beschizza at January 21st, 2014 10:07 — #1
telecinese at January 21st, 2014 12:10 — #2
Keep at it guys, we're not quite out of the valley yet.
Interesting that only the very latest body shape seems to be modeled in actual human anatomy instead of endless refinement of the v1 caricature, and how much visual identity is lost as a result. I wonder how long before the trend reverses as people realize the ridiculous sexpot cartoon proportions make the character much more recognizable than any amount of realism.
telecinese at January 21st, 2014 12:14 — #3
Also, yay for evolutionary design charts.
hallam at January 21st, 2014 13:25 — #4
The models have to be rendered in real time on reasonably common PC platforms. The game consoles have only just become capable of supporting the latest version of TR properly and even then only at HD resolution at best.
That is why there is so much difference between the in-game and box art figures in the early versions.
ianmcloud at January 21st, 2014 13:44 — #5
Sadly though, #1, #2, and #3 are far superior to the later Mercedes models. My personal favorite is #2.
nox at January 21st, 2014 14:28 — #6
Interesting thought, though recognizably of character is of questionable value as evidenced by the GTA series.
I would be interested to see a similar comparison of her breasts and hips though.
beschizza at January 21st, 2014 15:44 — #7
I think you could fit 2013 Lara's entire head inside of 1996 Lara's breast.
seki at January 21st, 2014 17:05 — #8
What the heck is wrong with the third one's face? Her features look all wonky and especially creepy.
peregrinus_bis at January 21st, 2014 17:08 — #9
echolocatechoco at January 21st, 2014 20:22 — #10
Not only that, until until HD television became commonplace console games were rendering to a screen resolution of 640 x 480 at best, with interlacing and blur hiding a lot of the detail on top of the already jaggy triangles and dodgy texture mapping.
Character models had to have very exaggerated features to read properly through all the noise.
It's the same way that blocky old video game sprites never looked like that back in the day--all the edges were smoothed off in the passage to a flickering cathode ray tube.
anthonyc at January 21st, 2014 20:53 — #11
Ok, not really, I don't actually care that much.
themudshark at January 22nd, 2014 04:46 — #12
I.m.o. we´re very much out of the valley, I can find nothing creepy about 2013 Lara.
telecinese at January 22nd, 2014 06:52 — #13
Even while moving? Because to me that's the last barrier between 'fully convincing human' and 'creepy living realistic wax model'. Or 'animated corpse', as if I recall lies very near the bottom of the classic uncanny valley graph.
Playing LA Noire was a good example to me. At first you're all 'neat, It's yet another guy from Mad Men' and it's pretty fun to recognize real faces. But for all the touted motion capture advancements, when they start speaking in close-up, there's a very real amount of je ne sais quoi missing.
telecinese at January 22nd, 2014 07:21 — #14
For all the increased focus on plot and character in GTA, I don't think the protagonists are meant to be iconic characters as much as representing familiar archetypes from movies and TV that are relevant to the chosen setting.
Did you notice the Lara Croft chart comes from a costume resource? Tank top and shorts in the v1 colors and gun holsters in combination go a long way for a Lara costume (I always felt that getup was meant to be a 90s 'hip' version of Indiana Jones' hat and whip).
It would take a very discerning eye to recognize most GTA-themed halloween costumes, I think. If anything people would more likely recognize the 'original' guy from Miami Vice, Boys'n'the'Hood, Heat or Sons of Anarchy they seem to be modeled after.
chentzilla at January 22nd, 2014 08:57 — #15
How would you even scale one against the other?
echolocatechoco at January 23rd, 2014 15:05 — #16
LA Noire is an especially odd one because the facial animation is essentially video capture--there's geometry deformation, but it's captured directly from real actors and not "re-targeted" onto a character model.
The animation of the characters' bodies is done separately with traditional motion capture and the two combined later. It works pretty well most of the time but you're still left with this feeling of a real man wearing an ill-fitting man suit--especially as the character designs and lighting are incongruously stylised compared to the faces. There's very little crumpling of fabric in their sausagey arms, for example.
The most startling thing for me about LA Noire was seeing the tendons and muscles in the actors' necks. You never see that in game animation.
telecinese at January 23rd, 2014 16:09 — #17
I thought it was a good step forward in technology, and it's pretty obviously superior facewise to the cutscenes in GTA V, for instance. But as I said, to me there's a whole lot of climbing ahead in terms of convincing realistic humans before the summit of canny hill.
hallam at January 25th, 2014 12:29 — #18
Thats not sexist though. For sexist, try the nuderaider patch.
beschizza at January 26th, 2014 10:07 — #19
This topic was automatically closed after 5 days. New replies are no longer allowed.