rendering points with a fixed pixel size

[jonp]

Is there any way to render points using Mantra at a consistent image pixel size, regardless of their distance?

[Mario Marengo]

Is there any way to render points using Mantra at a consistent image pixel size, regardless of their distance?

In a shader, it would be something like:

max(1e-6,width*Pz);

But because this has to happen at the attribute level, you'll have to calculate Pz in SOPs when assigning the width attribute. (BTW, Pz represents the distance to the camera plane, not to the center of projection).

[Jason]

Is there any way to render points using Mantra at a consistent image pixel size, regardless of their distance?

Not at the flick of a switch, nope.

But you could project the points onto a plane in front of camera, perhaps? (Be careful of velocity motion-blur, of course)

Or you could change the width of the points to counteract perspective? (EDIT: Mario answers this, I see)

Cheers,

Jason

[jonp]

I guess that's the way to go, but to keep it from being dependent on any particular camera, I'd need have access to some kind of variable that tells where the Eye is in SOP space.

[Jason]

I guess that's the way to go, but to keep it from being dependent on any particular camera, I'd need have access to some kind of variable that tells where the Eye is in SOP space.

Yeah, you could construct a SOP HDA that you can point to your render camera. (Magic math not included;))