PDA

View Full Version : Nvidia faceworks video



Ernest
03-22-2013, 01:48 AM
A video from the GPU technology conference a couple of days ago, showing a realtime 3D head (no, not the fairy, a little after that).

http://www.youtube.com/watch?feature=player_embedded&v=5d1ZOYU4gpo


On a totally different topic, did you see the mess that happened because of a joke about dongles? I think half this forum needs to get fired, me included!
http://techcrunch.com/2013/03/21/a-dongle-joke-that-spiraled-way-out-of-control/

cresshead
03-22-2013, 06:39 AM
skin and lighting were excellent but the eyes looked poor. the animation looked canned but was fairly convincing.

my main problem with their approach is that it only recreated existing humans that are scanned...so you won't get believable results unless you
can scan someone...so why not just film them...why bother with 3d...

safetyman
03-22-2013, 07:20 AM
Good point Cress -- unless it's really really cost effective to do 3d humans or if it can be done very easily there's not a lot of benefit. Safety of the actors, virtual stuntmen, or actors in situations where it would be impossible to do in camera seem to be the most common situations at the moment.

cresshead
03-22-2013, 08:43 AM
yeh 3d stuntmen are an excellent use for this technology

prometheus
03-23-2013, 05:43 AM
yepp very nice, I was wondering if they are talking about the expressions saved, and being able to reuse and combine in such way that you could extract it on to other figures, otherwise it isnīt of very much use for most people not having acess to all equipments around.

Rendering of shadows/lights, skin etc and movements in face are absolutly great, and as mentioned eyes doesnīt look good, but it isnīt anything fault with the render of it, just that from such
demos ..you always have the character with eyes starring out at nothing and not following a human interaction with another character or or following the camera interviewer, ergo dull and cold computer eyes, but
that has nothing to do with the rendering..just interaction movement.

Michael

prometheus
03-23-2013, 05:46 AM
yepp very nice, I was wondering if they are talking about the expressions saved, and being able to reuse and combine in such way that you could extract it on to other figures, otherwise it isnīt of very much use for most people not having acess to all equipments around.

Rendering of shadows/lights, skin etc and movements in face are absolutly great, and as mentioned eyes doesnīt look good, but it isnīt anything fault with the render of it, just that from such
demos ..you always have the character with eyes starring out at nothing and not following a human interaction with another character or or following the camera interviewer, ergo dull and cold computer eyes, but
that has nothing to do with the rendering..just interaction movement.


Forum time out double post cackup as usual:(

Michael

safetyman
03-23-2013, 06:34 AM
yeh 3d stuntmen are an excellent use for this technology

LOL ! Yeah, real efficient use. Much better, however, than the video conferencing scenario that the CEO outlined. A better use would be as a "face" for your computer: Instead of a Blue Screen of Death, you get a 3D rendered head representing your computer that says, "You're screwed".

LW_Will
03-23-2013, 09:30 AM
I'd like to see the tech applied to Jeff Bridges. That would amp up Flynn... and give more realism to TRON...

JCG
03-23-2013, 05:52 PM
my main problem with their approach is that it only recreated existing humans that are scanned...so you won't get believable results unless you
can scan someone...so why not just film them...why bother with 3d...I think the 3 main reasons would be: file size, resolution independence, and sequels.

If you film every possible movement of each of your game characters, the file sizes of the complete games are huge. The low resolution, fairly small FMV games of the 90s were already several GBs. Saving the likeness and expressions to 3D models uses much less space.

People have monitors with different native resolutions, but if you film all your characters, then they will have a fixed resolution and they will look stretched or fuzzy in other resolutions, or will require black borders. If you save the data to 3D models, it can be rendered at any resolution so everyone can play the game at their native res in full screen.

Games take a long time to be developed and they depend on a long number of sequels. If you film your actors, it's likely that you'll have to change your characters' appearances by the 3rd sequel.