QTGMC and VFR question about mixed framerate from source

[LaserBones]

I have a DVD series I'm up scaling. The source is 576i 25fps. The video is shot at 25fps however some of the special effects and motion graphics are true interlaced footage so I'd like to keep them at 50fps. Should I use something to like srestore to selectively remove duplicate frames if so wouldn't I also need to convert the file to be VFR since some of the video would be at 25fps and other parts would be at 50fps? Or should I just leave it all double framerate and not worry about duplicate frames?

[johnmeyer]

I don't see the need for SRestore.

Also, your description makes no sense. The first part sounds normal: PAL video DVDs are 25 fps interlaced. However, the sentence "some of the special effects and motion graphics are true interlaced footage so I'd like to keep them at 50fps" doesn't compute. A DVD cannot contain 50 fps material. It can contain 25 fps interlaced, which is 50 fields per second, but it cannot store 50 frames per second.

Finally, I don't understand whether you have duplicate frames in the original material (somewhat unusual for a PAL video DVD) or whether something you are doing is creating duplicates.

So, while I'd like to help, I don't understand what problem you have.

[LaserBones]

You are correct I'm reffering to the 50 fields and letting GTGMC create a frame for each field. Some of the motion graphics in the video use the full 50 fields while the video footage was shot at 25p so it doesn't use the full 50 fields and yields duplicate frames on that footage during QTGMC. In other words the motion graphics that were created for the intro are at 25i while the footage they edited with was 25p.

[johnmeyer]

OK, so it sounds like you have a mix of film and video and that the film portions have been sped up from 24 fps to 25 fps, as is common for PAL.

If you want the result to be 50 fps progressive, then you just use QTGMC to change your 25 fps interlaced to 50 fps progressive. However, I think the 25 fps progressive portions will not end up having duplicate frames, but instead I think QTGMC will do some sort of interpolation on those as well. Someone else can correct me if I'm wrong. Even if QTGMC did simply duplicate the frames from the 25 fps progressive portions, you wouldn't want to delete those duplicates, if your goal is to have a 50 fps progressive end result.

Someone else who has actually handled what I think are called "hybrid" PAL DVDs can probably suggest a

[LaserBones]

I understand what you're referring to and that isn't my question. I apologize for not being clear. I'm not sure what words to use. I've included two clips. One from the intro and one from the middle of the episode. If you run them through QTGMC you'll notice the intro can be "bobbed" when deinterlaced but the main footage of the story is progressive and can't be converted to 50fps without having double frames. https://www.dropbox.com/sh/pnevb8v8iw3zqe8/AAB1HY2d_CiF1nDB1dClktIoa?dl=0

They've done this to keep the motion graphics more fluid. They used to do this a lot with interlaced footage when showing still images as well to keep the motion more fluid on interlaced screens.

[jagabo]

QTGMC will duplicate frames of the 25p sections to make 50p. But the duplicates aren't exact. Some of the edge smoothing algorithms will smooth with the frames before and after. But the near duplicate frames consume very little bitrate -- not worth worrying about.

If you really want exact duplicates you could use a more complex script that detects comb artifacts and only uses QTGMC on those frame, otherwise simple duplicates with ChangeFPS(50):

v1 = LWLibavVideoSource("valentina - title_t03.mkv") 
v2 = LWLibavVideoSource("valentina - title_t03 - footage.mkv") 
v1+v2
ColorYUV(gain_y=-25, off_y=3)

q = QTGMC().Subtitle("QTGMC") # for debugging purposes, remove subtitle for production
d = ChangeFPS(50)

ConditionalFilter(d, q, d, "IsCombedTIVTC(cthresh=5)") # requires TIVTC package

And it's very common for film based TV shows to have interlaced titles overlaid onto them as video. Or even fully interlaced video during the intro or closing credits.

[LaserBones]

Thanks Jagabo!

[Luke M]

Also, your description makes no sense. The first part sounds normal: PAL video DVDs are 25 fps interlaced. However, the sentence "some of the special effects and motion graphics are true interlaced footage so I'd like to keep them at 50fps" doesn't compute. A DVD cannot contain 50 fps material. It can contain 25 fps interlaced, which is 50 fields per second, but it cannot store 50 frames per second.

That was a correct way of speaking in CRT days, but not now. Since all displays are noninterlaced, "fields" are merely an intermediate step in the decompression; the final output is always frames.

[Sharc]

Also, your description makes no sense. The first part sounds normal: PAL video DVDs are 25 fps interlaced. However, the sentence "some of the special effects and motion graphics are true interlaced footage so I'd like to keep them at 50fps" doesn't compute. A DVD cannot contain 50 fps material. It can contain 25 fps interlaced, which is 50 fields per second, but it cannot store 50 frames per second.

That was a correct way of speaking in CRT days, but not now. Since all displays are noninterlaced, "fields" are merely an intermediate step in the decompression; the final output is always frames.

Johnmeyers statement was absolutely correct.

[manono]

Since all displays are noninterlaced, "fields" are merely an intermediate step in the decompression; the final output is always frames.

Even the 25fps progressive frames are doubled to make 50fps. But the subject is the DVD, not how its output is sent to a modern progressive display.

Johnmeyers statement was absolutely correct.

Yep.

[Luke M]

Since all displays are noninterlaced, "fields" are merely an intermediate step in the decompression; the final output is always frames.

Even the 25fps progressive frames are doubled to make 50fps. But the subject is the DVD, not how its output is sent to a modern progressive display.

Since fields are always decoded into frames, it's meaningless to say that DVDs don't "really" store frames. They do store frames, it's just that the frames are compressed using interlacing.

[Cornucopia]

@LukeM, regardless of whether you are dealing with analog or digital, interlaced or progressive, the NOMENCLATURE for framerate still holds true - FPS as 25i or 25p, or 50p, would be correct. 50i would not. Not sure why you were arguing that.

WE here all know that digital media files are frame-based, and that modern displays are all frame-based (though there are still many installations of CRTs in operation), but when dealing with interlaced-captured & compressed footage, we also know one must always honor that and treat it appropriately with kid gloves. That applies even when you consider it to be "intermediate".

Also,johnmeyer never said DVDs don't store frames, he said DVDs cannot store 50p frames.


Scott

[johnmeyer]

Also,johnmeyer never said DVDs don't store frames, he said DVDs cannot store 50p frames.

Yes, that is what I was attempting to say. I guess I wasn't clear enough. Thanks for clarifying.

[LaserBones]

QTGMC will duplicate frames of the 25p sections to make 50p. But the duplicates aren't exact. Some of the edge smoothing algorithms will smooth with the frames before and after. But the near duplicate frames consume very little bitrate -- not worth worrying about.

If you really want exact duplicates you could use a more complex script that detects comb artifacts and only uses QTGMC on those frame, otherwise simple duplicates with ChangeFPS(50):

v1 = LWLibavVideoSource("valentina - title_t03.mkv") 
v2 = LWLibavVideoSource("valentina - title_t03 - footage.mkv") 
v1+v2
ColorYUV(gain_y=-25, off_y=3)

q = QTGMC().Subtitle("QTGMC") # for debugging purposes, remove subtitle for production
d = ChangeFPS(50)

ConditionalFilter(d, q, d, "IsCombedTIVTC(cthresh=5)") # requires TIVTC package

And it's very common for film based TV shows to have interlaced titles overlaid onto them as video. Or even fully interlaced video during the intro or closing credits.

Jagabo, could you please explain how you determined the ColorYUV settings?

[jagabo]

Jagabo, could you please explain how you determined the ColorYUV settings?

The video had some overly bright areas (Y > 235). So I reduced the gain (gain_y=-25). That left some of the dark shades a bit too dark (Y < 16) so I pulled Y up (off_y=3). I used Histogram() to view the levels.