seeking details on perturbed normals mapping

[darkblu]

Simon, Kristof,

any chance you have lying around the 'Bump Mapping For Series2.doc' from the PowerVR Series2 SDK? or links to the original Blinn's paper?

[Simon F]

Simon, Kristof,

any chance you have lying around the 'Bump Mapping For Series2.doc' from the PowerVR Series2 SDK? or links to the original Blinn's paper?
Blinn's original paper? I've only ever seen that on the ACM website but I'll have a look.

[darkblu]

thanks, Simon!

[darkblu]

btw, Simon, did series2 employ linear interpolation for normal maps? IIRC, normals in those maps were in polar coords, right?

[Simon F]

btw, Simon, did series2 employ linear interpolation for normal maps? IIRC, normals in those maps were in polar coords, right?
You're stretching my memory here but it was roughly based on polar coords. I initially thought this would be cheaper in HW than cartesian coords (with the cost of renomalisation etc) but I soon realised I was wrong. By then it was too late to change the HW design. <shrug>

[darkblu]

btw, Simon, did series2 employ linear interpolation for normal maps? IIRC, normals in those maps were in polar coords, right?
You're stretching my memory here but it was roughly based on polar coords. I initially thought this would be cheaper in HW than cartesian coords (with the cost of renomalisation etc) but I soon realised I was wrong. By then it was too late to change the HW design. <shrug>

from what you say i infer series2 did employ linear interpolation while sampling in a normal map, yes? so how did you handle interpolating between opposite vectors in polar coords? *

* correct me if i'm wrong here: those polar coords had an implied vector norm, i.e. the texture stored just the two angles whereas norm was implied to be a unit?

ps: hope i'm not tiring you, the motives behind my bugging is that i just did the same choice of normal maps in my sw rasteriser as you did back then, and i'm pondering over the consequences now ; )