For the stations that have made the transition to digital facilities, generating HD graphics often remains a challenge. Stations usually have a large library of graphics designed specifically for SD material used repeatedly on news and sports broadcasts, as well as in interstitials.
Unfortunately, there is no straightforward way of generating video by an SD graphics system to HD. Scaling up an SD signal with 4:3 aspect ratio to an HD signal with 16:9 aspect ratio will both blur the graphic and stretch it horizontally. For example, a sharp SD circle will look like a blurred oval on an HDTV set. While there is still a large audience for analog or SD signals, you don't want to alienate the HD audience. Fortunately, there are ways to make both sets of viewers happy.
The graphic system should be resolution-independent, meaning that the images are implemented as vectors rather than raster-based. This allows the shapes and colors to be mathematical formulas residing in the brains of the software, and they are rendered (or created) as pixels at the desired frame rate.
For example, a circle that fills half the screen renders as 320 × 320 pixels, and a twice-larger circle renders at 640 × 640. Both circles are based on the same mathematical formula, but are handed over to the video board as a different area of pixels, so the edges maintain sharpness.
This is the case with material (color) when applied to geometrical objects. Following the same concept, materials are also represented by formulas and rendered in real time. This process maintains the desired resolution and image smoothness and sharpness while preventing color banding. The material will calculate its color values per frame and interpolate the values it allocates to pixels based on the size the shape takes on the screen.
Editing and playout
Once you have an engine that can create the high-resolution graphics, the next step is to provide both editing and playout control over them. Stations sometimes employ two types of control systems. One focuses on content creation tasks, and the other focuses on graphics playout.
Broadcasters may consider deploying either or both platforms, depending on the needs of their organization. It's typically recommended that the CG system reside on top of the same render engine so the control room operator sees precisely how the next graphics fit on-screen and can see immediately if safe limits are exceeded.
An important characteristic in a graphics design environment is the wysiwyg (i.e., what you see is what you get) aspect. Without this feature, the designer has to guess how the graphics will look in the final output. For example, take a designer who is trying to compensate for the anemographic distortion between 4:3 and 16:9 images without the benefit of both types of viewers to see the images. Without the wysiwyg capability, the designer may end up with either great-looking SD or HD graphics, but seldom both. In a multiformat environment, maintaining the wysiwyg capability during the design process can save many working hours and result in more compatible images over different types of TV screens.
Once a new graphics hardware and software platform has been installed, the art of graphic design can be tackled. When broadcasting graphics in multiple video formats, there are three main challenges (in increasing order of difficulty): frame rate, resolution and aspect ratio.
When graphics systems handle animation speed independently from frame rate, no special considerations are needed for graphics without animation. Broadcasters adopting an HD standard with the same frame rate as their existing SD standard will incur no frame rate issues.
The resolution of the video standard is relevant when working with pixel-based graphics such as stills and video clips. Obviously, an HD image has a greater resolution than one created in SD. The 1080i HD format has almost six times more pixels than that of an NTSC SD signal. Other graphics elements will scale gracefully between resolutions when using a vector-based graphics system. When presenting pixel-based graphics in higher resolutions, you may find that the result will look blurred and diffuse if the pixel graphics are not of sufficient resolution.
If you have a small stock of high-resolution stills, now is the time to begin building a library with a resolution sufficient for HD. The library of HD-ready stills can be used for SD viewers as well, because the software will automatically downsample stills to the appropriate resolution.
That being said, downsampling might also show some artifacts if the graphics device does not support mipmap. Mipmap is a method where once an image is loaded, several versions of the image are automatically created in lower resolutions. Mipmap blends between the resolutions as per the total area that the image takes on-screen, creating an average-based resolution. (See Figure 1.)
Designers should carefully consider their image's displayed aspect ratio when the image needs to cover specific portions of the screen. Consider a lower-third banner graphic designed for SD that extends across the whole width of the screen. What portions of an HD widescreen would you want the graphic to cover? This is comparable to the difficulties encountered when adapting widescreen moves for 4:3 broadcast, except the width and high dimensions are reversed.
What do you do for your HD audience when you don't have enough 16:9 graphics for broadcast? There are four main strategies for dealing with aspect ratio issues:
- padding the sides or the use of ears;
- trimming the height;
- stretching the height; and
- changing the layout of individual structural elements.
The first three are essentially stop gap measures. They provide methods of outputting HD graphics, which will keep your HDTV audience satisfied until you can start sending them true HD, 16:9-friendly graphics.
Padding the sides of the graphics will retain the aspect ratio of the graphics, but will leave blank areas on either side of the graphics. This strategy requires no effort because it can be completely handled by the graphics engine. However, graphics that previously stretched across the full width of the screen may look strange when they no longer reach the side screen's edges. This solution works well for graphics installations that only display side panels on either the left or right edge of the screen, because you can easily keep the panel aligned with its screen edge.
Because the market is still in transition for both the consumers and the content provider, it is likely that we will continue to see the HD ears for some time. The blank HD ear space can be used for sponsor logo, schedule information, game scores, etc. (See Figure 2.)
When designing new graphics or redesigning existing graphics in a system that uses this strategy, the artist can accommodate both SD and HD aspects by extending background plates and other nonessential graphics to the edges of HD, while keeping important graphical elements within the safe area of SD.
In this case, the whole production is completed in HD and then downsampled to create the SD version. This approach is less expensive than producing both SD and HD graphics in parallel. Plus, only one HD graphics device is required, as long as it can handle both SD and HD.
Stretching the height of an SD signal to span across the height of an HD signal can be done without cropping or changing the aspect ratio of graphics as long as the graphics don't use the full height of the screen. This also requires no effort. The downside is that the proportion of the screen height used by the graphics will increase. The lower-third banner graphics designed for SD would cover only 4/9 of the height on an HD screen — not a desirable result.
Stretching the whole graphics horizontally to fill the HD widescreen also requires no effort. The downside is that it changes the aspect ratio of the graphics. Stretching will change perfect circles into ovals. However, unlike converting an SD signal to an HD signal, there will be no blurring, because the graphics system will render it with the appropriate resolution. Many graphics look fine after being stretched. Although, certain graphical elements such as circles, squares, natural photos and some text fonts will look odd after being stretched. Also, sponsors tend to dislike their logos being stretched or changed.
The best results are achieved by changing the layout of the individual structural elements of graphics designed for SD before broadcasting those same graphics in HD. While this requires more effort, it surpasses all other strategies on an aesthetic level.
The human factor
Good graphics design tools can simplify the process of converting SD graphic layouts to HD. And, this eliminates the need to recreate the HD graphics from scratch. However, someone still needs to inject the artistic sense and sensibility that will make graphics look good in 16:9. The artist will pick elements and adjust their size, position and aspect ratio to make the graphics look better. Easy-to-use graphic design tools that can arrange elements in nested groups and allow the graphics artist to operate on both individual elements and groups of elements can greatly simplify this process.
The front end
Another important aspect must be considered: the database and front-end CG airing tools. When shooting to air two or more sets of graphics, two or more sets of files (scenes or pages) are called simultaneously. When a designer fixes something in the HD page, he must be able to track and determine whether these fixes need to automatically take place in the SD page. This means that a single CG front end needs to be configured to work with multiple databases simultaneously.
Once you've got all these issues in mind and have identified a solution, you're not done. Broadcasters are learning that they have a tremendous opportunity to serve other channels. Those new channels could be 3G broadcasts, podcasts, webcasts and VOD on the Web.
Each of these platforms has unique graphic and image needs. Many of today's graphics platform manufacturers are working now to develop these solutions. While that shouldn't drive today's HD graphics purchase decisions, a unified graphics platform serving all these new channels may be a good solution down the road.
Nir Goshen is the creative director at Vizrt.