Multimedia
Blogs
Forum
Affordable HD
Whitepapers
Newsletters
Events
Advertisers
DCP Directory
MillimeterQuality assurance
Jul 1, 2001 12:00 PM, By Frank McMahon
Process before you compress, and begin to explore the ways you can improve the look of your web video projects.
Before you compress your finished production into any of several streaming formats, you may want to stop and do some processing on it. Processing optimizes your production and gets it ready for web viewing. There is a dizzying array of choices in most encoder programs, and it's hard to decide what to use and what to leave out. Plus, invariably, each additional process you apply to your clip leads to more rendering time. This month we'll take a look at some of the options you should be aware of after digitizing and before encoding. All these options are worth considering if you want your web video to be as pristine as possible.
One question that may arise: What is the best way to transfer movies shot on film to web video? That's because while the frame rate of NTSC video is 29.97fps, the rate for film is only 24fps. You may have used a telecine machine to incorporate 3:2 pulldown and add in the extra frames when transferring your film material to video, but it's always best to have your encoded video be the same frame rate as the original production.
Some producers do this in a separate compositing program using a plug-in such as DigiEffect's Cine-Motion. However, most encoders are now offering this as a built-in feature. Media 100 Cleaner 5 sports an option called Intelecine that removes the additional frames introduced during the telecine process and reverts back to the 24fps film rate. Pulldown frames should always be removed before encoding to web video for a number of reasons. The extra frames are just dupes padding out to 29.97fps, so they are merely adding more heft and more frames to your web format movie. Not only will the action in your clip be smoother when you go back to the film rate, the process will help eliminate the interlace motion artifacts that are more pronounced in these additional frames.
If you decide you want to go back the other way and apply the telecine process to move the frame rate from film back to video, most encoders let you do this as well.
As far as intellectual property goes, you may want to consider protecting your content by embedding a watermark on the video stream. Some editing systems, such as iFinish or Media 100i, have an overlay feature to superimpose an uncompressed watermark image. If you decide to add a watermark, there are a few rules to go by. First of all, by nature a watermark is small, and since web video can also be diminutive at lower rates, it is important to size it large enough to be noticeable but small enough to be unobtrusive. At the same time, it's important to make sure that this tiny graphic does not fall apart during the encoding process, where heavy compression is applied.
When you create your watermark, you should also add an alpha channel. This will allow you to adjust the transparency of the logo within your encoding program. A logo that is 100% opaque will be a distraction. One set for 50% transparency will be a little easier to take.
The developer version of the Sorenson codec supports an improved form of watermarking, but it is only available when using QuickTime. It can place an uncompressed grayscale image over a movie, residing on the video track, and you can specify the amount of transparency. This method produces no compression artifacts. However, since it does not use an alpha channel, its edges will be hard. Basically, the codec just knocks out the true black in the image. The end result, with properly adjusted opaqueness, will be much more legible and sharp than a standard watermark in web video form.
Sharpen and blur are two common effects that can be applied post-encoding, but their use should be tempered by viewing the master footage to see if the results will be worth the extra rendering. Encoding your video compresses your footage and tracks the changes from frame to frame. If you ever compress older analog legacy productions into web video format, you'll quickly see your encoding times increase and your final web video will be typically larger and include more artifacts because of the extra grain and color noise. Most sharpening filters available in encoders boost the grain a bit and often introduce noise into the picture. This could be troublesome if your footage already contains grain, but it could be beneficial if you have a fairly simple shot (perhaps a talking head against a solid color) and want to boost the detail.
If your encoder offers an unsharp mask, this would be a better option since that command only works with edges and generally produces less grain. Sharpen should be avoided in most cases, and if your movie really needs some work in this area, it is recommended that you go back to your finishing system and process the footage on the desktop. Most NLEs or plug-ins will do a much better job than an encoder will in this respect.
Blur is another process that should be used in special cases. It actually can improve compression because it knocks out a lot of the grain. However, your movie may start to have large areas of solid colors introduced, which could adversely affect the mood of your original footage. The solid color areas will compress better and leave more of the encoding power for your detail areas, but do a test run to see if this is the effect you want. A much better option is adaptive noise reduction, assuming your encoder supports it. This process is like a smarter version of blur and sharpen, and actually manages to do both techniques at once. The procedure keeps the edges sharp while smoothing over some of the noise introduced in large solid areas. So you have the benefit of having better compression and eliminating unwanted noise, as well as keeping the detail in the movie.
Temporal processing is an extension of adaptive noise reduction. Basically, this process performs the same noise reduction but averages it out over time. The advantage of smoothing the noise or grain out over time is that it eliminates pixels that suddenly appear. For example, if your original footage was shot on film and you had several dirt specks or marks on the footage, temporal processing would take care of that and could be used to eliminate most of the unwanted flecks.
Two additional processes that I use periodically are saturation and contrast. Often, footage made for a big-screen monitor that is slapped down to a few-inch web production needs a little bit of a punch to capture that original feel. Sneaking up the contrast and adding a little color saturation can work wonders. However, it is best to experiment and see what the final results will look like. This method introduces minimal extra encoding time, so it may be worth the tradeoff.
One feature that is especially well implemented in the Cleaner 5 encoder is the option to include a static mask. In simple terms, it can mask out items in your video that you don't want to change or compress with every frame. If you have watched web video, you know that if you look closely, lots of pixels are changing from frame to frame — even in areas that are totally solid. You may have also seen highly produced web video where certain solid areas are rock solid, with no pixel change whatsoever. Chances are that the producers incorporated static masks.
As an example, think of a video of a news report that has a reporter with a stationary over-the-shoulder graphic. This graphic will stay the same throughout the clip, but if you encode the video, most encoders will change pixels in this over-the-shoulder image. To incorporate a static map, you would grab a frame of this video and then import it into a graphics program such as Adobe Photoshop. Create a new layer and then draw a black box over the over-the-shoulder graphic. Export it as a mask image and then load it into your encoder. When you encode the final movie, this area will not change at all. The software has been instructed to leave this area untouched and will only encode it once on the first frame, and not compress it from frame to frame.
You can experiment and actually mask out the entire news set, leaving only the reporter and a slight area around the person to be compressed. In theory, this could save encoding time and make for a better-looking web video, but there are a few caveats to keep in mind. First of all, obviously the talent can't lean over into the masked-out area, or else they will be blocked out.
Also, it's important to note that most productions are shot with camera people who are often not locked down on their tripod. The slightest move or change in camera position, and the deal is off. The mask will be stationary, and if the camera moves, the mask obviously won't move with it. So it's best to experiment with this process on a non-critical production before you use it for an important project. If used correctly, it could save rendering time and give your web video a sharper, more solid look.
A few more options to be aware of are deinterlacing and aspect ratio. As you know, NTSC utilizes two alternating fields for every frame. If you shrink the video to say, 320×240, you may end up with interlaced artifacts when there is a lot of movement in your footage. The better option, and most encoders offer this, is to deinterlace the movie and then scale it down. This will remove unwanted motion interlacing effects, and you will not lose any more detail than normal once you scale it down to web video size. If anything, motion sequences should play back smoother and clearer.
Also be aware of aspect ratio. D1-resolution files often have a different aspect ratio than web video files. A 720×486 movie scaled to half will be 360×243, not 320×240. Basically, any format that contains non-square pixels should be adjusted so it displays correctly on a desktop screen, which has square pixels. Most encoders offer a resize option or a way to select aspect ratios, but be aware of the differences — especially if you are encoding a client's clip.
There are other process options I'll discuss in future columns, but start to explore the ways you can enhance your web video projects. Software has advanced a lot in the past year, and most of the bases are covered for producing absolutely pristine-looking web content. So journey off the wizard path and explore the advanced features of your favorite encoders. There are a whole lot of options between digitizing and encoding.
Frank McMahon is a media artist specializing in directing, editing, animation, and graphic design. He has been working in broadcast for 15 years and his client list includes FOX Television, Microsoft, Yahoo, Time Warner, and Media One. He can be reached via his media company at www.fmstudio.com or through the Portland Media Artists at www.mediaartist.com.
Feedback
To comment on this article, email the Video Systems editorial staff at vsfeedback@intertec.com.
