Mark Lyn­der­say

Prof Er­ic R Fos­sum is prob­a­bly the most im­por­tant man in the world of mod­ern pho­tog­ra­phy, but he al­most walked right past me with­out my know­ing it.

Lost on my way to the lec­ture he was sched­uled to give last Mon­day, I spot­ted an old school friend, who al­so hap­pened to be the chair of the Open Lec­tures Com­mit­tee that brought Prof Fos­sum to T&T.

In an un­char­ac­ter­is­tic mo­ment of men­tal agili­ty, I de­duced that the per­son am­bling along be­side him had to be the fea­ture speak­er.

As the pro­fes­sor gripped my hand firm­ly af­ter in­tro­duc­tions, he qui­et­ly said, "Thanks for com­ing to the lec­ture."

"Huh," I might have said if I were 40 years younger, "as if."

Prof Fos­sum is the sci­en­tist who lever­aged the ground­break­ing charge-cou­pled de­vice (CCD) tech­nol­o­gy that un­der­pins al­most all mod­ern vi­su­al telecom­mu­ni­ca­tions to cre­ate a more en­er­gy-ef­fi­cient and ul­ti­mate­ly more scal­able re­vi­sion called com­ple­men­tary met­al-ox­ide-semi­con­duc­tor or CMOS tech­nol­o­gy.

Moore's Law, the ax­iom that's rather ef­fi­cient­ly pre­dict­ed the rate of tech­no­log­i­cal progress (it dou­bles every year), has been kind to these light-gath­er­ing tech­nolo­gies. In 1971, it was pos­si­ble to put 2.3 light-gath­er­ing sen­sors on a chip, now tran­sis­tors are pack­ing in up to 2.6 bil­lion of them.

The learned pro­fes­sor quick­ly pow­ered through the tech­ni­cal part of his pre­sen­ta­tion, a dense thing made of high-lev­el physics and some folksy char­ac­ter­i­sa­tions of his in­ven­tion.

"Light is a wave," he not­ed as a slide came up, "but light is some­times al­so a par­ti­cle."

This was the sort of thing that al­ways drove me a lit­tle loopy in sci­ence class and at this lev­el, with dis­cus­sions that ref­er­ence "pho­ton shot noise" and "the Pois­son Process," things were drift­ing firm­ly in the di­rec­tion of hard sci­ence.

One sen­tence peeped through, how­ev­er: a men­tion that sci­en­tists seem to be agree­ing that we are ap­proach­ing the lim­its of physics when it comes to pix­el sizes.

Pix­els, or pic­ture el­e­ments, are the tiny light-sen­si­tive sites packed on­to a cam­era's sen­sor that ac­tu­al­ly cap­ture the light rep­re­sent­ing your pho­to and trans­late it in­to the elec­tri­cal im­puls­es that be­come the bits of an im­age.

Prof Fos­sum be­gan his work on the tech­nol­o­gy that would be­come the CMOS sen­sor at NASA's Jet Propul­sion Lab­o­ra­to­ry, where he was charged with mak­ing dig­i­tal-cap­ture de­vices that were small­er, lighter and more en­er­gy ef­fi­cient to fit in­to the small­er space­craft that the space agency was build­ing.

Galileo launched in 1989 with an 800 x 800 pix­el sen­sor, but in 1992, the pro­fes­sor's team de­vel­oped In­trapix­el Change Trans­fer, a method of re­duc­ing noise and am­pli­fy­ing sig­nal.

This be­gan the era of a cam­era sys­tem on a chip, where the chip it­self car­ried the tech­nolo­gies need­ed to de­code the light it was cap­tur­ing.

His work hav­ing ar­rived at a sat­is­fac­to­ry stage for NASA's needs, JPL took the next step with its dis­cov­ery and be­gan the process of de­vel­op­ing tech­nol­o­gy de­vel­op­ment agree­ments with the gi­ants of the day, AT&T, Ko­dak, Schick Tech­nolo­gies and Bell Labs through Cal­Tech.

The tech­nol­o­gy trans­fer proved slow and re­quired con­stant in­ter­ven­tion by the team at JPL, so NASA took the un­com­mon step in 1995 of al­low­ing the frus­trat­ed team to li­cense their in­ven­tion and mar­ket it as Pho­to­bit.

Pho­to­bit ran as a self-fund­ed con­cern from 1995-2001, fu­elled by de­sign con­tracts with pri­vate in­dus­try and filed 100 new patents on imag­ing tech­nolo­gies.

It was ac­quired by Mi­cron Tech­nolo­gies in 2001 and the in­tel­lec­tu­al as­sets re­vert­ed to Cal­Tech, but by then, 30 com­pa­nies were work­ing on CMOS based tech­nolo­gies.

"A patent nev­er stops any­one from steal­ing your idea," Prof Fos­sum not­ed wry­ly, "but you may be able to make them pay."

To­day, Sony and Sam­sung are the world lead­ers in CMOS chip pro­duc­tion, ranked at one and two, re­spec­tive­ly, and two bil­lion dig­i­tal cam­eras are be­ing pro­duced every year, rough­ly 60 cam­eras per sec­ond.

Cal­Tech has suc­cess­ful­ly en­forced its patents against mul­ti­ple play­ers and now the sto­ry is com­ing full cir­cle, and NASA is now us­ing the tech­nol­o­gy in space.

As Prof Fos­sum gets ready for his third re­tire­ment, he is fas­ci­nat­ed by the im­pact of imag­ing tech­nol­o­gy, which he has seen bring new so­cial is­sues, vi­su­al over­load, rapid so­cial change, in­stant com­mu­ni­ca­tion and some in­ap­pro­pri­ate use since its in­tro­duc­tion.

He's al­so work­ing on a new tech­nol­o­gy, the Quan­ta Im­age Sen­sor (QIS), which promis­es to de­liv­er dig­i­tal im­ages with a more film-like ex­po­sure, through cap­ture sen­sors that re­spond to light more like tra­di­tion­al emul­sions did, putting a bil­lion sen­sors on a chip while pulling a watt of pow­er.

But he's al­so look­ing at the di­rec­tions that com­pu­ta­tion­al imag­ing is tak­ing, with tech­nolo­gies like light field tech­nol­o­gy (ex­em­pli­fied by the Lytro cam­era) invit­ing em­pha­sis now.

"Un­der­stand­ing busi­ness isn't rock­et sci­ence," he ar­gues, "it's a lot eas­i­er than en­gi­neer­ing."When there's a need that peo­ple want sat­is­fied, when you can an­swer, who is go­ing to buy this. There is your op­por­tu­ni­ty."

Look­ing back at his work with the CMOS sen­sor, he boiled his learn­ings down in­to five brief sen­tences:

�2Cre­ate the in­ven­tion.

�2Suc­cess­ful­ly com­mer­cialise it.

�2De­liv­er on the promise com­pelling­ly.

�2Sell the com­pa­ny.

�2De­fend the patents.