High Speed Photography and Micrography

  • Published on

  • View

  • Download


<ul><li><p>High Speed Photography and Micrography</p><p>J. S. Courtney-Pratt</p><p>A brief review is given of those methods of high speed photography that are or may be useful for micro-graphy. When we try to record an event at a high lateral magnification, the image is moving muchfaster than the object, so that exposure times must be proportionately reduced. Moreover, at high lateralmagnification, the available light flux per unit area of the photographic emulsion is in general much less.These two effects make high speed micrography the most difficult branch of high speed photography. Insome ranges of speed and magnification, relatively conventional equipment is adequate, but there is nosingle technique that will do all that we want.</p><p>In many problems in research and industry today, itis necessary to take pictures with a very short exposuretime to study what is happening, or to see how certainkinds of fault occur, or to see whether equipment isbehaving as it should.</p><p>If we were photographing, for example, a car travelingfast, we might need an exposure time for any frame be-tween 0.01 see and 0.001 sec. The exposure that isneeded is dictated partly by the speed of the car, andpartly by the relative size of its image in the camera.Suppose the car was traveling at 100 mph and that itwas 75 yards (68.58 m) away. If we had a camera thattook pictures of the same size as professional 35-mm inecameras, each frame would measure 16 mm X 22 mm.With a 6-in. (15.25-cm) focal length lens, the image ofthe car would be about half as long as the frame, andthe speed of the image relative to the emulsion wouldbe 1/450 of the speed of movement of the car. If, todetermine the detail we want, we decided that the imageblur should not be more than 0.1 mm, we obviouslywould need an exposure time as short as 0.001 sec. Ifthe car was more distant or moving toward the camerarather than across the line of sight, the exposure timewould not need to be so short. Figure 1 shows threesuccessive frames of a racing car traveling at speed.The picture is of high quality and acceptably free fromblur, both with respect to the car and to the back-ground. This picture was taken at the ordinary rate of24 frames/sec, and with an exposure time for eachframe not much shorter than 0.01 sec. It is because thecar is a large and relatively distant object that we getsharp pictures at normal exposure times.</p><p>Suppose, however, that we wanted to study the de-formation of the tread of the tire. A possible waywould be to make a section of the roadway out of glassand to photograph the car as it passed over the top.</p><p>The author is with Bell Telephone Laboratories, Inc., MurrayHill, N. J.</p><p>Received 22 September 1964.</p><p>The tread pattern on tires in ordinary use may havedetail to be examined that is only a few tenths of aninch across. We would need to take pictures at near-natural size rather than reduced by a factor of 450,thus the rate of movement of the image in the camerawould be 450 times faster. The exposure time that wewould need would have to be 450 times briefer.</p><p>High Speed MicrographyThe extraordinary shortness of the exposure that is</p><p>needed when we begin to study fine detail is indeed ageneral problem. It is aggravated when we considerthe problem of taking pictures through a microscope.In a typical example, an insect's foot can move at oneor two ft/sec. Suppose we wished to study the legmovement of a fly. We might wish to magnify animage of its foot say a thousand times and so the imageon our film would be moving (intermittently) at speedshigher than 1000 ft/sec. To keep the blur on the filmto the same figure as before, i.e., 0.1 mm, we wouldneed an exposure time shorter than 0.3 X 1-7 sec.</p><p>Again, we think of the movement of the column ofmercury in a thermometer as a slow process. In aclinical thermometer there is a constriction just abovethe bulb, so that after taking a patient's temperaturethe top of the column of mercury remains at the maxi-mum height, even though the temperature of the bulbdrops, and the bulk of the mercury contracts within thebulb. A number of suggestions had been put forwardto explain the way in which the mercury column wouldbreak at the constriction. These explanations allseemed rather unsatisfactory in one respect or another.Some years ago, R. McV. Weston decided to study theproblem in more detail. To overcome the awkwardrefraction due to the cylindrical curvature of the glassstem, he had flat surfaces ground on the sides of thethermometer. This made it possible to examine theconstriction with a microscope. To his surprise hefound that even when the mercury was expanding</p><p>November 1964 / Vol. 3, No. 11 / APPLIED OPTICS 1201</p></li><li><p>Fig. 1. Racing car driven by Stirling Moss traveling at highspeed. Repetition rate, 24 frames/sec; linear magnification '/50;</p><p>exposure time/frame 0.01 sec.</p><p>there was not a continuous thread of mercury throughthe constriction. As the temperature in the bulb rosethe mercury was pushed into the convergent part of theconstriction. The curvature of the mercury surfacebecame greater and the pressure in the bulb, to over-come the surface tension effects, correspondingly in-creased. After the pressure had reached a value suf-ficient to force some mercury past the narrowest part ofthe neck, a globule of mercury would be shot rapidlyfrom the narrowest part of the neck out into the parallelpart of the bore of the stem of the thermometer. Therate of movement of these globules was not tremen-dously high, but they would travel a distance as greatas their own diameter in a very short time, and to fol-low these movements it was necessary to use cinecameras working at rates of about 3000 frames/sec.Even so, one can barely see the movement of the drop-lets; but one can follow the early stage of their forma-tion and can form some estimate of their speed andfrequency. The whole of this phenomenon has been</p><p>elucidated by this elegant series of high speed pictures.Three successive frames from one of the records arereproduced in Fig. 2. One can see the stages in theformation of the droplet, and the deformed surface ofthe column of mercury after the droplet has crossed thegap.</p><p>When we are studying faster phenomena such asdetonation, explosion, deflagration of solid and liquidpropellants, the atomization of carburants, the grindingor cutting of metals, the penetration of armour plate,the formation of electric sparks, the "explosion" of athin wire by a surge of current, the fracture of brittlematerials, the vibration and fatigue of metals, theformation of shock waves, the ablation of hypersonicprojectiles-to name but a few examples-the rates ofmovement are much higher and the significant detail isoften just as small. In such cases, it is even morenecessary to make use of sophisticated and ingeniousforms of camera.</p><p>Limitations on the Speed of CamerasIt is necessary to apply research and development</p><p>efforts to the improvement of cameras, for it is notpossible to continue increasing the speed of operationof a camera simply by increasing the speed of movementof the parts. Standard 35-mm cine cameras can beincreased in speed to about 200 frames/sec, but if wetry to go faster with these intermittent movementcameras the claw tears the film. If we are prepared toput up with smaller pictures, we can increase therepetition rate. The resolution measured in line pairs/mm may be no worse, but with the smaller format theresolution in line pairs across the whole frame will prob-ably be lower, and the information content of eachpicture will be less. For example, we can achieve aspeed of 400 frames/sec with a conventional 16-mmcamera which stops the film while each frame is taken.However, the over-all resolution of each frame (mea-sured in line pairs across the frame) is less than half ofthat of the 35-mm camera mentioned above.</p><p>A big advance in speed is possible if, instead of tryingto stop the film while we take each picture, we move thefilm continuously and introduce some form of imagemovement compensation so that the image and thefilm move along together during the exposure. Thebest known of the methods of image compensation isthe use of a rotating block of glass. Such a cameramoves the film continuously through a gate and rotatesa small block of glass between the lens and the film.Figure 3 shows how this principle works. A Fastaxcamera can take pictures on 16-mm film, each of whichis 10.4 mm X 7.5 mm, at a rate of about 10,000/sec.It can take pictures half this height twice as fast.*</p><p>* There is an extensive literature on the design and use of rotat-ing prism cameras, and in any survey of that field one would wishto pay tribute to the pioneer work of John Waddell. However, inthis present paper on high speed micrography, the author has notincluded a general bibliography on rotating prism cameras, asthat work is covered better in the books on high speed photo-graphy.2 4 6' 84</p><p>1202 APPLIED OPTICS / Vol. 3, No. 11 / November 1964</p></li><li><p>//</p><p>4C</p><p>/ /n</p><p>B ///</p><p>F</p><p>D-Ed/</p><p>Fig. 2. Fastax recording of movement of globule of mercury through constriction of clinical thermometer. Repetition rate, 3000frames/sec; magnification, 70 X. A, mercury in bulb; B, mercury in stem; C, constriction; D, globule forming, E, globule de-</p><p>taching from mercury; F, mercury surface at rest; G, mercury surface deformed by impact of globule.</p><p>If we try to drive a camera such as this faster still, wefind that again there is a limit set to the speed by thestrength and resistance to flexure of the mechanicalparts.</p><p>Rotating Mirror Framing CamerasIf we wish to go even faster, one of the best general</p><p>principles is to duplicate, or actually multiplicate, theequipment. One could easily take a set of four picturesat short intervals if one arranged four cameras in a rowand triggered their shutters one after the other withshort phase delays between them. Some of the earlystudies of animal locomotion were made with a schemenot very different from this by Muybridgel 2 from 1872onward.</p><p>It is reasonably easy to set up such multiple equip-ment if one is trying to photograph a fairly distant ob-ject. However, if we wish to study something close by,it is difficult to place all the lenses and cameras suf-ficiently close together to be able to get the pictures.In any case, the pictures would be taken from somewhatdifferent viewpoints and we would have problems fromparallax. We should not be able to project such asequence to give a steady picture.</p><p>A most valuable solution to this problem was in-vented at about the same time by Miller3 and Bowen,4working independently. They arranged a row ofcameras, each with a lens and film and all with theiraxes pointing toward one common point. Theyformed a real image of the event on a mirror at thiscommon point, and as they rotated the mirror thebeam of light pointed to the camera lenses one after theother. The mirror could be small and could be rotatedat a very high speed, and the camera lenses could befairly small and close together, so that one couldachieve an extraordinarily high repetition rate.</p><p>This kind of camera is now called a rotating-mirrorframing camera. Some very elaborate cameras workingon this principle have been built in a number of labora-tories in several countries.3 -31 Beckman &amp; Whitley,Inc., and the Cordin Co. in the United States, and Barr &amp;Stroud, Ltd., in Scotland, manufacture cameras of thiskind commercially. One of the Beckman &amp; Whitley</p><p>cameras can take a sequence of 25 pictures at 1.2 X106/sec (or with half-height frames at 4.3 X 106/sec).Another can take 80 pictures at intervals of 0.7 Msec.A camera built by the U. K. Atomic Energy Author-ityIi l2 takes a sequence of 30 pictures at 107/sec. TheBarr &amp; Stroud cameras are based on this and laterwork31 and, using interchangeable sets of lenses, cantake 117 pictures at 8 X 106/sec, 59 pictures at 4 X106/sec, or 30 pictures at 2 X 106/sec. The limit towhich you can drive a camera of this kind is again setby the strength of mechanical parts. If you drive therotor too fast, it bursts because of centrifugal stresses.</p><p>With a rotating-mirror framing camera, if we use awide-aperture lens, we need a large mirror. If thewidth of the mirror is large, we cannot rotate it at sucha high angular speed. If we use very narrow apertures,not only is it difficult to get enough light but we findthat the resolution of any picture is limited because of thediffraction of light at the small apertures. Schardin3 2showed that it would not be possible to increase theproduct of the repetition rate and the number of linesresolved across each frame beyond a certain figure,which would depend on the strength-to-weight ratio ofthe mirror material. t</p><p>Cameras that have been made so far are still wellbelow Schardin's fundamental limit. But even so, ithas been possible with rotating-mirror framing camerasto achieve a resolution of 280 line pairs across eachframe and a repetition rate of 4.3 X 106 pictures/sec.(The product nin = 1.2 X 109 sec-i.)</p><p>Rotating-mirror framing cameras are excellent for thephotography of brilliantly illuminated events. Suchimage brillance can be achieved when photographingobjects at or near natural size. Sultanoff,3 3 for ex-ample, has been able to record a sequence of pictureswith a camera of this kind in color at sub-usec inter-</p><p>t Schardin's Fundamental Limit: for presently availablemirror materials, the product nm cannot be greater than 4 X 109sec-, where m is the number of line pairs resolved across theframe in the direction of sweep, and n is the repetition rate inpictures/sec.</p><p>November 1964 / Vol. 3, No. 11 / APPLIED OPTICS 1203</p></li><li><p>AA-</p><p>-</p><p>I i GATE</p><p>U'.8 1 </p><p>*1l</p><p>GLASSSHUTTER BLOCK</p><p>j2 </p><p>'1.2</p><p>CONTINUOUSLYMOVING FILM</p><p>Fig. 3. Diagram showing the principle of image compensation bymeans of a rotating block of glass. As the block turns, the imageis displaced, and, for the duration of the exposure, the image</p><p>remains in register with the continuously moving film.</p><p>vals-a really remarkable achievement. A few photo-graphs from one of his records are shown in Fig. 4.The record shows the movement of a detonation frontdown a stick of explosive, and the way that the detona-tion front crosses a gap between two sticks of explosive.</p><p>The problem of light level becomes much more dif-ficult when one tries to photograph at considerablemagnification. There are some subjects which arebright enough. Zernow and Hauver3 4 have taken aseries of pictures at a magnification (from object tophotographic emulsion) of about twenty-five times, ofthe explosion of a wire when a large surge of current waspassed through it. Figure 5 shows a few framesselected from one of their recorded sequences. Muchfurther work in the study of exploding wires and foilshas been reported since this.3 5 -40 However, for reallyhigh magnification, it is almost impossible to getenough light through a rotating-mirror framing camerato take a record.</p><p>An Image Dissection CinemicrographSome years ago, the author conceived a different ap-</p><p>proach to this problem.4 4 2 An ordinary microscopewas set up, but, instead of placin...</p></li></ul>