Imagine a computer as thin as paper and no larger than 8 1/2 by 11 inches but with all the power of a room-sized, Cray-class supercomputer.

Imagine a cigar-box-sized device that converts the heat of a campfire into enough electricity to run a television set.

Imagine a chalkboard that, at the push of a button, produces a paper copy of anything written on it.

Imagine a roll-up solar panel that can be unfurled on the dashboard to recharge a dead car battery in half an hour while plugged into the cigarette lighter.

Imagining such things and, more significantly, creating them, has been the full-time preoccupation for more than 30 years of Stanford R. Ovshinsky, a self-taught inventor from Detroit who has almost singlehandedly created a new kind of solid state physics.

His discoveries of certain peculiarities in the behavior of atoms, once scorned by many scientists, have led to the creation of materials that can produce the effects that make these devices work. All of these devices and dozens of others are either in commercial production or in development -- either short-range or long-range -- at Ovshinsky's company, Energy Conversion Devices Inc., based in this Detroit suburb.

Although Ovshinsky's early claims met with considerable doubt among most U.S. experts in traditional solid state physics, his unorthodox ideas found ready acceptance among the leaders of Japan's high-technology industries. Many of ECD's research, development and production programs are being carried out under various forms of collaboration with major Japanese companies such as Matsushita, Sony, Hitachi, Canon and Nippon Steel, the world's largest steel manufacturer.

Over the years Ovshinsky has made about 80 trips to Japan to negotiate contracts, to lecture before packed halls of scientists and industrialists and to meet with prime ministers and cabinet ministers. One Japanese company, reversing a trend of recent years, has asked ECD officials to reorganize its research programs to conform with ECD's work.

"We've tried for years to get American companies interested in our work, but we haven't had much luck until recently," Ovshinsky said. "Once the Japanese starting coming aboard, the Americans began to take a look."

IBM and Raytheon now have licenses to manufacture some of the ECD devices.

Ovshinsky has also won support from many of this country's top academic specialists in solid state physics. David Adler, an electrical engineer at MIT, said that in 1968 when Ovshinsky published his first paper in Physical Review Letters, the most prestigious U.S. physics journal, many scientists said, "It is unstable, it is uneconomical and it doesn't work."

Nowadays, Adler said, former critics are claiming, "I invented it first."

"The most eloquent testimony to the importance of Ovshinsky's work," Adler said, "is the fact that his paper has become one of the five most cited publications in the history of Physical Review Letters."

The peculiar properties underlying Ovshinsky's inventions are those of amorphous materials, manufactured solids whose atoms are linked not in the regular latticework patterns of crystalline solids, the basis of most of today's electronic wizardry, but in irregular, disordered arrays.

Ovshinsky and the scientists he directs have learned how to select and alloy different elements into a single amorphous solid, usually in the form of a thin film, or layers of different alloys, so that they produce a variety of effects.

Some alloys make electricity when heated, others when light shines on them. Some alloys are electrical insulators that become conductors when an electric field is applied. Some function as transistors, others as the kind of switches that constitute a computer memory but without the conventional computer's annoying requirement for a continuous power supply to retain data.

Ovshinsky's success at making such materials has largely borne out his early claims that low-cost amorphous materials can do almost all the things that high-cost crystalline silicon can do and many more.

The silicon crystals in conventional transistors, computer chips and solar cells have their silicon atoms spaced at regular intervals, each linked to its neighbors by a pattern of atomic bonds that is the same for every atom. The atoms in a crystal may be arrayed as if they were the corners of millions of tiny cubes, or other regular shapes, stacked together.

An amorphous solid, by contrast, has no regular spacing of atoms. It is a jumble, each atom's bond angles reaching in different directions and distances before linking up with neighboring atoms. By planning the jumble so that atoms of various elements take up predictable three-dimensional relationships within the solid, various special effects can be achieved.

A given element can be made either crystalline or amorphous, but the conditions required to produce either state differ greatly. It is cheaper to make amorphous materials, and they have the advantage of allowing different elements to be alloyed, an impossible step with crystals.

The classic example of amorphous silicon is glass, which forms when melted silicon cools. Ovshinsky has pioneered manufacturing methods that produce solar cells and the substrates for integrated circuits and other products in large sheets almost as easily as glassmakers produce windows.

The sheets are 18 inches wide and 1,000 feet long, but could be any size. The manufacturing process deposits a microscopically thin layer of amorphous material on a thin stainless steel or plastic sheet that comes off a big roll and passes through a series of sealed boxes containing electrically charged vapors of the relevant atoms.

The atoms stick to the moving sheet, forming a film of mixed elements. If the material is to have three different layers, the sheet passes through three such boxes, each holding its own mix of atoms. At the end, the sheet emerges and is rolled up again or cut to the desired size.

Unlike conventional solar cells, which are thick, heavy and breakable, Ovshinsky's solar panels are flexible, paper-thin sheets. The larger the sheet, the more power. ECD's current solar panels produce about 5.5 watts per square foot or about 1,100 watts per pound of panel, nearly double the goal the National Aeronautics and Space Administration set for achievement by 1995 to power a space station.

The panels now in production convert about 8 percent of the solar energy falling on them into electricity. In the laboratory an efficiency of 13 percent has been achieved, comparable to crystalline solar cells. New amorphous alloys currently in research promise 16 percent efficiency, said ECD's research director Stephen J. Hudgens, and 20 percent efficiency is considered achievable in the next five to 10 years.

Tiny pieces, less than a square inch in area, already run calculators and watches. Larger panels are powering remote irrigation pumps in Spain and being planned for previously unelectrified villages in India.

A curved solar panel that can be laminated onto a car's roof is being tested to power fans that would automatically ventilate a closed car parked in the hot summer sun. In winter, the power could keep the battery charged.

Solar cells make electricity because around the nuclei of certain atoms are electrons that are, in effect, only loosely bound. Once the electrons absorb energy from light, they become "excited" and escape.

Because electricity is simply a flow of electrons, wires attached to the cell channel the free electrons into some device, such as a light bulb or radio. The electrons flow through the device, making it work as they move along, and back through another wire into the solar cell.

The device that generates electricity from heat works much like a solar cell except that the atoms in it are chosen to produce their optimal release of excited electrons when they absorb heat energy rather than light energy.

Because solar cells do not work in the dark, ECD has also developed a new kind of rechargeable battery, using amorphous materials, that for a given size will run twice as long as existing rechargeable batteries.

While some applications of amorphous technology offer mere convenience, Ovshinsky's thin-film computer, still in the research stage, promises a virtual revolution. It could achieve the next major step in computer miniaturization and cost-reduction, cramming microscopic versions of all the components of today's supercomputers, which fill a small room, into a film no bigger than a sheet of typing paper.

Existing computer technology is essentially two-dimensional. All the thousands of transistors and memory switches in a conventional chip are arrayed over a flat surface. To grow means increasing the surface area by adding more chips. Thin-film technology makes it possible to grow simply by adding layers of amorphous components on top of each other. Several hundred layers, each containing thousands of components, would still be microscopically thin.

Layering conserves space but also permits true three-dimensionality because the hookups between components can be not only horizontal, as in existing chips, but vertical too. A component in one layer could "talk" directly to a component in a layer above, thus cutting the internal communication time that limits the speed of today's computers.

The Pentagon is interested in amorphous computers because they are free of the crystalline computer's vulnerability to radiation damage. They could, in theory, keep working in a missile's guidance system or an orbiting "Star Wars" battle station in a nuclear war.

Long before the thin-film computer arrives, ECD already has a thin-film, flat-screen display aimed at the portable computer market. It is a liquid crystal display but with such small picture elements, or pixels, that the display has much higher resolution than existing LCDs. Behind each pixel is a microscopic amorphous switch to turn it on or off as needed to form the image. ECD engineers have used the same technology to make a prototype flat-screen television display, two inches on a side.

Not all Ovshinsky's amorphous alloys find electronics applications. One turned out to combine a kind of surface slipperiness with the hardness of a diamond. It is now marketed as a coating for machine tools that extends their working life by up to 40 times.

"All the properties we find are in the atoms, if you know how to find them," said Ovshinsky, who professes an intuitive feel for what will happen if various given elements are alloyed. "I think I know the atoms like friends. I really visualize them. Sometimes I feel the atoms talk to me."