Animated displacement & normal maps for a CGI character

hisshadow

Hi guys
I'm coming to the end of my dissertation project and I'm looking for some critical feedback. The project is about procedurally combined, animated displacement & normal maps for a CGI character. The idea is you can activate as many or as few expressions at different levels of intensity at the same time. Its not just blended...

The purpose of the animation is for incremented wrinkling etc. So if the forehead wrinkles up slightly its not five faint lines but one or two, appearing one at a time in a more natural way.

Maya Version
[ame="http://www.youtube.com/watch?v=KS9Y_dggIwk"]MSc Project - 3D Head - Maya - YouTube[/ame]

UDK Version
[ame="http://www.youtube.com/watch?v=6ryOXKUJ058"]MSc Project - 3D Head - UT3 - YouTube[/ame]

Let me know what you think

sltrOlsson

The smile looks extremely creepy! He's not smiling at all with his mouth, Just looks very strange. Other then that, cool stuff man

Ace-Angel

So...what would you like us to tell you?

Wesley

Er... he asked for critical feedback and any thoughts?

gauss

save out the UDK version again with a less extreme fov. notice how it looks like a normal head in the maya version but in the udk shot the nose is a mile long and the eyes are bugging out? too close, too wide of an fov. results look pretty good in the engine but the fov makes it too distracting.

otherwise really interesting stuff, nice work

Ace-Angel

Wesley wrote: »

Er... he asked for critical feedback and any thoughts?

Without any technical information, I can't tell if the setup is optimized and what the limitations are, just showing people a bunch of videos doesn't help. There is a reason stuff like white papers at least try and explain the 'idea' behind the tech being used without outright giving the information, so other people can implement it, if they have the skill or give back critical feedback in a more constructive way.

For example, I don't see any difference between this and a 'blended' normal approach, other then it having a much 'quicker' and harder change pace, which can also be done for blends with extreme values, time variations and exponent pulls.

Having each wrinkle show separately on the forehead? You can use procedural masks for that which is tied with your 'wrinkle' amount constant, have it mask out exponentially so by the time your reach 1 = full wrinkle, you have mask = 0, etc, in a gradient fashion.

So...yeah...