Recode He-Man

[XadoX]

I would like to recode my He-Man Episodes.
MeGUI's analysed that's partially interlaced.
But I am not realy sure if its right.

Here is a sample (https://www.dropbox.com/s/i4sfrfyyy1r6yg0/intro.m2v?dl=0) of the intro.
Here is a sample (https://www.dropbox.com/s/2pwh7ihhpnjccpe/scene.m2v?dl=0) of the middle of one episode.

Maybe someone could lend me a hand finding the right deinterlace filter for this content?

[LemMotlow]

I would like to recode my He-Man Episodes.Why? Other than the fact that they've been through some careless conversions and processing, what else is wrong with them?

Both samples are progressive content encoded as interlaced. It has been speeded up from the original film speed of 23.976 or 24 fps to PAL 25fps. Sloppy processing has blurred some lines and details. I doubt there's much you can do about it.

[SaurusX]

I haven't downloaded the samples, but if these are the NTSC discs then I believe they've undergone an NTSC->PAL->NTSC conversion. You have to deinterlace and then use Srestore(frate=23.976). Or take the easy way of AnimeIVTC(mode=2,bbob=4,normconv=true).

[Sharc]

I haven't downloaded the samples, .....
Please do. It's usually pointless to give advice based on wild speculations when samples are available....

[SaurusX]

Please do. It's usually pointless to give advice based on wild speculations when samples are available....

In this case my wild speculation was incorrect. However, I think the advice is still good for the NTSC discs. :)

[XadoX]

Why? ... what else is wrong with them?
Size.

[LemMotlow]

Size.Oh, I get it. This is another one of those "how can I re-encode my lossy video and reduce the file size by using a lower bitrate but still keep the quality of a poor quality original?" Answer: you can't maintain the original quality, period. Re-encode at a lower bitrate. Don't come back complaining that your new encode looks like crap.

There isn't anything to deinterlace. According to your samples the video is already physically progressive. Just re-encode at a lower bitrate, or reduce the frame size to 704x576 and re-encode for a new DVD at a lower bitrate. Or reduce the frame size to 640x480 and re-encode at a lower bitrate (it will be much smaller afterward, and will no longer be valid for DVD disc, but you probably don't want a disc anyway). Instead of re-encoding as MPEG, make the frame size smaller and re-encode at a low bitrate with h.264. That should look pretty horrible by the time you're finished with it, but you'll definitely have a smaller file.

Or save yourself some time and trouble and get it off YouTube at a much smaller size.

[XadoX]

@LemMotlow do you had some kind of a bad day?

Nevermind like you discovered before:
...Both samples are progressive content encoded as interlaced. It has been speeded up from the original film speed of 23.976 or 24 fps to PAL 25fps. Sloppy processing has blurred some lines and details. I doubt there's much you can do about it.
So this should help me. Maybe I have to try srestore again.

Edit: While using CRF19 I already reduce filesize by half.

[LemMotlow]

No, not a bad day. Just more low-quality garbage coming at us again and another question about how to make garbage look worse. It gets annoying.

Run QTGMC, then run sRestore. You'll end up with the same thing you started with. The video is already progressive, so every frame is unique. There are no duplicate frames to remove.

[XadoX]

But you are right, maybe I just remux the DVD files to seperate MKV or MP4 files.
I could use the saved time for watching those episodes, finaly :)

[FranceBB]

Actually, you could downscale, denoise and deband in order to make compression at lower bitrate easier for your preferred codec (MPEG-2? H.264? HEVC?).



FFmpegSource2("scene.m2v", fpsnum=25000, fpsden=1000, seekmode=1, rffmode=0, colorspace="YV12")

Dither_convert_8_to_16()
ly = debicubicy(704,576,lsb_inout=true)
lu = utoy().dither_resize16(704,576,kernel="spline64",invks=true,invkstaps=3,src_left=0.25,u=1,v=1)
lv = vtoy().dither_resize16(704,576,kernel="spline64",invks=true,invkstaps=3,src_left=0.25,u=1,v=1)
ytouv(lu,lv,ly)
dfttest(sigma=64, tbsize=1, lsb_in=true, lsb=true, Y=true, U=true, V=true, opt=3, dither=0)
f3kdb(range=15, Y=80, Cb=60, Cr=60, grainY=0, grainC=0, keep_tv_range=True, input_depth=16, output_depth=8)



https://www.gtothegreatsite.it/Personale%20di%20FranceBB/He-man.PNG

[GMJCZP]

FranceBB, excuse my ignorance, is it better to do denoising at 16 bit than at 8?

[FranceBB]

FranceBB, excuse my ignorance, is it better to do denoising at 16 bit than at 8?

Well, yes, not for the denoise operation itself, but because denoise generally leads to banding: if you denoise at 16bit, you'll reduce the risk to introduce banding.

[GMJCZP]

It was what I suspected within my limited knowledge, thank you for responding so quickly :D

PD: by the way, subtle looks spectacular your contribution.

[LemMotlow]

Actually, you could downscale, denoise and deband in order to make compression at lower bitrate easier for your preferred codec (MPEG-2? H.264? HEVC?).



FFmpegSource2("scene.m2v", fpsnum=25000, fpsden=1000, seekmode=1, rffmode=0, colorspace="YV12")

Dither_convert_8_to_16()
ly = debicubicy(704,576,lsb_inout=true)
lu = utoy().dither_resize16(704,576,kernel="spline64",invks=true,invkstaps=3,src_left=0.25,u=1,v=1)
lv = vtoy().dither_resize16(704,576,kernel="spline64",invks=true,invkstaps=3,src_left=0.25,u=1,v=1)
ytouv(lu,lv,ly)
dfttest(sigma=64, tbsize=1, lsb_in=true, lsb=true, Y=true, U=true, V=true, opt=3, dither=0)
f3kdb(range=15, Y=80, Cb=60, Cr=60, grainY=0, grainC=0, keep_tv_range=True, input_depth=16, output_depth=8)




Cool. People put a lot of time, research, and even lifestyle alterations into this sort of thing. Would you consider that more processing and more encoding of flawed over-processed and re-encoded originals = better?

Or would Garbage In = Garbage Out still apply?

[GMJCZP]

I was analyzing the values of dfttest a bit more and it seems incredible to me that with a value of sigma = 64 practically not lost details of the image, how can that be possible given the script?

[SaurusX]

Cool. People put a lot of time, research, and even lifestyle alterations into this sort of thing. Would you consider that more processing and more encoding of flawed over-processed and re-encoded originals = better?

Or would Garbage In = Garbage Out still apply?

Since when does GIGO apply to video processing?

[Overdrive80]

Actually, you could downscale, denoise and deband in order to make compression at lower bitrate easier for your preferred codec (MPEG-2? H.264? HEVC?).



FFmpegSource2("scene.m2v", fpsnum=25000, fpsden=1000, seekmode=1, rffmode=0, colorspace="YV12")

Dither_convert_8_to_16()
ly = debicubicy(704,576,lsb_inout=true)
lu = utoy().dither_resize16(704,576,kernel="spline64",invks=true,invkstaps=3,src_left=0.25,u=1,v=1)
lv = vtoy().dither_resize16(704,576,kernel="spline64",invks=true,invkstaps=3,src_left=0.25,u=1,v=1)
ytouv(lu,lv,ly)
dfttest(sigma=64, tbsize=1, lsb_in=true, lsb=true, Y=true, U=true, V=true, opt=3, dither=0)
f3kdb(range=15, Y=80, Cb=60, Cr=60, grainY=0, grainC=0, keep_tv_range=True, input_depth=16, output_depth=8)



https://www.gtothegreatsite.it/Personale%20di%20FranceBB/He-man.PNG

Overdenoised, maybe you should use masking on dark/mid areas.

[GMJCZP]

Overdenoised, maybe you should use masking on dark/mid areas.

So what do you suggest as far as masking? Well you have to admit that the value of sigma = 64 in dfttest is somewhat intimidating.