The Joy of Interlacing from Videopia on Vimeo.

As it happens, interlacing is not a diabolical technology invented just to make your life miserable by creating those annoying Venetian Blind patterns on digital videos. (Who knew?)

The wonderful people of Videopia don’t just explain interlacing – they defend it, starting with its early history. Then they explain how to deal with removing interlacing in the progressive-scan world of Internet distribution. And if you’re still not clear on when that horizontal pattern of lines on your video is a good thing and when it’s a bad thing, this will make it clear.

It’s by far the single best explanation I’ve seen, and they’ve done it all with fantastic production values.

Now if people will just watch the darned thing, maybe we won’t see all this poor deinterlacing in online videos on YouTube. (Stats were surprisingly low when this came online, so have at it, Visualist Nation, and spread the love around!)

Any further tips (or questions) to add to their interlacing advice, ye tech-savvy visualists? Let us know in comments.

Lots more smart advice at Videopia. Via Jamie Wilkinson’s FriendFeed

Updated: Richard Lainhart writes with a still-better technique. I agree, absolutely – got so distracted by the elegant explanation of interlacing itself and its history that I neglected to pay as much attention to what they were actually suggesting! Of course, this won’t work in all cases, meaning you’re back to the video technique. But since a lot of you have cameras capable of shooting as Richard describes, this could be helpful.

The deinterlacing techniques mentioned in the video all will introduce artifacts of some sort in the image. If you use leave the fields in, you’ll still see interlace combing on the edges of objects in motion, even if the frame isn’t paused. Interpolated interlacing can be better, but you’ll still often see blockiness, sawtooth effects, or other such artifacting on straight lines and hard-edged objects, as no interpolation method is perfect.

If you can, you’ll get better results with this method – shoot everything in full 1080i HDV, and reduce the frame to one-half resolution in After Effects. When you bring the HDV footage into AE, convert it to square pixels but tell AE to not deinterlace it (in the Interpret Footage dialog.) Then scale that image to half-size in a 960×540 comp. This has the effect of throwing out every other field and reducing the frame to widescreen SD format, and you’ll get perfect, clear progressive full frames. From there, crop to 4:3 for standard SD, or scale up to 1280×720 for 720P HD – scaling the image up in AE will introduce some softness, but it will still look better than 720i footage when viewed on a computer screen.

All the footage on my YouTube site was processed this way, and none of it has any visible field artifacting.

http://www.youtube.com/rlainhart

  • http://www.otownmedia.com Richard Lainhart

    The deinterlacing techniques mentioned in the video all will introduce artifacts of some sort in the image. If you use leave the fields in, you'll still see interlace combing on the edges of objects in motion, even if the frame isn't paused. Interpolated interlacing can be better, but you'll still often see blockiness, sawtooth effects, or other such artifacting on straight lines and hard-edged objects, as no interpolation method is perfect.

    If you can, you'll get better results with this method – shoot everything in full 1080i HDV, and reduce the frame to one-half resolution in After Effects. When you bring the HDV footage into AE, convert it to square pixels but tell AE to not deinterlace it (in the Interpret Footage dialog.) Then scale that image to half-size in a 960×540 comp. This has the effect of throwing out every other field and reducing the frame to widescreen SD format, and you'll get perfect, clear progressive full frames. From there, crop to 4:3 for standard SD, or scale up to 1280×720 for 720P HD – scaling the image up in AE will introduce some softness, but it will still look better than 720i footage when viewed on a computer screen.

    All the footage on my YouTube site was processed this way, and none of it has any visible field artifacting.
    http://www.youtube.com/rlainhart

  • http://www.otownmedia.com Richard Lainhart

    Or, of course, the best method of all – shoot progressive scan frames in the first place if your camera supports it, as more and more do these days.

  • Pingback: aefreemart : The Joy of Interlacing

  • Pingback: fortschritt.tv — The Joy Of Interlacing

  • http://sleepytom.co.uk sleepytom

    wow how totally underwhelming…

    all of the methods mentioned in that video are basic have serious drawbacks. Richard's method (which IS mentioned in the video – it's point 2 in the deinterlacing methods text graphic)is ok if you don't mind throwing half your resolution away but for people with SD cameras or people aiming at full resolution workflows it is unacceptable.

    More advanced methods for deinterlacing include "Bob + Weave" and "Motion Adaptive" deinterlaceing.

    to find out what these are visit the excellent site at http://www.100fps.com (it goes beyond the basics and doesn't have any offensively blue shirts to put you off the hard facts!)

    for a good motion adaptive deinterlace filter with lots of options get Mike Crash's Smart Deinterlace filter for vegas from http://www.mikecrash.com/modules.php?name=Downloa

    you might also like the deinterlace in the magic bullet suite for after effects

  • http://createdigitalmusic.com Peter Kirn

    Sorry about the comment bug.

    I agree, it's good to go all-out with better deinterlacing techniques. For targeting web video, though, what they're saying as *general* advice (and what Richard is saying) is just fine; the target here is people who just want to get their video up on one of these sites in a fairly low resolution.

    And in defense of the video, I think they first needed to explain the concept to people who may never have seen it before. So basic is okay in some cases. Think of this as the video you can go show those annoying beginners. ;)

  • http://www.otownmedia.com Richard Lainhart

    Just a point of clarification – Method 2 in the video is not the same as my technique. Method 2 maintains the clip at its original resolution by discarding one field and replacing it with the other, which would result in visible stair-stepping on the edges of objects. My method discards one field and scales the resulting clip, resulting in no edge artifacting.

    You're right, of course, that it halves the resolution, and so this technique isn't useful if you need to maintain full res. It's intended only for footage that will be delivered in standard or lower resolutions, which is most computer-playback situations. It's especially useful if you're aiming for lower than SD resolutions, like YouTube and other public video sites, as scaling to less than SD frame sizes is especially likely to produce interlace artifacts.

    For full original-resolution delivery, specialty deinterlacing tools like Magic Bullet or ReVision Effects FieldsKit are definitely the way to go.

  • http://sleepytom.co.uk sleepytom

    yeah there is nothing wrong with basic – but i was expecting the video to end all videos from your write up and the video opening sequence. Sadly they fail to mention any of the more advanced methods which do make your video look a lot nicer – if your aiming for the best quality on Vimeo HD then you need the hardcore methods! But as everyone else has said too their really is no substitute for shooting in native progressive scan.

  • http://sleepytom.co.uk sleepytom

    Richard – i think your wrong their – you method IS the same if you scale the video back up to full height, your suggestion is actually to reduce the horizontal resolution by an equal amount before rescaling it which rather than resulting in no edge artifacting actually results in symmetrical artifacting (which may look better if you are going down to a low resolution output anyway but certainly won't look better if you scale the resulting video backup to it's full frame size (which you do if you play it on a TV for example)

  • http://www.otownmedia.com Richard Lainhart

    No, you misunderstand my method. You start with 1920x1080i footage. (Actually, if it's HDV, it's 1440×1080, but you compensate for that first by interpreting the HDV pixel aspect ratio to square from non-square.) You scale that to 960×540 in AE (exactly half in each dimension) to discard one field. For standard definition video, whether 720×480 or 720×486 (non-square pixels), you crop the 960×540 horizontally to 720×540, which is SD video using square pixels. You then scale this again vertically to 720×480 or 720×486, depending on your delivery format, for standard NTSC digital video delivery. You can also go lower to 640×480 for YouTube and similar sites, or lower yet to 320×240 or any of the other standard multimedia formats.

    The point is, because you've removed the fields first, scaling down from the higher resolution (especially in After Effects, which does it beautifully), doesn't introduce any of the typical field artifacting you see when scaling footage with fields. And, because After Effects scales so well, you can even scale up to 1280×720, and the results will still be better than if you scaled down to 1280×720 from 1920×1080, because you've removed the fields. I wouldn't recommend going any higher than that, because the softness introduced scaling up would become unacceptable, but I think it is acceptable up to 1280.

    But scaling down to standard NTSC resolution is perfectly acceptable, I think, and would certainly give better results than shooting at standard def with fields and removing those fields with interpolation. I know, because I've tested it.

  • Pierre A Aebischer

    Hi,

    Well, even if desired or useful or else, interlacing is rather complex, annoying and very much 20st centuryish.

    Why can't we all move to 1080P, ProRes or uncompressed, and altogether be done with the i for good, then with that we'd still have all the fun of messing, mixing and interpoling between 50, 60 and 24 frame rates…

    … and I am not mentioning (well I am!) the ugly HDV and the likes codecs with 1440 pixel wide, so as to make sure you get horizontal AND vertical artifacts, sort of the cherry on the cake of the video industry :-)

  • http://abstrakt.vade.info vade

    Just use motion adaptive de-interlacing and move on…

  • http://videopia.org Videopia

    Thanks for posting this and thanks for the comments. My goal was to explain interlacing in 5 minutes and give people a few general one-click solutions and I entirely agree with the discussion here that there are specific situations and methods that advanced users can use to get better results, albeit with more time, effort and expert knowledge. – Eric

  • Anil Kumar

    I have read your article and enjoyed it a lot since my current job is to scale the Interlaced video.

    I have one software libraray which scale up or down based on the bi cubic interpolation for progressive.
    My job is tune the scaling for the interlace video such that for bottom field.

    Please let me know what should be my though process in order to overcome this situation.