avisynth error: Cannot convert if width is not mod2?

[marcorocchini]

Hi, the cats
but

FFImageSource("C:\Users\Administrator\Desktop\Nuova cartella (7)\1_E.E.-Hiolle_Narciso.jpg")

ConvertToYUY2(interlaced=false) 

colorMatrix(mode="Rec.601->Rec.709")

ColorYUV(levels="PC->TV")

[Attachment 86327 - Click to enlarge]

what can it be?

[jagabo]

4:2:2 YUV video must be mod2 in width. So crop a pixel from the width, or add one.

[marcorocchini]

I'm a poor cat.


In what way I have to crop a pixel from the width, or add one ?

[jagabo]

Use Crop() or AddBorders().

[krykmoon]

I modified your script so it should crop 1 pixel from the left of an image IF its width is not mod2 (making it mod2)

clip=FFImageSource("C:\Users\Administrator\Desktop\Nuova cartella (7)\1_E.E.-Hiolle_Narciso.jpg")

cropV=int(Fmod(clip.Width,2))

clip=clip.crop(cropV,0,0,0)

clip=clip.ConvertToYUY2(interlaced=false) 

clip=clip.colorMatrix(mode="Rec.601->Rec.709")

clip.ColorYUV(levels="PC->TV")

[marcorocchini]

clip=FFImageSource("v:\SMPTE bars .jpg")

cropV=int(Fmod(clip.Width,2))

clip=clip.crop(cropV,0,0,0)


return (clip)


#converttoYV24() #convert to planar
converttoYV16() #convert to planar
z_convertformat(interlaced=false,colorspace_op="170m:709:709:l=>709:709:709:l")
ColorYUV(levels="PC->TV")

ah wow
[Attachment 86338 - Click to enlarge]

but can tell to the z_convertformat

in the "z_convertformat(interlaced=false,colorspace_op="1 70m:709:709:l=>709:709:709:l"" line

to do what do the "ColorYUV(levels="PC->TV")" ? thanks

[Sharc]

clip=FFImageSource("v:\SMPTE bars .jpg")

cropV=int(Fmod(clip.Width,2))

clip=clip.crop(cropV,0,0,0)


return (clip)


#converttoYV24() #convert to planar
converttoYV16() #convert to planar
z_convertformat(interlaced=false,colorspace_op="170m:709:709:l=>709:709:709:l")
ColorYUV(levels="PC->TV")

ah wow
[Attachment 86338 - Click to enlarge]

but can tell to the z_convertformat

in the "z_convertformat(interlaced=false,colorspace_op="1 70m:709:709:l=>709:709:709:l"" line

to do what do the "ColorYUV(levels="PC->TV")" ? thanks

See in your other thread how to convert your SMPTE bars .jpg
https://forum.videohelp.com/threads/417936-comparting-frame-to-frame-in-edius-between-...th#post2771234

You don't need colorYUV(levels= ....)

Conversions are described here
http://avisynth.nl/index.php/Avsresize
Scroll down to the Examples.

[marcorocchini]

ww thansk, I have modify to don't execute the line ColorYUV(levels="PC->TV")

[jagabo]

A simple trick to crop one pixel off the width of a clip:

Crop(0,0,width/2*2,height)

That will crop one pixel if the width is odd, do nothing if the width is even.

For conversion to YV12 (YUV 4:2:0) you can do the same with the height.

[marcorocchini]

ok but I wonder:

can ffmpeg do the same thing of avisynth this

z_convertformat(interlaced=false,colorspace_op="170m:709:709:l=>709:709:709:l")

colorspace conversion from 170m to 709?

[jagabo]

https://ffmpeg.org/ffmpeg-filters.html#colormatrix

[marcorocchini]

[Attachment 86369 - Click to enlarge]

I use

ffmpeg.exe -an -y -i "C:\Users\Administrator\Desktop\Nuova cartella (7)\SMPTE bars .jpg" -vf scale=w=-2:h=1080:in_range=tv:out_range=pc,colormatrix=bt470bg:bt709,pad="width=ceil(iw/2)*2:height=ceil(ih/2)*2" -frames:v 1 -update 1 -q:v 1 -huffman 1 -intra_matrix 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8 -color_range 2 v:\ref.jpg

to generate a ref.jpg OK (levels ok)

but If I try to use the same line but with the final output file ref.TGA

I get that: levels of ref.TGA are different to the ref.JPG, even if the commandline if the same


Can Someone explain to the poor cat if is there a way to get the same levels?


[Attachment 86370 - Click to enlarge]

thanks

[marcorocchini]

also have try

ffmpegOCT2024.exe  -an -v 32 -stats -loglevel fatal -y -i "C:\Users\Administrator\Desktop\Nuova cartella (7)\SMPTE bars .jpg" -vf zscale=w=-2:h=1080:f=spline36:r=full:d=ordered:in_range=tv:out_range=pc,colormatrix=bt470bg:bt709 -frames:v 1 -f image2 -update 1 -c:v targa -rle 0 uscitatarga.tga

but the video levels between the previous output is the same

[Sharc]

You change the levels range from TV(limited) to PC(full) or vice versa.

When you do colorspace conversions you should be clear about
1. The colorspace and range of the source
2. How your source filter possibly affects the colorspace and range
3. What exactly you want to do (from->to)
4. Set up the conversion (avisynth, ffmpeg ......) accordingly, knowing what you are doing
5. Watch the result using the correct matrix and range (decoder, renderer).

If any of above is not clear your conversion becomes a surprise.

[marcorocchini]

ffmpeg.exe -an -y -i "SMPTE bars .jpg" -vf scale=-2:1080,colormatrix=bt470bg:bt709 -sws_flags lanczos+accurate_rnd -frames:v 1 -f image2 -update 1 -c:v mjpeg -q:v 1 -huffman 1 -intra_matrix 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8 out.jpg

I get a out.jpg file "raw 420 JFIF from .JPG"

Can I get a 422 or 444 instead of 420?

[Sharc]

Your source "SMPTE bars .jpg" from your other thread is 4:2:0.
If you want 4:2:2 or 4:4:4 you need to upsample it, or use a 4:2:2 or 4:4:4 pixel format right from start.
https://ffmpeg.org/ffmpeg-filters.html#allrgb_002c-allyuv_002c-color_002c-colorchart_0...02c-yuvtestsrc

For 4:2:2 bars, for example

ffmpeg -f lavfi -i smptehdbars=duration=30:rate=25.0 -vf scale=1920:1080,setdar=16/9 -an -c:v utvideo -pix_fmt yuv422p "smpte_HDbars.avi"

or in Avisynth, for 4:4:4

colorbarsHD(1920,1080,"YV24").

(And what's the purpose of using the mjpeg encoder?)

[jagabo]

levels of ref.TGA are different to the ref.JPG, even if the commandline if the same

I get the same levels for both. The problem may be in the way your editor is interpreting them.