Quality on CUDA IVTC and inverse pulldown

tormento · 7th June 2022, 08:24

I have never worked with bad anime sources, so I need your help.

I have bought the BD box of Stand alone complex and it shows up it's both interlaced and pulldown. Sample here.

Good ol' pal FranceBB suggested me the lines

Code:

tfm(mode=1,pp=5,slow=2,micmatching=2,clip2=tdeint(mode=2,type=3))
tdecimate()

and it works fine but I have a very old CPU and gave a try to DG tools with

Code:

DGTelecide(mode=1, pthresh=3.5)
DGDecimate()

and they work really fine and fast too.

My eyes are not trained to get quality issues and my brain not educated to work with old anime material.

Can you help me to tweak and discern?

takla · 7th June 2022, 09:34

You could also try this (ffmpeg only, no AVS)

Code:

ffmpeg -y -benchmark -i INPUT.mkv -c:v ffv1 -vf "fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate" -an OUTPUT.mkv

It will tell you your speed

ReinerSchweinlin · 7th June 2022, 11:59

Quote:

Originally Posted by tormento

I have never worked with bad anime sources, so I need your help.

I have bought the BD box of Stand alone complex and it shows up it's both interlaced and pulldown. Sample here.
...?

The download says 6.6 GB - you might want to consider to only provide a small sample of a few seconds

tormento · 7th June 2022, 12:27

Quote:

Originally Posted by takla

It will tell you your speed

My request was between 2 different methods, not a third one

tormento · 7th June 2022, 12:28

Quote:

Originally Posted by ReinerSchweinlin

The download says 6.6 GB - you might want to consider to only provide a small sample of a few seconds

Here, with a very nice horizontal panning at the very beginning.

tormento · 11th June 2022, 09:24

*bump*

takla · 12th June 2022, 11:47

Everytime I work with interlaced content I'm reminded again why I never bother: Quality won't ever reach that point where I think to myself "That looks good and was worth the time"...

From rainbowing to missaligned fields, it is always the same issues.

7th June 2022, 08:24	#1 \| Link
tormento Acid fr0g Join Date: May 2002 Location: Italy Posts: 3,129	Quality on CUDA IVTC and inverse pulldown I have never worked with bad anime sources, so I need your help. I have bought the BD box of Stand alone complex and it shows up it's both interlaced and pulldown. Sample here. Good ol' pal FranceBB suggested me the lines Code: tfm(mode=1,pp=5,slow=2,micmatching=2,clip2=tdeint(mode=2,type=3)) tdecimate() and it works fine but I have a very old CPU and gave a try to DG tools with Code: DGTelecide(mode=1, pthresh=3.5) DGDecimate() and they work really fine and fast too. My eyes are not trained to get quality issues and my brain not educated to work with old anime material. Can you help me to tweak and discern? __________________ @turment on Telegram

7th June 2022, 09:34	#2 \| Link
takla Registered User Join Date: May 2018 Posts: 242	You could also try this (ffmpeg only, no AVS) Code: ffmpeg -y -benchmark -i INPUT.mkv -c:v ffv1 -vf "fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate" -an OUTPUT.mkv It will tell you your speed Last edited by takla; 7th June 2022 at 11:35.

11th June 2022, 09:24	#6 \| Link
tormento Acid fr0g Join Date: May 2002 Location: Italy Posts: 3,129	bump __________________ @turment on Telegram

12th June 2022, 11:47	#7 \| Link
takla Registered User Join Date: May 2018 Posts: 242	Everytime I work with interlaced content I'm reminded again why I never bother: Quality won't ever reach that point where I think to myself "That looks good and was worth the time"... From rainbowing to missaligned fields, it is always the same issues.