f0/df/Nlp16rcQ_o.png

Relevant part of the VS code:

clip = core.fb.FillBorders(clip, left=0, right=0, top=1, bottom=1, mode="fillmargins") 
clip = havsfunc.bbmod(clip, cTop = 2, cBottom = 2, cLeft = 0, cRight = 0, thresh = 128, blur = 999)
clip = core.resize.Spline36(clip, width=clip.width-0-0, height=clip.height-1-1, src_left=0, src_top=1, src_width=clip.width-0-0, src_height=clip.height-1-1)

I used the portable version of R47.2 and R48-RC2 (both x64) with VSEditor r19 and Python 3.7.5 (embedded). Video is loaded with dgdecodenv.DGSource, most recent version.

Can't reproduce. CPU and input video format+resolution?

[l00t]

Originally Posted by Myrsloik

Can't reproduce. CPU and input video format+resolution?

CPU: Intel i7-9750H
Video format: YUV420P8; 25fps; input resolution: 1920x1080; final resolution: 1920x1038

(added 2 more lines, which might be important as well...)

[Myrsloik]

Originally Posted by l00t

CPU: Intel i7-9750H
Video format: YUV420P8; 25fps; resolution: 1920x1038

(added 2 more lines, which might be important as well...)

Added how? Ideally I want a way to reproduce it using blankclip as the source and a complete script.

[l00t]

Originally Posted by Myrsloik

Added how? Ideally I want a way to reproduce it using blankclip as the source and a complete script.

Sorry for being a bit misleading, these were already in the script, when the pictures were taken... (I just wanted to emphasize, that something is happening in bbmod, the script looks fine without it). Nevermind, here's the complete script:

clip = core.dgdecodenv.DGSource(r'some_hd_video.dgi')
clip = core.resize.Spline36(clip, matrix_in_s="709", transfer_in_s="709", primaries_in_s="709", range_s="limited")
clip = core.std.AssumeFPS(clip, fpsnum=25000, fpsden=1000)
clip = core.std.SetFrameProp(clip=clip, prop="_ColorRange", intval=1)
clip = core.std.SetFieldBased(clip, 0)
clip = core.std.CropRel(clip=clip, left=0, right=0, top=20, bottom=20)
clip = core.fb.FillBorders(clip, left=0, right=0, top=1, bottom=1, mode="fillmargins")
clip = havsfunc.bbmod(clip, cTop = 2, cBottom = 2, cLeft = 0, cRight = 0, thresh = 128, blur = 999)
clip = core.resize.Spline36(clip, width=clip.width-0-0, height=clip.height-1-1, src_left=0, src_top=1, src_width=clip.width-0-0, src_height=clip.height-1-1)
clip = core.remap.ReplaceFramesSimple(baseclip=clip, sourceclip=core.f3kdb.Deband(clip, range=20, grainy=32, grainc=24, sample_mode=2, dither_algo=3, keep_tv_range=1, blur_first=1, dynamic_grain=0), mappings="[0 1070]")
clip.set_output()

[Myrsloik]

R48-RC3

Fixes the aformentioned pink line at top and bottom (general expr bug) and another 32bit bug in expr as well. Keep testing it.

[l00t]

Wow, so fast, thank you very much! The pinkies are now gone, hooray

[Jukus]

Is there any way to use spline144 with Vapour?
Is it true that this is the best resizer for any situation?

[poisondeathray]

Originally Posted by Jukus

Is there any way to use spline144 with Vapour?

With fmtconv kernel="spline", taps=6

"spline"

Generic splines, number of sample points is twice the taps parameter, so you can use taps = 6 to get a Spline144Resize equivalent.

Is it true that this is the best resizer for any situation?

Probably not. Likely some ringing artifacts, oversharpen halos

[Myrsloik]

Originally Posted by Myrsloik

R48-RC3

Fixes the aformentioned pink line at top and bottom (general expr bug) and another 32bit bug in expr as well. Keep testing it.

Keep testing it! So far the only discovered bug in RC3 is in Convolution with float input, everything else should work fine. I'll probably do daily builds from now on until every single regression is gone.

[poisondeathray]

If you have RGBS and use Expr to add 1 to channel R , is it suppose to round the value ? Or is this the same thing as the Convolution with float input issue ?

clip2 = core.std.Expr(clip, ["x 1 +", "", ""])

e.g. in vsedit, the color picker shows R=0.921468 for clip, R=1.92147 for clip2

I would have expected 1.921468 (G, B remain unchanged at 0.921468)

core.std.SetMaxCPU("none") does not affect the result, R48-RC3 x64

[Myrsloik]

Originally Posted by poisondeathray

If you have RGBS and use Expr to add 1 to channel R , is it suppose to round the value ? Or is this the same thing as the Convolution with float input issue ?

clip2 = core.std.Expr(clip, ["x 1 +", "", ""])

e.g. in vsedit, the color picker shows R=0.921468 for clip, R=1.92147 for clip2

I would have expected 1.921468 (G, B remain unchanged at 0.921468)

core.std.SetMaxCPU("none") does not affect the result, R48-RC3 x64

That's how floating point math works. It's not exact.

[StainlessS]

From VS2008 Float.h

#define FLT_DIG         6                       /* # of decimal digits of precision */
#define FLT_EPSILON     1.192092896e-07F        /* smallest such that 1.0+FLT_EPSILON != 1.0 */
#define FLT_GUARD       0
#define FLT_MANT_DIG    24                      /* # of bits in mantissa */
#define FLT_MAX         3.402823466e+38F        /* max value */
#define FLT_MAX_10_EXP  38                      /* max decimal exponent */
#define FLT_MAX_EXP     128                     /* max binary exponent */
#define FLT_MIN         1.175494351e-38F        /* min positive value */
#define FLT_MIN_10_EXP  (-37)                   /* min decimal exponent */
#define FLT_MIN_EXP     (-125)                  /* min binary exponent */
#define FLT_NORMALIZE   0
#define FLT_RADIX       2                       /* exponent radix */
#define FLT_ROUNDS      1                       /* addition rounding: near */

[poisondeathray]

Thanks Myrsloik, StainlessS ; If it's using FLT_DIG 6 , shouldn't it round to 1.921468 ? Or does it count decimal digits differently ? Or was this covered in grade school math that I might have slept through?

[StainlessS]

Its really the binary digits that count, and 10 is not a power of 2, so it works (rounds) a bit weird.

EDIT: If it were viewed in binary, it would make sense.