FLASH FOR­MA­TIONS: a brief conversation with Texas In­stru­ments

Sound+Image - - Test - JF

It sounds easy enough — use four flashes of 1920 x 1080 to make the UHD res­o­lu­tion of 3840x 2160. The num­bers are easy. But the ge­om­e­try is less ob­vi­ous. Af­ter sig­nif­i­cant de­bate be­tween au­thor and Edi­tor over the pos­si­ble im­ple­men­ta­tions, we came up with three po­ten­tial an­swers.

The first is the sim­plest. The DMD could de­liver 1920 x 1080 to four en­tirely sep­a­rate quar­ters of the screen. But this would re­quire mov­ing the en­tire DMD around sig­nif­i­cant an­gles, and might also cre­ate bor­ders or po­ten­tial time-smear­ing as the sep­a­rate sec­tions ap­peared. So un­likely.

Sec­ond: the pix­els on the DMDs might oc­cupy only a quar­ter of the mi­cromir­ror area, so they can be shifted half a mir­ror in each di­rec­tion to de­liver four ad­ja­cent pix­els. But this would make the pix­els tiny, po­ten­tially af­fect­ing bright­ness.

Or third, full-size pix­els (fill­ing most of each mi­cromir­ror) are de­liv­ered to four sep­a­rate po­si­tions that over­lap. This would seem to re­quire im­age pro­cess­ing of each frame prior to de­liv­ery (to cal­cu­late and negate the ef­fects of over­lap­ping), but then this ge­om­e­try is the only way to make sense of pre­vi­ous DMDs which dou­bled 2716 x 1528 to make 4K.

We con­tacted Texas In­stru­ments, cre­ators of the mirac­u­lous DMD chips. “The DLP470 TE and DLP470TP4K UHD chip sets dis­play four pix­els on the screen per mi­cro mir­ror ”, they said, and when we asked for fur­ther clar­i­fi­ca­tion, they

replied that: “

ar­ray of mi­cro mir­rors, which switch on the or­der of The DLP470 TE uses a 1920x 1080 mi­crosec­onds, to cre­ate four dis­tinct ad­dress able pix­els on the screen dur­ing every frame to de­liver full 4 K UH D res­o­lu­tion ... The com­plete set of pix­els are over sam­pled and pro­cessed by unique TI im­age pro­cess­ing al­go­rithms lead­ing to a crisp, col­or­ful& de­tailed full 8- megapixel 4 K UH D dis­play .”

Can you con­firm that the ad­dress­able pixel spa­ces over­lap each other? Is the im­age pro­cess­ing re­quired to pre­dict the over­lap ef­fects?

“I ap­pre­ci­ate your in­ter­est in how our new­est 4 K UH D tech­nol­ogy works ,” replied our cor­re­spon­dent ,“butt he DLP im­age pro­cess­ing al­go­rithms are pro­pri­etary and I’ m afraid I can’ t go into any­more de­tail .”

But from that, to­gether with a di­a­gram from Ben Q, we reckon it’s op­tion 3 — over­sam­pling and pre-pro­cess­ing, then 240Hz flash­ing to four clock­wise-ad­ja­cent po­si­tions, with over­laps.

Newspapers in English

Newspapers from Australia

© PressReader. All rights reserved.