Top Tech Tools and Trends of 2018

Animation Magazine - - Vfx - By Todd Sheri­dan Perry

We are back again with a recap of the best tech of 2018. As al­ways, some things are new, some have been around, and some we’ve just start­ing to in­cor­po­rate into our tiny lit­tle world of an­i­ma­tion and film­mak­ing. Vir­tual Film Pro­duc­tion. With game en­gines and real-time ray­trac­ing, vir­tual pro­duc­tion is hit­ting its stride, and will prop­a­gate its way back­ward through pre-pro­duc­tion and con­cept, all the way back to pitch­ing a film. And ideally, the tools on the ac­tual pro­duc­tion mi­grates seam­lessly to post-pro­duc­tion.

Real­time Ray­trac­ing. Nvidia re­leased its Quadro RTX graph­ics card this year at SIGGRAPH. Too many ap­pli­ca­tions to list, but the afore­men­tioned vir­tual pro­duc­tion is a key use, as is rapid con­cept de­vel­op­ment. See­ing your changes in real time with re­flec­tions, DOF and GI is a game changer.

Sharing. With NASA re­leas­ing its archives, Dis­ney of­fer­ing up its tech­nol­ogy as open source along with ac­tual pro­duc­tion as­sets like the is­land in Moana, more com­pa­nies are show­ing that sci­ence can’t grow with­out col­lab­o­ra­tion, and art can’t grow with­out sharing.

Quadrupedal AI Mo­tion Anal­y­sis. Yet an­other ex­am­ple of us­ing ar­ti­fi­cial in­tel­li­gence to learn com­plex be­hav­iors.

3D Metal Print­ing. No more plas­tic 3D prints! We needed some­thing stronger, and we got it.

Sen­sor City. We have seen the fu­ture, and it in­volves city plan­ning us­ing sam­pled big data from sen­sors to best de­sign it for the com­mu­nity. Univer­sal AI. Yes, they were once the play­ground of the rich and well-funded, but AI and ma­chine learn­ing have be­come more ac­ces­si­ble thanks to the ef­forts of Ama­zon’s AWS, Google’s Cloud ML and Cloud AI tools.

GANS: Duel­ing Neu­ral Net­works. Gen­er­a­tive ad­ver­sar­ial net­works are what hap­pens when two AI sys­tems are in ca­hoots. One tries to make fake images based on its learned sam­ple set. The other gives it con­struc­tive feed­back based on the same sam­ple set. Both learn more as one gets bet­ter at fak­ing it.

Hou­dini 17. Vel­lum, char­ac­ter tools, UV tools, bet­ter fluid sims, groom­ing tools, fa­cial auto-rig­ging, built-in Bul­let Solver soft con­straints for col­li­sion, ma­te­rial-based frac­tur­ing ... The list goes on and on. What’s not to love?

Thanos. Dig­i­tal Do­main cre­ated a fa­cial mo­tion-cap­ture sys­tem that uses deep learn­ing to fi­nesse in the fine de­tails of the year’s big­gest block­buster vil­lain. Dis­ney Re­search’s Stun­tron­ics. Ev­ery­one says they love real stunts. But Dis­ney Imag­i­neer­ing gave us au­tonomous ro­bot stunt dou­bles this year. Does this mean less work for real-life stunt pro­fes­sion­als?

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.