Interlaced or fields out of phase?

[vertical]

Hello everybody,

I would appreciate your advice whether the picture of the movie is interlaced or not. Here is the link to the clip:

http://rapidshare.com/files/6121670/Dragon-Test.demuxed.m2v.html

If it is not interlaced, I believe I can use the following filters in the script to correct it:

AssumeTFF()
Telecide()

If somebody would be so kind to explain how to determine whether the picture is interlaced or it so called the fields out of phase. This would give me a chance not to post the question again.

A few details of the movie just in case: PAL DVD 720 x 576 (4:3).

Thank you!

[jagabo]

Use AVISynth's

Bob()

To perform a Bob deinterlace (you may need to add AssumeTFF() before the Bob() if the source is TFF). Then step through the frames one by one with something like VirtualDUb. If you see pairs of nearly identical frames the video is either not interlaced or simply captured out of phase. If each frame is unique then you have a pure interlaced source.

Your source is out of phase PAL. You can use Telecide() to restore the progressive PAL frames.

[vertical]

jagabo,

Thank you. Though I am not sure I will be able to use the Bob (I have no idea what it is), it gives an idea how to determine what I need.

Thanks again.

[manono]

Though I am not sure I will be able to use the Bob (I have no idea what it is), it gives an idea how to determine what I need.

Bob is an included AviSynth filter. You don't have to load a Plugin, the way you would if you wanted to use LeakKernelBob. It's good for determining what you want to know here. It's no good at all if you really need to use a Bob for something (like for RePAL, or SmartDecimate).

A bobber, either the built in Bob command, or a much better one like TDeint(Mode=1), or LeakKernelBob, bobs a frame. It splits the frame into 2 fields, doubles the framerate, and resizes each field to the original resolution. If, for example, your source is NTSC DVD at 29.97fps, a Bob will create a 720x480 59.94fps video.

http://www.avisynth.org.ru/docs/english/corefilters/bob.htm

[vertical]

manono, thank you!

[edDV]

Deinterlace is an imperfect art. Techniques each have pro/con.

Weave works perfectly for still video but gives the notorious mice teeth when anything moves.

Bob flickers for stills but looks OK for small to moderate motion.

Single field looks good for fast motion but becomes jerky in slow moving areas and shows low vertical resolution for stills.

Fancy deinterlacers apply all three to a frame on a motion block vector/prediction basis. Slow moving areas get a weave, moderate motion blocks get a bob and fast action gets a single field. Motion detection takes time as does blending adjacent motion blocks. The time to process and royalty payment costs for the technology limits the average Joe to choose one at a time for software deinterlacing.

When all this is built into hardware, realtime performance can be achieved. The only problem with realtime is all this picture analysis consumes fields to frames of time delay before the filter can be employed. Frames of video delay are expensive so some shortcuts can be made using motion vector prediction, that is predicting the direction of motion so filters can be applied there on future frames. Motion analysis errors are analyzed and corrected adaptively.

[vertical]

Thank you for the information. It gives a broader picture of what we are trying to achieve and what is possible to achieve, at least for now.

[edDV]

The pros use hardware deinterlacers similar to the "fancy" models described above. The good news is much of this hardware technology is beginning to show up in mainstream HDTV sets, progressive DVD players and in the latest generation computer display cards. As patents expire and chip technology improves, more of this will become available at the low end.

Deinterlace is lossy and irreversible to the same quality*. If you deinterlace in software, you are locked to that current method and quality. If you leave the video interlaced to the DVD, future progressive players, TV sets and display cards will do a better deinterlace job with each technology generation.


*Archive the interlace original for the future.

[vertical]

Thank you! It is really very interesting.

What if my DVD player (Panasonic) is progressive and my TV set is HD (Panasonic plasma HD)? Are they supposed to do the job, I mean deinterlacing?

[edDV]

Thank you! It is really very interesting.

What if my DVD player (Panasonic) is progressive and my TV set is HD (Panasonic plasma HD)? Are they supposed to do the job, I mean deinterlacing?

Any progressive DVD player that plays an interlace 480i DVD and outputs 480p needs to deinterlace. The method used ranges from low rent single field or blend up to motion adaptive weaves and bobs. The major feature to look for is "cinema" mode for inverse telecine when a 2:3 telecine field sequence is detected (i.e. film source) and then converted to progressive. Inverse telecine isn't a deinterlace per se. It just rearranges fields to restore the 23.976 progressive frame sequence. For 480p output, the 23.976 progressive frames are repeated in a 3:2 sequence to 59.94 fps.

LCD and Plasma displays are native progressive and fixed resolution. That means any interlace source needs to be deinterlaced or inverse telecined, then rescaled to the native resolution of the display. It used to be cheap HDTV displays didn't have "cinema" IVTC and used minimal quality deinterlace techniques. Today some higher end techniques are reaching mainstream or even low end displays (e.g. Faroudja dcdi in some low end Westinghouse LCD models).

[vertical]

Thank you!