Unlimited Detail - gaming / CGI future or vaporware?

[Jervan]

http://www.youtube.com/watch?v=Q-ATtrImCx4&feature=player_embedded#

Post speculations

[Ad Bot]

[Thorax]

Interesting. Seems like it would work, but wouldn't it need the same amount of processing power as current technologies already have?

[yuki]

Interesting. Seems like it would work, but wouldn't it need the same amount of processing power as current technologies already have?

No, it would need even more processing power.

[Skele]

That's pretty impressive.

[JJA0]

I'm not following this well.

Basically only the dots your looking at are computed? The others float off somewhere else?

[heedicalking]

It seems cool, but I don't think building things dot by dot is a progressive way to model. I would hope there would be some kind of model program to make it easier.

[dkamm65]

No, it would need even more processing power.

Less. Did you even watch the video?

[Skele]

It seems cool, but I don't think building things dot by dot is a progressive way to model. I would hope there would be some kind of model program to make it easier.

Obviously...

[Sirrus]

Yeah, as a mapper I can't say I'm excited about this. I just don't see a practical approach to mapping/modeling with point cloud.

Edit: Obviously I'm excited about the end products, but what I'm not too happy about is how someone like me could make something using this system.

Also, another question, what's the difference between hardware and software rendering?

[yuki]

Less. Did you even watch the video?

Yes, saw it yesterday. I was referring to the original idea that it would need more porcessing power, not that UD has created a method for it to use less.

It seems cool, but I don't think building things dot by dot is a progressive way to model. I would hope there would be some kind of model program to make it easier.

The site says it has converters so you can import polygon models and export point-float.

Basically only the dots your looking at are computed? The others float off somewhere else?

It draws points, instead of those tiny triangles.

Yeah, as a mapper I can't say I'm excited about this. I just don't see a practical approach to mapping/modeling with point cloud.

Edit: Obviously I'm excited about the end products, but what I'm not too happy about is how someone like me could make something using this system.

Obviously it's in their best interest to make using this tech user friendly.

[Sirrus]

Obviously it's in their best interest to make using this tech user friendly.

Obviously.

[tay104]

neat-o

[rkynick]

Mapping / Modeling with point-clouds wouldn't be extremely different.
You could model as you do now and have it changed into a point-cloud model. A face would just be made into a large number of points on a plane.

Mapping would probably be similar to Torque's(and others but this example comes to mind) terrain system of a plane of points raised and lowered with faces drawn to and from each one. It'd be the same thing, take out the faces and up the points. Mapping would be more powerful and smoother.
And, a height map could be more accurate and easier to use, every pixel could be a point (then of course the heightmap will be very very massive, but it could be divvied into sections)

Also detail mapping and all that good stuff would be interesting; it could be transferred into actual 3d displacement.

These concerns are unfounded.

[Tokthree]

I'm not following this well.

Basically only the dots your looking at are computed? The others float off somewhere else?

It probably dynamically renders those which you can see, and hides those which you cannot.. When you think about it that's an astonishing thing, it doesn't take much processor power to render 1280x1024 pixels.. The only major CPU usage would be the actual loading/rendering algorithms.

I am looking forward to future developments on this, quite immensely

[Radial543]

Sounds great, I can't wait to mess around with the soon-to-be-released SDK.