As speech makers at both political conventions drove home their messages, television cameras cut away from the podium and zoomed in for close-up shots of someone -- a minority delegate, a mother with children, an AIDS activist -- who seemed to illustrate a particular point.
When this technique works, viewers don't even notice; they aren't aware of the back-and-forth camera work the networks use to make political speeches visually interesting.
But the foibles of audience reaction shots became apparent this week during television coverage of the Republican National Convention.
Time and again, as Patrick Buchanan, Ronald Reagan and Jack Kemp spoke on Monday and Tuesday nights, network cameras focused on black delegates' faces. So many times, in fact, that the repeated images of black delegates may have left some viewers with the impression that the group of Republicans gathered in Houston was a much more diverse group than it actually was. But black delegates accounted for only 4 percent of the 2,200 or so delegates to the convention.
"I don't think we went overboard on showing minorities or blacks," said Bob Furnad, executive vice president at CNN, the network that provided the most thorough coverage of the convention.
But Furnad acknowledged that there is "absolutely a danger" of choosing images that misrepresent what's actually happening in the convention hall and that the danger is greater with the GOP.
"The Republicans are visually more difficult," he said in a telephone interview from Houston. "There's greater variety with the Democrats . . . more people in outrageous attire . . . more minorities . . . more people who are openly gay. It's just naturally more visual."
The naturalness -- or unnaturalness -- of the pictures viewers see during the speeches is the key.
Television producers at the Democratic convention who wanted to visually dress up a point being made by a speaker about gays and lesbians, found it relatively easy, said Furnad. They pointed the camera at the area on the convention-hall floor where people are holding signs that say, "Gay and Lesbian Rights."
At the Astrodome, however, there were no gay and lesbian signs. Instead, cameramen were trying to zero in on gay or lesbian delegates to see their reactions to Republican speeches. But there were only two delegates at the convention who acknowledged being gay, and they have been interviewed and photographed repeatedly. Besides, Furnad added, they weren't wearing buttons, and it would have been awkward to flash the words 'Gay Delegate' under their images.
It's a complicated process, these convention pictures of people wearing goofy hats, parents' holding up babies, men and women with tears in their eyes.
But it's a carefully planned process: The selection of images starts well before the speeches, Furnad said. About half the time, the network producers and directors get an advance copy of the speech. If someone is identified or singled out by name in the speech, the producers will often try to find where that person is going to be seated, so that the camera can show him or her.
"During the speech, you have 5 or 8 or 9 cameras scanning the crowd," Furnad explained. "And, generally, you look for shots that are logical. If there's a mention of blacks in the speech, you look for blacks. Hispanics, Hispanics. If there's mention of abortion -- even though it's not only a women's issue -- you look for younger women. . . . Or, you look for signs -- pro-life, pro-choice. . . . Or, if it's a rather lifeless speech or session, you might look for people not paying attention."
The goal of such pictures, producers say, is to make speeches interesting visually and emotionally to a prime-time audience, which generally is accustomed to more stimulating fare than a speech from Texas Sen. Phil Gramm.
As David Roberts, the news director at WBAL, put it: "They cover conventions like football games, and those are the shots of people in the stands; they are supposed to capture the emotion of the moment, tell us what the moment means."
The most obvious -- and perhaps the most moving -- example of this came Wednesday night during the speech by Mary Fisher, a 44-year-old, HIV-positive woman. As she spoke of the need for compassion for AIDS victims, the cameras showed Fisher family members and several delegates silently wiping tears from their eyes. The coverage was virtually identical on CNN, C-SPAN and PBS, the only three networks that showed the speech -- even though CNN and PBS had their own crowd cameras in operation and were not using pool pictures.
While the selection of such shots can play a crucial role in determining how millions of Americans feel about such speeches or the convention and the political party, television producers say that deciding what to show is a highly imprecise matter. And it is a matter that continually demands what Furnad calls "course correction" -- midstream adjustments adding or subtracting certain kinds of pictures.
There is a reminder that for all its marvelous technology, the machinery of TV is still run by human beings, and its images need to be questioned.
"One of the most common problems is that most of my camera operators are young men," Furnad said, "and, too often, their cameras are showing us pictures of young women in the hall."