Howard sipped his coffee as she started the video. By now he knew every second by heart, and he could watch it without emotion. He no longer grieved for the dying family, and he could listen to the woman’s last words of comfort to her son without cringing inside. They watched it together, in silence.

“This is the original video,” she said, “the same version you’ve seen in Phoenix. I’ve taken the signal on the tape and programmed it into the computer, then used it to boost the definition several-fold. To see it we’ll need a very high-definition television monitor, this one here.” She flicked a switch on a console and the video played again on a wider television screen. Howard could see the improvement immediately. Bonnie kept her hand on the pause button and as the camera panned to the ground below she pressed it. The frozen picture was much sharper than on a standard video-recorder, too; there was no fuzzy line or flickering. On the screen was one of the towers, and Howard could clearly see a figure with a rifle. The face was still blurred.

“The quality is much improved, but there are limits to what we can do with analog methods,” said Bonnie. She let the video play on. “We can get better results by digitising the video and storing it on a CD.” She patted one of the computers, an unimposing white box. “This is our image processor, which takes the video signal and digitises it. We call it a frame grabber. It can digitise images in real-time – about one-thirtieth of a second each – and save them into storage. Then we use a computer to handle and process the data. Once we’ve processed the images and cleaned them up, we can choose the frames we want on the monitor, and print them on a film recorder. That gives us much better definition. That’s what’s in the file – computer-generated pictures.”

Howard opened the file and looked inside. There was a stack of more than twenty glossy eight-by-tens. He went through them. There were photographs of the towers, and close-ups of the snipers. None was clear enough to make out their faces, however.

“That’s the best I can do with my equipment,” Bonnie said as she saw his face fall. “Not much help, I’m afraid. Though you can see that they’re better than the images we had on the screen.”

“What have you done to them?” Howard asked.

“I tried neighbourhood averaging first, but that wasn’t too helpful,” she said. “Those pictures are after I used a technique called median filtering. I could probably enhance them more if I used pixel aggregation, but that’s going to take me more time.” She smiled as she saw the deep furrows appear on Howard’s forehead. She took a sheet of paper and a pencil, and drew a square box with deft strokes. “Imagine this is a tiny piece of the screen,” she said. “That unit is as small as you can go. It’s indivisible. We call it a pixel. The camcorder in the plane was a Toshiba TSC-100, with a Canon 12:1 servo zoom lens, and it records 410,000 pixels with seven hundred horizontal lines. That’s a lot more than the average camcorder, some have fewer than 300,000 pixels on the image sensor. He had some pretty specialised equipment.”