PLEASE can you transcode in Avisynth this 2 monster because...

[marcorocchini]

because I'm a *** and I'm not able


Monster 1

https://www.dropbox.com/s/irzeackb2xrpwr8/NTSCmonster.MXF?dl=0

Video
ID : 2
Format : MPEG Video
Format version : Version 2
Format profile : 4:2:2@High
Format settings, BVOP : Yes
Format settings, Matrix : Custom
Format settings, GOP : M=3, N=15
Format_Settings_Wrapping : Frame
Codec ID : 0D01030102046001-0E06410201040301
Duration : 32s 299ms
Bit rate : 25.0 Mbps
Width : 1 920 pixels
Height : 540 pixels
Display aspect ratio : 16:9
Frame rate : 29.970 fps
Standard : Component
Color space : YUV
Chroma subsampling : 4:2:2
Bit depth : 8 bits
Scan type : Progressive
Compression mode : Lossy
Bits/(Pixel*Frame) : 0.805
Stream size : 96.3 MiB (89%)
Color primaries : BT.709
Transfer characteristics : BT.709
Matrix coefficients : BT.709


Monster 2

https://www.dropbox.com/s/b3waqvoh2kpaj4i/PALmonster.MXF?dl=0

Video
ID : 2
Format : MPEG Video
Format version : Version 2
Format profile : 4:2:2@High
Format settings, BVOP : Yes
Format settings, Matrix : Custom
Format settings, GOP : M=3, N=12
Format_Settings_Wrapping : Frame
Codec ID : 0D01030102046001-0E06410201040301
Duration : 37s 760ms
Bit rate : 25.0 Mbps
Width : 1 920 pixels
Height : 540 pixels
Display aspect ratio : 16:9
Frame rate : 25.000 fps
Standard : Component
Color space : YUV
Chroma subsampling : 4:2:2
Bit depth : 8 bits
Scan type : Progressive
Compression mode : Lossy
Bits/(Pixel*Frame) : 0.965
Stream size : 113 MiB (90%)
Color primaries : BT.709
Transfer characteristics : BT.709
Matrix coefficients : BT.709


target is 1920x1080 50i
and 720x576 50i

duration of the NTSC ones have to be 16 seconds
duration of the PAL ones have to be 18,5 seconds

Please can you tell me both the script? Thanks a lot

[LMotlow]

How did these videos become distorted and in 1920x540 frame size at 1/2 vertical resolution?

Perhaps I'm missing something from some of your earlier posts about what you intend to do with these videos, but I have to admit that I have no idea what your purpose could be, other than reducing image quality. So I'm afraid I'm not much help here.

[poisondeathray]

Probably they are shot in overcrank mode, during which this camera reduces the resolution

Just don't use it = problem fixed

[Selur]

Where is the problem?

Assuming it's okay to have 4:2:0 output, you could
a. load source (FFVideoSource or LWLibavVideoSource both work fine)
b. adjust frame rate to 25 using SalFps3 (this requires a ConvertToYv12(), since SalFps3 only supports 4:2:0)
c. resize to the resolution you want as output (if you want to fix the PAR issue, here is where you have to resize accordingly)
d. separate fields and weave them
e. encode to the format you want to with the application you want to (remember to set the proper PAR, avisynth has no PAR signaling)

Here's an example for the 10920x1080@50i output:

SetMemoryMax(768)
SetMTMode(5,16) # change MT mode # not needed when input is 25fps PAL
LoadCPlugin("G:\Hybrid\avisynthPlugins\ffms2.dll")
LoadPlugin("G:\Hybrid\avisynthPlugins\Motion_12Dec05.dll")  # not needed when input is 25fps PAL
LoadPlugin("G:\Hybrid\avisynthPlugins\mt_masktools-26.dll") # not needed when input is 25fps PAL
Import("G:\Hybrid\avisynthPlugins\SalFPS3.avs") # not needed when input is 25fps PAL
# loading source
FFVideoSource("H:\NTSCMO~1.MXF",fpsnum=30000,fpsden=1001)
SetMTMode(2) # change MT mode  # not needed when input is 25fps PAL
# to 4:2:0
ConvertToYv12()
# adjust frame rate using SALFps3
PointResize(1920,544) # not needed when input is 25fps PAL
SalFPS3(FPS=25,mode=1,protection=80,protection2=20,iterate=2,reset=50,initialise=6) # not needed when input is 25fps PAL
PointResize(1920,540) # not needed when input is 25fps PAL
# scaling
Spline36Resize(1920,1080) # output par is 1:1 now
# interlace
AssumeTFF().SeparateFields().SelectEvery(4, 0, 3).Weave()
return last

If 4:2:2 or 4:4:4 needs to be preserved, another FrameRateChanger than SalFPS3 has to be used.

@LMotlow: the PAR info of the clip is wrong, it indicates 1:2, but it should indicate 2:1.

Cu Selur

[marcorocchini]

@ LMotlow: yes, as poison, is the camera has overcrack all to produce this
I know I cannot do nothing of this but I try to understand for principles a correct procedure^^,

thanks Selur, please can I ask you:

- what alternative FrameRateChanger can manage 4:2:2 properly?

- this script can also work for the NTSC .mxf file? and do a properly framerate conversion with motion interpolation?

- if I apply some process "not needed" it affects quality? quality can be lower if this process remains? thanks

[Selur]

this script can also work for the NTSC .mxf file? and do a properly framerate conversion with motion interpolation?

The script I used was ment for the NTSC file,... ("H:\NTSCMO~1.MXF")

what alternative FrameRateChanger can manage 4:2:2 properly?

No clue, never had to change the frame rate off 4:2:2 content with avisynth. So either someone else post a suggestion or you have to search for yourself.

[marcorocchini]

but it can work on the PAL one?

[Selur]

You don't need SalFPS3 if your content is PAL.
25p -> 50i is done through the interlacing step.
=> how about looking at the output of the avisynth script and decide if it's what you want or not?

[marcorocchini]

I'm finding all plugins and I try as soon as possible to see results, overall I'm interested to the ntsc

[poisondeathray]

1) You don't want to apply motion interpolation blindly on everything. Too many problems and flaws / artifacts. Very unprofessional results



2) If certain filters only work in 4:2:0, you can preserve 4:2:2 by using UtoY (and VtoY) , Converting to YV12, applying your filter, Converting back to YV16

You don't lose anything for the chroma (it's not down sampled), because the full resoution of each U, V plane is copied to the Y plane for each instance (same resolution as the original chroma channel)

src=last

u_chroma = src.UToY().ConvertToYV12().FILTER().ConvertToYV16()
v_chroma = src.VToY().ConvertToYV12().FILTER().ConvertToYV16()
YToUV(u_chroma, v_chroma)
MergeLuma(src)

#Replace FILTER() with whatever filter you wanted in YV12 4:2:0

[Selur]

hmm,.. so for 4:4:4 you could use Yv24 instead of Yv16, right?

[poisondeathray]

hmm,.. so for 4:4:4 you could use Yv24 instead of Yv16, right?

Yes

And you can verify with subtract() or amplified differences, or any method. There is no loss from that conversion itself

It's the same idea as lossless RGB filtering (with YUV filters)
http://forum.doom9.org/showthread.php?p=1647926

[Selur]

@poisondeathray: still a native 4:2:2 and 4:4:4 FrameRateChanger would be preferable, do you by any change know one?

[poisondeathray]

What do you mean by "frameratechanger" ? Do you mean some interpolation method (new frames generated) ? like interframe, svpflow, mflowfps etc... ?

I really wouldn't bother in his scenario - especially since overcranking is meant for slo motion ; but I don't think there is any native avisynth implementation, you would need that workaround above

[Selur]

Yes, I ment basically all typical filters that are meant for frame rate conversion. (http://avisynth.nl/index.php/External_filters#Frame_Rate_Conversion)

[poisondeathray]

If you look in that link, they basically all say YV12. I suppose you could write a wrapper function to incorporate that workaround above



"frame rate conversion" can be categorized into 3 basic types:

1) duplicating or decimating frames e.g. This would be like ChangeFPS() in avisynth

2) blending frames e.g. This would be like ConvertFPS() or similar filters in avisynth

3) interpolation of generation of "new" inbetween frames i.e. "optical flow" ; svpflow, interframe , mflowfps etc... in avisynth

[Selur]

I suppose you could write a wrapper function to incorporate that workaround above

I know.

[poisondeathray]

But he's trying to do all these "automatically" in batch with scripts.

I would definitely not use approach #3 (optical flow) for that without manually checking. He will get disasterous results in some clips for certain

[Selur]

But he's trying to do all these "automatically" in batch with scripts.

Okay, he didn't mention that anywhere, but that is probably from another thread.

[poisondeathray]

But he's trying to do all these "automatically" in batch with scripts.

Okay, he didn't mention that anywhere, but that is probably from another thread.

It's implied Haven't you seen any "cat" threads ?

[marcorocchini]

YES: present! ^^

[Selur]

And that is probably the last word for that, since all useful stuff has probably been said now.

[LMotlow]

Probably they are shot in overcrank mode, during which this camera reduces the resolution

OK, that answers part of the question. Thanks.

@LMotlow: the PAR info of the clip is wrong, it indicates 1:2, but it should indicate 2:1.

Yeah, I doubled the height early on to get 1920x1080 and figured that one out. Another part of the question answered.

Still haven't figured out where these final output videos are going. I'll just watch and wait.

[marcorocchini]

I wonder: artifacts are impossible to avoiding due to the native resolution reduced?

In practice: what is useful, this "special" mode?

i mean: what utility have the camera in this mode of working?

[jagabo]

nnedi3(dh=true)

[marcorocchini]

Double High the input=true

[jagabo]

Since you missed the point (4x point resize after):