QTGMC vs TFM for Progressive Source

[TheLastOfThem]

Hi Everyone,

Been a bit confused but I have a source that has a lot of combing (because the fields are from different moments) and it's in no discernible pattern (3:2, etc). What would be the right step in removing the combs from this file? Run it through TFM() - which seems to do a good job - or QTGMC?

I heard TFM() is for progressive material and de-interlacers are for materials that have fields source from different moments...but the whole thing doesn't make sense to me. TFM() can recover progressive frames from combing - which obviously implies different field moments..

The source is also interlaced (not progressive)...so not sure when people say "this is for progressive material" they mean before the source was put onto a DVD?

Just a lot of confusion as to this thing

[TheLastOfThem]

From VIT - creator of QTGMC -

The script makes no attempt to detect combing. It runs a temporal binomial smooth on every frame to remove bob-shimmer when deinterlacing. This same process is also effective in remove deinterlacing residue from progressive video. But it isn't selective on which frames to process.

I imagine this can be very bad if TFM() can make your video look progressive albeit with duplicate frames and no combing?

[TheLastOfThem]

In the end I can't very much tell the difference between QTGMC (Slower) and TFM()....they look nearly identical for this source I worked with though I'm sure technically it may have been a wrong move to do QTGMC ....thoughts anyone?

[TheLastOfThem]

WITH QTFMC Slower



WITHOUT QTGMC (TFM ONLY)

[manono]

Instead of talking to yourself, you might post 10 seconds from your source. No offense intended. One difference you'll get is 23.976fps from a full IVTC (TFM + TDecimate) and 59.94fps from QTGMC.

And you can always edit your posts to add more details, rather than making additional posts.

[TheLastOfThem]

Instead of talking to yourself, you might post 10 seconds from your source. No offense intended. One difference you'll get is 23.976fps from a full IVTC (TFM + TDecimate) and 59.94fps from QTGMC.

And you can always edit your posts to add more details, rather than making additional posts.

Oh drats. You're right! Sorry. I wanted to mention that when one does TIVTC - Isn't that incorrect to slowdown PAL to NTSC? They just speed up footage by 4% so wouldn't it make more sense to just AssumeFPS(23.976) on something that was progressive source to begin with?

I notice doing TIVTC works...but then you end up keeping the same audio and just altering pitch instead of pitch and also tempo to the correct NTSC broadcast of a show...

[manono]

If your source is PAL then, yes, you wouldn't do a full IVTC and sometimes TFM alone works to realign the fields. Sometimes you might also need to remove 1 of every 25 frames to restore it to 24fps. In any event, no one can (or should have to) guess. Sample please. One shouldn't have to ask for a sample either. When asking about how to handle a video it's standard practice to include a short sample

[hello_hello]

I heard TFM() is for progressive material and de-interlacers are for materials that have fields source from different moments...but the whole thing doesn't make sense to me. TFM() can recover progressive frames from combing - which obviously implies different field moments..

Not quite. When TFM recovers progressive frames from PAL sources that appear purely interlaced it's not because the fields of each frame are different moments in time, it's because the fields were encoded out of phase. Instead of the top and bottom fields of a frame being together, the bottom field from one frame is paired with the top field of the next.... or something like that. The fields of each frame are still the same moment in time though, and for that type of source, TFM() just puts them back together.

Don't forget TFM can de-interlace and does so by default when it detects combing (in order to repair any combed frames that might come out of the field matching process), but if that was happening you'd probably know about it. I assume it's not detecting any combing after field matching so that's all it's doing in this case, but you can make sure it doesn't de-interlace by using TFM(pp=0) while you're diagnosing.

[Skiller]

If TFM(pp=0) doesn't do the job here it's certainly a normconversion with blended fields (I bet it is, cartoons from the 90's always are).

In such case, there are 2 options besides leaving it interlaced:

1) SRestore

2) Bob to 50 fps and encode it like that. Use QTGMC for the bob-deinterlace.