00:00:02.740 - 00:00:54.360

okay, Every Yeah. Hello. I'm Mike Cannon, and this is conversation on our program this evening. We're going to be talking about computer graphics and especially those graphics which you see on your home television screens. By that, I mean pictures that leap out of you as though from some third dimension various forms and colors that combine in a startling in an attractive way.

00:00:54.640 - 00:01:12.660

And all the myriad other things which you have come to identify as network logos or introductions to big time sports broadcasts, Uh, to give you a better idea, I suppose we could run a few examples of that. So let's do. Let's call upon our engineers to run a few graphics and you'll have a better idea of what I mean.

00:01:13.140 - 00:01:39.160

And these are graphics that are the result of, uh, computers. Well, our tape is not ready on that, but we'll get it, uh, sooner or later in the program. These are pictures, if you will, that are the result of, uh, computers, hardware and programming, which causes the hardware to produce a particular type of Well, I suppose picture is the best word, but the picture is far more complex.

00:01:39.350 - 00:02:22.200

And in order to find out just what a computer graphic picture is. I'll return in a moment with my guest. Uh huh. Yeah, Yeah. My guest on this conversation is Doctor John Stout Hammer, professor of electrical engineering at the University of Florida. John, welcome to conversation.

00:02:22.530 - 00:02:51.130

Thank you. I'm sorry that our tape didn't roll at the beginning, but I suppose the the imagery of five live as it came up on the screen superimposed on the picture of you and me was a modest kind of computer graphics. Yes, it was. It was a way of taking two images and superimposing one on top of the other and switching from one part to the other part of the image by means of a very simple switch, much like you have at your home that is activating a light.

00:02:51.140 - 00:03:08.480

Except it has to operate very rapidly to be able to give you a transition from one picture element to another picture element. Well, the pictures that we're going to see in a few moments, I hope our pictures that are not the result of photography, right? That is correct.

00:03:08.490 - 00:03:37.880

Well, um, and they're not videotapes either as such or their videotapes by the time they come to this, uh, studio, of course. But before, they were really nothing more than a sequence of numbers, and all we did do is arrange these numbers in such a way that the television, uh, display device, a television tube can pick it up and produce a picture from it.

00:03:38.230 - 00:03:56.120

You see, the picture that comes to you from your TV cable is nothing more An electrical signal and all I have to do with a computer is compute that signal and do it cheaply and rapidly. That's what the computer enables you to do. I wonder if that first tape is ready, because if it is, it would give us a good idea.

00:03:56.270 - 00:04:12.450

Um, we'll go to our tape in just a moment, then. Um, John, when you talk about numbers, that is, in essence, what a computer does it manages. Numbers may be may be of some interest to note just how many numbers you have to come up with.

00:04:13.140 - 00:04:45.530

The television image that you look at on your TV screen is made up of 485 lines, approximately and across each of the 485 lines you have approximately 600 points, and each of these points are distinct, distinct from each other. So if you wanted to come up with a totally random picture on your screen, you would have to come up with something like 485 times 600 or order of 300,000 points for each time a TV picture appears in front of you.

00:04:45.540 - 00:05:09.960

Now, on the system used in the United States, that's 30 times a second. So you come up with order of 300,000 times 30 calculations per second. Now that's the kind of thing we've been doing with a computer. But of course we have to find some shortcuts because, after all, on anyone TV picture, not every point is different from its neighbor.

00:05:10.540 - 00:05:32.370

So what we came up with is a scheme of using the neighborhood coherence of the image. And that way, all we have to worry about is not the picture content itself. But the changes in the picture content that we have to calculate and thats much less and therefore can be done with a modest computer at modest cost at reasonable computing speeds.

00:05:32.380 - 00:05:56.430

And if I understand correctly. This is your special contribution to the field. Yes, I've been working in this for order of 10 years now, and some success. Well, I certainly know of that. And you have begun now the development of a laboratory here at the University of Florida that will be driven by 2 $300,000 computers in order to produce this kind of work.

00:05:56.620 - 00:06:16.550

Now, that doesn't sound like a modest or inexpensive set of computers. Well, of course, modest is your framework. Remember, Now they computers that today cost order of a couple of $1000.2000 dollars. A TRS 80 type machine 10 years ago could not be had for half a million dollars.

00:06:17.520 - 00:06:49.160

The costs simply have dropped very rapidly. And today, what we're looking at is not a way of producing pictures, but in the way of investigating the process of produce producing pictures. So the way of investigating requires really more hardware than actual production. Because we want to study the system and want to see what the system does and thereby find maybe some shortcuts, maybe some better equipment to do better pictures with less cost.

00:06:49.640 - 00:07:14.990

Now you have developed similar programs for other universities and research centers in the country. I understand. Prior to coming to Florida, I had built such equipment and copies of that equipment that were sold two other, uh, National Science Foundation supported groups. And, uh, these groups have made some very nice use of this equipment in producing Saturday morning like cartoons.

00:07:15.000 - 00:07:31.270

All right, we're going to see, I hope now, uh, with our hardware, some examples of Saturday morning type cartoons that have been developed by computers according to the kinds of procedures or languages that you have innovated. So let's go to that tape and see if it will not run for us.

00:07:34.350 - 00:07:53.960

I am John Stout Hammer, a professor in the College of Engineering. I have been at the University of Florida since August 1980. For the last decade, I have been working on means to have a computer produced pictures so that we may have a way of visualizing anything that may be computed by a computer.

00:07:54.840 - 00:08:17.560

Computers are great at the calculation of minutia of great precision. However, to make this useful to humans, one must have images, graphs and dynamic sequences of images My students and I have built special colour television signal generators, which are controlled by modest computers. In essence, we have a synthetic camera.

00:08:18.340 - 00:08:41.360

I can make images of any complexity that may be shown on the television screen. The sequence that you will see is nothing more than a big collection of numbers, which are carefully sequenced to give you the illusion of things in space and in time. Now, John, we're on live now.

00:08:41.620 - 00:09:07.260

Would you tell me what's happening on the screen? On the screen, there will be a movie sequence generated. You see a star falling there, for example. There will be some elephants moving their trunks, and there's a a motion around the tent. This sequence is called the Ringmaster, and it comes from Ohio State University, which uses the equipment I built for them.

00:09:08.140 - 00:09:36.660

And if you look at the image carefully, you will note that there are not really very many changes in a given line in the TV image. Mostly, there's a blue background, and there is a bear without moving its arms, For example, Um, we have a way of coming up with these images in real time, so it appears to us if it would be a motion that is a Saturday morning cartoon.

00:09:37.340 - 00:10:03.660

And one of the advantages of this kind of a system is that if you note in the motion sequence something not quite right, you could go back in the calculation process and say, Hey, the bear didn't move its arms in C in sequence from 1 80 to 224 frame numbers and you could recalculate those and grief, uh, generate the entire sequence without ever to have to use photography.

00:10:04.240 - 00:10:26.260

See, what are the problems in making animated Saturday morning cartoons or Fantasia cartoons even, is that you really don't know the spatial relations and the emotions until the film has been processed. And here we have a way of just simply calculating that the calculation is done, not in real time.

00:10:26.640 - 00:11:00.100

But the calculation will give you sequences, and you can see if anything is wrong. The pictures you have seen were computed images of objects made up of little pieces of tiles, and the objects appear somewhat as if they were hacked out with an axe. Visibility coloring and motion of these objects was calculated by a medium scale computer to give better appearance.

00:11:00.100 - 00:11:32.520

Two objects, 1 may use many more such tiles. The only difference is that we now have to calculate many more numbers, but that is precisely what a computer does very well. All we need to do is to allow more time for the calculations. To produce these images, one need to define objects and objects sequences.

00:11:32.940 - 00:11:55.920

We developed procedures. We call that a language for the definition of object parts and object assemblies. This language consists of type in commands and English like phrases. The sequence of pictures may also be specified in that language. Light source locations, reflections and surface texture may be added to the computation process.

00:11:58.220 - 00:12:21.000

I will now show you how these language commands are used to generate an image sequence. The anime to language provides complete and structured control of images through time to this first sample anima to script, which positions an object in the screen. Commands are successively added, which changed position, rotation, shape and size.

00:12:21.640 - 00:13:14.460

All commands execute sequentially and take effect in parallel so that animation of any complexity is controlled by simple independent commands. Without looping. Other language features provide control over color, intensity and background path movement and changing rates of motion and seem directives like pans, zooms and perspective to specify more complicated images.

00:13:14.840 - 00:13:41.860

The same types of commands are used at a wholesale rate. This walking skeleton is someone's master's thesis at Ohio State University. The figure can walk very convincingly over a level terrain. There are 22 equations just for the skeletal positions, which must be solved for every frame being shown.

00:13:45.340 - 00:14:27.850

My work has application in terrain modeling and showing geographic surfaces. The terrain really, maybe anything hardest. Iraq. Soft as a butter, maybe even suitable for a land shark. My work has also uses in architecture, where building shapes and placement can be simulated, as was done in this sequence from Cornell.

00:14:36.240 - 00:15:20.960

Locations and perspectives in relation with surrounds can be judged without having to construct an actual building. One could build an entire city, walk down its streets, look down its alleys, look up into the sky and look around. The city may consist of buildings, lots of buildings, tall ones, short ones, maybe even who knows a garden.

00:15:25.040 - 00:16:33.110

More recently, I have worked on molecular structure, visualization equipment and procedures developed by myself and others is capable of showing details of complex molecules. As a television sequence, we can simulate molecular interactions this way. Refinements in calculations can add reflections, reflections, shadows and smoothness. Remember, everything you saw was done with numbers, lots of numbers and with the guiding input of some perspiring humans who wrote the dreams of required programs with the plant capabilities of the Computer Graphics Research Laboratory.

00:16:33.650 - 00:17:09.950

My work will continue here in Florida and the remain interesting and perhaps productive to our college. I thank you very much, John. Those were startling pictures, but I emphasize again what you emphasize. They are not photographs, their computer images, and that's correct. They are computed point by point, and essentially a set of commands are put aside, uh, put away in a disk storage, and they're brought back from the computer to simply drive a television signal generator.

00:17:09.960 - 00:17:29.390

But there's everything there. There are curves and shadows, trans Lucien's that it's remarkable what the computer is able to do. The building. Yes, go ahead. That is not the computer that does it. It's a computer. Just simply calculates a sequence of what a person had told the computer to do, and the video screen does the rest.

00:17:29.600 - 00:17:51.010

No video screen simply responds to stimuli much like it with a response from a camera. But you have to think of these sequences you have to do to come up with the with the artistry in it that makes a person a computer doesn't do it. A computer just follows commands blindly and rapidly, and this is a creative use of the computer.

00:17:51.260 - 00:18:17.500

I would hope so, yes, yes, for example, in architecture, that building that the Cornell Project developed could be moved into different settings in order to discover if it had a correct proportion proportion if it had proper harmony with its surrounds, uh, it would fit with the with the building contour with the terrain contours it would fit with, possibly with the other buildings in the neighborhood.

00:18:17.640 - 00:18:39.670

It might clash. It might cast shadows. I showed, you know, real pictures here about shadows, but that also can be calculated. And, uh, you might find that most of time a building cast shadow on, uh, area, which is normally used for painting. For example, in a fine arts program, you can't have that.

00:18:40.240 - 00:19:03.510

Ah, the sequence of the skeleton walking was very convincing. I understand that was a somebody's Masters thesis. Just that sequence it took approximately well. That and some other things that made this walking, uh, simulation possible took somebody about one year of, uh, computer programming using an established machine and established language.

00:19:03.740 - 00:19:18.790

All right, let's talk now about fabrics and go to some slides that you have, and these slides will appear in a certain sequence so that we all right, Okay, this is essentially a photograph, which we are showing on the TV screen now of another TV screen.

00:19:18.800 - 00:19:33.560

And on this TV screen appear. Say what a blank area and on this blank area can deposit any one of the colours aligned on the bottom of that screen and a single data spot there. A sequence of these can be made up, and this is not a set of stitches.

00:19:33.570 - 00:19:52.810

Think about this as being about a 30 by 30 postage stamp that ultimately will make up an entire woven cloth. Now, if you add more stitches to it of different colors, you come up with. These are actually random lines that were drawn or were put on the screen by a pointing device.

00:19:52.810 - 00:20:13.640

Incidentally, like Penn pointing device, we can change the colors, and now we have a view of how this how a different set of yarns would appear. And now we can say, Repeat, let's if you repeat this in particular orders. That's what the cloth pattern would look like if you repeat it in a different way.

00:20:13.650 - 00:20:34.460

That's what it looked like and you could, of course, have many, many different repeat features. Here are some actual patterns on cloth that have been developed by the computer procedure or language that you have described here is one such I'll try to keep it out of the shadow.

00:20:34.640 - 00:20:51.910

Let me do another here. This is an attractive pattern. Incidentally, all of these three you have with you were done with a very small computer and done by a young lady in the design school at the school I was at before, and she is currently a textile designer and working for Ford Motor Company.

00:20:51.920 - 00:21:07.620

Yes, she's designing. I understand the fabric that will go on the seats of many Ford automobiles. The 1983 here. Actually, some of them are already out, and she made the designs by the use of the computer. Fascinating. We have another sequence. John. Let's see it, too.

00:21:07.740 - 00:21:41.430

Also on slides. Okay, this is a simply an image of, uh, how the system works. On the left side, we have a line drawing device, which defines objects by simply their bounding surfaces and the edges of those only How do you get? Let's pause just for a moment, as how do you get that design on the screen with a pointing device, light pen and a set of numbers that say says, Here's a point so far up and to the left from the left edge, Okay, and then draw simply lines between them.

00:21:41.510 - 00:22:01.690

A number of these these lines can define bunch of planes, much like a blueprint of a here. A an object. And if you now calculate what's in front, what's behind what's included you then wind up with a scan image, a raster image very much like a model that a TV camera would see.

00:22:02.010 - 00:22:21.660

This is synthetic. It's a computed image. It's not a real thing right now. If you care to add some more computational steps to this, uh, you could come up with translucency like we did. In this case. It's a wine bottle and uh, the wine bottle is made simply translucent, and it has highlights on it.

00:22:22.140 - 00:22:50.730

See, there's a reflection off from a light source, really from up on the upper left hand side of this image, and you can look through the wine bottle and everything behind it is greenish all right Now, the amount of calculation involved is quite a bit, and here it's a similar sort of, uh, procedure was done by using just curved columns and a You were a water picture up in front, which looks amazingly realistic.

00:22:50.730 - 00:23:17.690

Looks very realistic. It has curves, shadows, highlights, actually has highlights and curse to it. The real shadows and shades, I think, will be the next couple of pictures. This remarkable. This is a picture of really the last sequence that we have shown on that videotape before, and here you can see reflections both from the table top and from the spheres, one from another, also translucency of this fear.

00:23:17.720 - 00:23:44.160

And here we have a textured set of objects which reflect and through shadows, and part of the shadowing is done on from safe from the red ball onto that white ish ball, which is a curved surface shadowing effect. It's all computed point by point 480 lines down 640 across roughly 300 points, and this particular image took approximately three quarters of an hour to compute.

00:23:45.140 - 00:24:14.060

But, of course, then you store the commands and bring it back in 1/30 of a second, and you have a movie made out of it. John, why did you relocate yourself and your laboratory work to the University of Florida? Well, that's a hard question. Um, I have looked around to various spaces where I could do my, uh, research work properly with the right computational support and with the right people.

00:24:14.440 - 00:24:40.170

And I found that Florida has a very nice set of people to work with an industrial base that is going into more and more computers and especially computers for design. And when you design something you need to visualize. And a computer, of course, is the primary ways of visualizing today anything that you might design from an automobile body to textiles.

00:24:40.540 - 00:25:06.060

Two. You name it so that you would expect that new industries would rise to take advantage of this kind of technology, but that what you've already accomplished is stands ready to serve industries that we already have. Many here in the state of Florida. Well, actually, old industries are using these newer techniques, probably more than new industries would use.

00:25:06.540 - 00:25:26.060

For example, automobile body design is being done that way. For example, textile patterning is done that with these are old old industries, uh, the simulation that is done for flight work, for example. It's a branch of this kind of graphics that I've been working with. Let's talk about that for a moment.

00:25:26.060 - 00:25:54.330

I've been fascinated by flight simulators. I've never seen one, but I've read a good deal about them. And I did see a television special one night on the flight simulators used by one of the major airlines. This particular setup was in California, and the pilot sat in a cockpit, which was identical to, say, the cockpit of a 7 27 which was his usual plane.

00:25:54.330 - 00:26:20.560

And he was going to take a check ride with an instructor to check his proficiency, particularly on emergency situations such as engines out and, uh, as he was given the go ahead to start his role on the runway. The plane, actually, as it were bounced because the, uh, the cabin or the cockpit area bounced somewhat, even though it was just on hydraulic springs or whatever.

00:26:20.590 - 00:26:37.960

And ahead there was a runway that moved to meet him, as as it actually would in an aircraft. And then, as the plane took off, you saw and this was a night sequence, the lights of Los Angeles below you in a very realistic way. And as the plane turned again, you saw the actual landscape.

00:26:38.340 - 00:26:57.540

Well, that is done by Yes, it's all done by computer, really Several parts to flight simulator. The one I have been talking to you about is the visual aspects of it, something that would be hung on the windows that the that the pilot would see. But there is another very important part of flight simulators.

00:26:57.550 - 00:27:24.360

By taking the cockpit itself and moving it, it's a limited amount of motion. You can't in fact accelerated to 500 miles an hour, but you can't shake it and buffet it. And that gives additional realism, of course, to the flight simulator. But the great advantage of a flight simulator is is well, maybe if I may paraphrase another one, a fight or flight simulator that I've looked at some years ago cost more than a fighter plane.

00:27:24.840 - 00:27:44.480

Considerably more. But you see, it has the advantage that you can crash a flight simulator and be able to walk away from it. Yes, and fly it again and fly it again. And hopefully, by putting the pilot into situations that are scary and realistically scary, he will learn to avoid those.

00:27:45.440 - 00:28:01.570

Certainly life ones will avoid it. John, this has been fascinating. And I'm very grateful to you for being with me on conversation to talk about computer graphics, your new laboratory and work here at the University of Florida. I wish you much success. And let's do this again sometime on conversation.

00:28:01.690 - 00:28:45.970

Thank you very much. It's been my pleasure. Yeah, yeah, no.