Rendering noninterlaced vs. GL2's Frame Mode

[David Ennis]

The GL2 has a frame mode, which I understand is not true 30p, although I don't understand the distinction. Anyway, if I shoot with the GL2 in normal (30i) mode and then render noninterlaced, is that the result the same as shooting in frame mode then rendering (and it does not matter whether interlaced or noninterlaced is chosen)?

[Rob Lohman]

It does matter whether interlaced or non-interlaced is chosen (what the result
of this is depends on your NLE => edit software).

Let us first define the difference between interlaced (which was built because
the electronics back when TV was invented couldn't work fast enough) and
non-interlaced.

The ONLY difference between interlaced and non-interlaced is that in interlaced
the frame is split in two and each half is recorded at a different time interval
(1/60th of a second apart in NTSC).

The two half are "interlaced", so you get:

-- line 1, half 1
-- line 1, half 2
-- line 2, half 1
-- line 2, half 2
etc.

So the lines on half 2 will be recorded ("seen") 1/60th of a second later. If
you play this back on a non-interlaced/progressive screen (like on a computer)
you will see jagging lines on movement (either the whole camera or a subject
in frame), you see the time displacement.

Usually people associate progressive/frame mode with a more filmic look (since
film is exposed one full frame at a time, not interlaced like TV [camera's]).

As you know, the frame mode on the Canon XL1(S) & GL1/2 is a faux progressive
(ie, non-interlaced) mode. What does this mean? It means that the CCD's
are still read out in two half frames 1/60th a second apart. However, what
they (kinda, the exact procedure is not known) do is only use one of these
halves and due some special algorithms and extra data interpolate this to a
full signal.

Yes, with that you will loose "apparant" (apparant, because the signal is still
plain 720 x 480 [for NTSC] DV, hence the interpolation) resolution, but it is
not 50% resolution you loose (since they can do some clever things with CCD
timings and whatnot).

Therefor on the Canon line of camera's I've always adviced to use the frame
mode if you want a progressive look instead of doing a de-interlace in post
(when editing). The post version cannot use the extra information from the
CCD's (pixelshift amongst one) since that information is no longer available
at that point in time. In my opinion the frame mode looks great and saves
you a lot of time in doing a de-interlace in post as well.

I've owned the XL1S and I've only shot in frame mode (except for some tests
every now and then).

So what is the difference between this frame mode and true progressive? As
indicated the frame mode still works with interlaced footage albeit with some
extra tricks. A truely progressive camera (like the XL2 in 24p or 30p mode)
will get all the lines from the CCD's in one go without any time difference at
all. This results in a truely progressive frame (like a single frame exposure on
film), like a (digital) still camera.

Hope that explained it a bit. Usually it is best to shoot in the mode that you
can keep through your entire workflow. So if you shoot something that is going
to be broadcasted at a TV station then shoot interlaced. Shooting weddings
might be done interlaced depending if the couple want a more realistic or
more "fictional/glossy" look for example.

Most people doing fictional movie work shoot in progressive/frame mode.

As always make sure you keep everything in your workflow in the same mode.
So if you shot frame/progressive make sure your NLE (edit software) is set
to progressive (or none for interlaced) as well, including the settings when
doing an export to the final format!

Hope that made some sense to you. Good luck!

[David Ennis]

Thanks, Rob. I did know what interlaced and non-interlaced were, but I did not know the distinction between Canon's frame mode and true noninterlaced (30p). You cleared that up.

But it gets cloudy again in the permutations. We have frame mode vs. 30i mode. Then we have rendering 30i vs. 30p. Then we have displaying on a non interlaced monitor vs. an interlaced TV screen. So it seems that the event we taped can wind up being viewed in eight different versions. I wonder how different they are?

Is frame mode about the same in all those permutations since the even fields are basically replicas of the odd fields?

Does rendering from 30i to 30p give true progressive? Does it even matter in any way (i.e., with respect to resolution or flicker) when viewed on an interlaced TV set?

I suppose I'll just have to test for myself.

[DJ Kinney]

That's what I said: "I'll just have to test for myself."

CUT TO: Six freakin' months later...

I'm shooting a documentary right now that may someday find itself output to film. It's been a never-ending quest to find, not only the filmlook, but the best filmic flicker that maintains clarity. The truth is, I have discovered that some subjects that look great shot in 30p mode, end up looking like muddy messes shot in 60i and deinterlaced and run through 2332 pulldown. Sometimes that deinterlace/pulldown workflow looks good.

It's enough to drive me mad. So best of luck to you. Also, you'll find that there are somehow more than eight permutations. Once you start playing, they are infinite, and there appears to be no single solution.

DJ

[David Ennis]

Originally Posted by DJ Kinney

...Also, you'll find that there are somehow more than eight permutations. Once you start playing, they are infinite, and there appears to be no single solution...DJ

Right. Now that you mention it, I guess there are eight permutations only when the image is absolutely unmoving and the light is absolutely unchanging. Even a very slowly moving subject of solid even color can be rapidly changing the character of the light it reflects to the camera in subtle and not so subtle ways. 30p and 60i will see that differently.

In the extreme, if a white strobe is operating at 60 Hz and in synch with the camera, 30p will have a screen white at the top and black at the bottom, while 60i has a fully lit but dim screen. If you slow the strobe down, 30p will show the white half moving down the screen but 60i will show the same motionless dim screen. Deinterlacing the 60i should show the movement, but less vividly I suspect, because there is probably some light bleed between the rows of pixels.

Or, if you alternately illuminate the top half and bottom half of the CCD with white light at 1/60 of a second, 30p will record a bright white screen and 60i will record a dim screen. Yikes!

The above model has to apply in varying degrees to all movement and illumination shifts. The filmic look must involve not only flicker of the whole screen, but shimmer within the screen. 60i misses or distorts the shimmer of light that 30p captures, and deinterlacing most likely doesn't restore it very well.

What effect interlaced vs. noninterlaced displays has on all that is too murky to think about. But I certainly want a cam with a true progressive shutter now.

[Don Donatello]

"if I shoot with the GL2 in normal (30i) mode and then render noninterlaced, is that the result the same as shooting in frame mode"

shooting interlace then rendering pregressive will give you sort of a progressive frame so it is sort of like frame mode .. you'll probably have to use "blend" as deinterlace method for rendering progressive .. the rendered progressve image will be softer then the interlace clip and not sure if it would be any better /worse then if you had shot frame mode in 1st place ??? but it should cut in with the frame mode clips .... IMO forget trying to get frame mode or a rendered progressive ( from interlace) to 24p unless you use it as an effect ...

60i (30i to some) rendered out to progressive is not "true progressive " but it is progressive "like" ....

[Rob Lohman]

Yes, there is a difference in true progressive since it captures the full frame
in 1/30th of a second instead of two halves in 1/60th of a second that are
1/30th of a second apart. No de-interlacing can ever reverse that without
dropping quality, and then it might not even give the desired effect (Canon's
frame mode is usually found to be better than de-interlacing).

However, you can watch the frame mode/progressive footage on an interlaced
system it basically will look the same as on a progressive system.

In the case of interlaced it will look different on a progressive system than
on an interlaced system (unless the device does a de-interlace as a lot of
DVD players can do these days [quality differs here as well]).

Personally for any fictional work I would work in progressive/frame mode
(as the "Hollywood" DVD's are made as well)

[Don Meers]

So Rob,if i was making a music clip for someone using an xl2/vegas and was after the film look,you would say shoot progressive and render progressive even though it was to be viewed on tv.

[Rob Lohman]

Yes, absolutely!

[Dennis Vogel]

Originally Posted by Fred Retread

The GL2 has a frame mode, which I understand is not true 30p, although I don't understand the distinction.

Fred, I'm not sure this article adds anything to the excellent info that has been presented already. Nor do I know if this is the same way the GL2 does frame mode but here is a pretty good explanation how frame mode is done in some cams: ftp://ftp.panasonic.com/pub/Panasoni...ressive-WP.pdf.

Good luck.

Dennis