Explain interlacing and deinterlacing?

Hey guys, just purchased a new Sony HDR-FX1 camera yesterday and will be getting it tomorrow. All the footage has seemed awsome until i saw a recent couple of clips (from supposodley the pre-release version of the camera) with streaky lines when something moves (camera pan, person walking etc..) there .m2t files and im using VLC player to view them. I noticed when i set de-interlacing to "blend" instead of "none" i get REALLY good results (no lines anymore, blured like it should in film)

I know this is an interlacing issue so could someone explain it to me more? I know that the camera captures two images and puts them together to get one frame and all but its still very vague.

What programs can de-interlace?

 

Here's a simple explanation.

Regular NTSC video, as in the VHS/DV and television video broadcast, plays at 30 frames per second, but is interlaced, meaning that every single frame is actually a combination of two adjacent frames (dual fields). That's why when you freeze-frame video-based material with something moving, you see a jittering/flickering effect like two blinking frames overlapped, along with whatever motion blur to the movement. This is one of the main things that makes video look like video, as opposed to film wich has a still, one-image frame for each..with just a motion blur for movement...resulting in a smoother, warmer moving image. De-interlacing allows you to make each video frame a still frame by showing one complete field (upper or lower) for each 1/30th of a second...hence helping to take away some of the video look, depending on the original image quality. One of the things that Hi-Def can do is shoot at the same 24fps as film with a still, non-interlaced image for each frame, bringing it closer to film (but you can still tell).

Programs like After Effects and Final Cut Pro can de-interlace...probably a bunch more that I'm not all that familiar with.

 

okay thanks, yea spiff cleared it up with me in MSN messenger as he has the camera. There are different modes each generating different results in the camera.

 

"Magic Bullet" does a pretty good job, as I´ve heard. You can also try "DV Film Maker".

 

Let me try to explain that a little more clearly.

Your TV never actually shows you a complete frame at any given time. Each frame is made up of horizontal scan lines, upper field, and lower field (a total of 480 lines). You are basically seeing every other scan line. Then, in the next 1/30th you see the other set of every other horizontal lines mixed with the previous - and so on.

In this way, your TV can show you a fairly clean motion by only actually showing you half the picture at a time.


Programs like After Effects can deinterlace footage by combining the fields together and blending them as best it can. This is why sometimes video footage looks blurry sometimes. And, in some cases, the audio can go out of synch.

 

yea, last night i played around with it in after effects. Used Video> (something about interlacing, lol already forgot what its called) and played around with the blur on that a bit.

I also seperated the fields in After effects. I realized it is possible to seperate the fields and turn them into 3-d footage (aka red and blue glasses. But camera must either be rotating around something, or moving by filming from a car out the side of the window) It actually does work except when the shot is still and all.

But thanks again, and my camera should be at my house as we speek (stupid last day of school before spring break) w00t!

 

Programs like After Effects can deinterlace footage by combining the fields together and blending them as best it can. This is why sometimes video footage looks blurry sometimes. And, in some cases, the audio can go out of synch.

So what do people generally use to deinterlace footage? From the sound of that AE isn't ideal, with potential blur and loss of sync. I've been looking and looking for info about deinterlacing and most of the info I've found generally pertains to DivX, which I'm not real interested in.

I'm at the point where I want to do some rotoscoping (trying Photoshop to begin with), and exporting the filmstrip is fine, but comes out interlaced in Photoshop. Some websites talk about using the DeinterlaceSmooth filter in VirtualDub. And Dorkman mentioned Magic Bullet on another thread.

So to reiterate - for the people who've done deinterlacing, how did you do it?

 

"I also separated the fields in After effects. I realized it is possible to seperate the fields and turn them into 3-d footage (aka red and blue glasses. But camera must either be rotating around something, or moving by filming from a car out the side of the window) It actually does work except when the shot is still and all."

Actually, you don't need red and blue glasses for that to work. If you have a shot like that, you can watch the normal full-colour footage wearing a regular sunglasses lens over one eye, and you'll see the 3D effect, which was "discovered" by Carl Pulfrich (who never saw it himself, because he was blind in one eye!).

 

So to reiterate - for the people who've done deinterlacing, how did you do it?

Anyone? Anyone at all?

 

Scott_M posted on 5/1/05 3:56pm

---

[i]So to reiterate - for the people who've done deinterlacing, how did you do it? [/i]

Anyone? Anyone at all?


[hr][/blockquote]

So what is it you're looking for? What programs can deinterlace? Answer: ALL OF THEM. ANY editing, effects, or compositing program on the planet can deinterlace footage. Just pick one and go, it makes no difference. If you really have to find one that works better than the others, then TRY them. Programs have demos for a reason. Just go out and do it.. I, for one, ain't gonna go download a dozen different demos just to try out deinterlacing just for you, bub.

 

This all explains alot for me

 

I was asking to share your experience. What's the best way you've found to do it?

Sadly it seems that was too much to ask.

 

I use After Effects.

 

60i footage shows forward movement in time every 1/60th of a second. Since every "frame" of a field only has half the vertical resolution AND is a different moment in time, you can't really combine the fields back together to form a frame that truly represents what a 30fps progressive captured frame would look like.

When you blend the fields of 60i footage it will create some minimal "ghosting" but it still seems to be the best way to "deinterlace" it.

I used to drop a field and then stretch it to it's proper vertical resolution to deinterlace but that doesn't look so good. Blending fields just looks better IMHO, not only for clarity but for movement smoothness.

That said, Adobe After Effects and VirtualDub both blend fields just fine. Looks exactly the same. It's a little easier to setup in VDub. Hope that helps.

 

I use After Effects.

Thank you.

That said, Adobe After Effects and VirtualDub both blend fields just fine. Looks exactly the same. It's a little easier to setup in VDub. Hope that helps.

Thanks. I've had a bit of a look at VirtualDub, but it's had issues with Type I and Type II DV-AVI's. It loads Type I without sound and doesn't like the stuff I've converted to Type II. Might try a different converter and see if that makes any difference.

 

Yeah, VDub has issues with DV codecs because of copywrite issues involving Microsoft and their codecs. The only way I can use DV AVIs with Vdub is to use it on my computer that has a DV capture card (Pinnacle DV500) in it. It installs the Pinnacle DV codec. Not sure what other DV codecs will work in VDub now on that particular computer.

 

I dunno if the audio will really be an issue at this point. I'm sure I'll be able to find a way to re-sync it if I need to. Thanks again for the tips.

 

Actually, very few editing programs deinterlace, 2016. A lot will reduce interlacing evidence by blurring a pixel up and down, but you're not deinterlacing, you're just making it less obvious.

 

The process you describe IS deinterlacing. Even the most expensive deinterlacing plugins in the world do just that.. they scrap one field and stretch the other vertically to "fill in the blanks" left by the missing field. THAT'S deinterlacing. The expensive stuff just has more little options and tweaks and stuff.

 

Not to be argumentative, but that's not what I've heard from several sources, including the first hit when you ask Google what deinterlacing means.

http://www.google.com/search?sourceid=navclient&ie=UTF-8&rls=RNWG,RNWG:2005-04,RNWG:en&q=define%3A+deinterlacing

 

Apparently, that's where we differ... I don't listen to "what I've heard".. I pay much closer attention to what I've seen with my own eyes when working with video. I've worked with just about every type of video interlacing there is.. 60i, 30p, 24p (3:2), 24p (2:3:3:2)... after a while, you get a good sense of exactly what's happening.. you learn the WHY. Then you don't need "definitions" anymore.

 

I keep forgetting why I should trust your opinion, though now that you've told me I should, I know.

Please explain to me what it is that you've done that makes you the supreme holy god of fanfilms, because while I do respect you, it has yet to be seen why I should.

 

What? All I did was give a description of deinterlacing, and now I'm some kind of "fanfilm god"? Did I miss a memo or something?

 

Read a book on compositing video, Fig. Fields have half the vertical resolution of a full frame. To deinterlace, usually one field is discarded entirely and the other field interpolated vertically to fill the full space. Some programs, like Commotion, actually separated them into discrete frames, giving, say, a 60fps video stream -- with all frames interpolated vertically to full 480 resolution.

There's a difference between theory and practice. Google is mostly theory, especially when the hit you choose is wikipedia definitions. People who actually work with this stuff is practice. Between the two, trust the practice.

M. Scott

 

Discarding a field and then stretching 640x240 to 640x480 look like crap IMHO. You can see "jaggies" when you do it that way. Also you don't have as true a motion capture doing it that way (when trying to get the 30fps progressive look) as you would get when blending fields. Just my opinion.