Dazzling pyrotechnics on a handheld 3D budget with OpenGL ES

[AdamBailey9]

Since the explosion is billboarded it could be precomputed back into a single texture animation.

[angelorohit]

Very awesome! I would love to try out these tricks and see what speed boosts I can get.

[cyborgtroy]

The fake global illumination looks cool. I'm going to have to try some of these tricks once I get my OpenGL stuff going.

[cyborgtroy]

So this video recommends doing CPU lighting? Or.. something..

[jordicoma]

Look this for the gp2x wiz, it will have opengl 1.1 es capable hardware!

[cyborgtroy]

Pandora supports OGL ES 2.0 IIRC. Sadly, it is quite expensive and still being produced. [Designed, not quite all shipped]

[icyresh2]

how do you egt opengl support your pc

[AndyGoRAWR]

Well if phones can do all of this with suhc high framerate, why are the only games i know of are snake, tetris, zuma, and other shit lol. I'd love an awesome 3D game racer or something on my phone lol.

[cyborgtroy]

It's because cell phones are a pain to develop for.

a PAIN.

[m1ck3n]

Nvidia the way it´s ment to be played it sure aint no bullshit

[hausenharry]

would rather look at smoke

cheaper easier

and a lot less stressful

and not nearly as creepy

good vid though!

[a1mint]

NVidia rules

[bajan13k]

Floating point requires a lot more hardware (transistors), thus more power. For mobile chips, power consumption is at a premium so even if there is floating point in hardware, it wouldn't be the power-hungry fast type, it'll be the slow type.

Integer is a lot faster not only because u don't have to shift the point, but you can multiply faster by using a combination of shifts and adds, thus allow the removal of a multiply (and divide) unit. It also draws less power.

[h2n0n]

Very god video! Very Interesting!!!

[chnemesis19]

ur joking right?

[FrankRizzo6662]

It's "you're", learn to type

[vgexpert]

nice

[RamonesKaraoke]

FPUs add size and hence cost and hence power consumption to your mobile phone CPU so have often been omitted, fixed-point is way faster on these CPUs, typically 6-10x.

The nasty case is floating-point add on an integer-only machine - its a single-cycle add in fixed-point, but for floats there's separation of mantissa and exponents, fix up based on sign bits, shift based on difference of exponents, add, post-add shift to renormalize...

[3DGeezer]

BOOM goes the dynamite!!!

[baileymog]

Nice refresher thanks man... :) encore encore