ATI's new Ruby?

[corysama]

You are limited to the performance on the light stage.

If in the performance you put your hand in your hair, your hair will move and have volumetric shadows and anisotropic highlights and everything else real hair has because you recorded it while doing it for real.

If instead you animate a voxelated version of the data and you move the voxelated hand into the voxelated hair, nothing will happen. You might be able to add in voxel lighting, shadows, physics, etc through some other technique. But, the light stage data is prerecorded and fixed.


Edit: More rambling.
I think they /can/ use the light stage as a technique to capture hirez normal and albedo maps (and maybe even voxel positions). That is a really nice alternative to hand modeling and painting and deserves great kudos, but it can only be used to create data for a traditional lighting pipeline. By beef with OTOY's presentations is that they show the 100% accurate (prerecorded) lighting and then show the albedo and normals that can be used for not-100%-accurate simulated lighting. Both techniques come from the same tech, but the fact that they are separate and incompatible is quietly left unmentioned.

 

[Iron Tiger]

I'm fairly certain that's actress Ivana Milicevic. I've never found her to be attractive, so if she turns out to be the next Ruby, I'll be very disappointed.

 

[Ailuros]

wow, could you show us a photo, please?

Are you sure any minors aren't watching?

 

[thop]

I'm fairly certain that's actress Ivana Milicevic. I've never found her to be attractive, so if she turns out to be the next Ruby, I'll be very disappointed.

Prepare for disappointment then.

 

[spacemonkey]

definitely an improvement on...

 

[Babel-17]

Prepare for disappointment then.

She has great bone structure.

She was on Pushing Daisies.

 

[jlippo]

Disadvantages:
*The data set is huge -even after compression. The video mentions 21,000 images.
*The subject is static (Unless you can handle enough images to make a movie. Then you can have a pre-recorded video like the brunette OTOY showed off a while back.) Also, the camera is static. Basically, all that you can change is the lighting.
*No local lighting.
*No mesh to work with. Just photographs.

You can use the data as a texture, lightspace has been used in movies for years.
That many images mean you have blend shapes which you use for animating character.
You can use one image with 48 or 64 light directions for animation, but you do not get 'perfect' lighting.

 

[homerdog]

definitely an improvement on...

Yes it's definitely an improvement on that

And wtf guys she doesn't look that bad. Definitely much better as a blond though

 

[Kaotik]

Yes it's definitely an improvement on that

And wtf guys she doesn't look that bad. Definitely much better as a blond though

Say what?
She's way hotter as a redhead, no contest there.
Red hair is something that makes a lot of otherwise average looking girls instantly hotties.

 

[corysama]

You can use the data as a texture, lightspace has been used in movies for years.
That many images mean you have blend shapes which you use for animating character.
You can use one image with 48 or 64 light directions for animation, but you do not get 'perfect' lighting.

Yeah, but I think when they use it in movies, they fix the camera angle and only use it for relighting. If the camera could move then you would need to multiply the number of images by the number of potential camera angles that you want to blend between. Otherwise, the view-dependent (specular) effects would be obviously wrong.

 

[spacemonkey]

Yes it's definitely an improvement on that

Sorry, the image link broke. I was referring to the previous Ruby.

 

[homerdog]

Say what?
She's way hotter as a redhead, no contest there.
Red hair is something that makes a lot of otherwise average looking girls instantly hotties.

Normally I would agree but for some reason, not this time.

 

[jlippo]

Yeah, but I think when they use it in movies, they fix the camera angle and only use it for relighting. If the camera could move then you would need to multiply the number of images by the number of potential camera angles that you want to blend between. Otherwise, the view-dependent (specular) effects would be obviously wrong.

They do capture the information from several angles and thus can use the data on objects.
http://projects.ict.usc.edu/graphics/Research/afrf/

It was used in Spiderman2 to get proper reflectance fields for character faces.

 

[Pressure]

Say what?
She's way hotter as a redhead, no contest there.
Red hair is something that makes a lot of otherwise average looking girls instantly hotties.

I doubt a bloke from Scotland and Ireland would say that heh.

But yeah, redheads have something going for them

 

[Alexko]

Normally I would agree but for some reason, not this time.

Maybe because the blond is real and the redhead isn't

 

[spacemonkey]

Maybe because the blond is real and the redhead isn't

Well, both are images (just produced using different processes). Neither one is really "real"

p.s. there is no spoon

 

[Unknown Soldier]

But yeah, redheads have something going for them

+1

 

[hoom]

I also am a redhead fan but do agree that she looks hotter blonde.

Possibly its because the facial expression dots they stuck on her weren't blended out in the recreated image & instead look like zits?
(also my long hair fetish > redhead fetish )

 

[I.S.T.]

I also am a redhead fan but do agree that she looks hotter blonde.

Possibly its because the facial expression dots they stuck on her weren't blended out in the recreated image & instead look like zits?
(also my long hair fetish > redhead fetish )

That and they turned her skin fucking yellow!

 

[Jawed]

I'm waiting to see how OTOY animates her hair. I want no excuses about "concrete-lock" hairspray in the future

Jawed