CompuServe Thread

I have tried doing that, ie using displace for brain mapping, but ran into a number of problems. However, I have developed some close approximations of many levels of the brain which I may be able to assist you in rendering. The first question is what view, section, and areas are you trying to represent? Are you trying to view the whole brain? the coronal sections, axial sections etc, etc ,etc. Anyway give me a better idea. Sam Medical Media

PMJI, but in reference to constructing a brain (or other 3D detailed objects) using Displace.pxp…..here's a technique I used months ago for a baby doll model. It's sort of a "poor man's 3D digitizer", and I've meant to write about it here before, but never did. Anyway, usually detailed * photographs * used lighting which was at some angle wrt the view, so using them for bump maps isn't always ideal. This is a little bit of work, but try this if you can obtain a real-world model of a brain: 1) Take the subject object to be scanned and paint it flat white. There are various types of paint which will come off later pretty well if necessary. 2) Using any sort of scheme for registration, place the brain (or most any object) on the bed of a common 24-bit scanner. Using the registration scheme, carefully scan front, sides, top & bottom views at pretty high resolution in greyscale. 3) In a good paint/processing program, crop each image tightly, convert to 24-bit RGB format, process for brightness & contrast, filter if necessary, and fill in the background to make sure it's consistently black. 4) Then, scale & distort the views as necessary for spherical mapping, & combine & blend them into one large displacement map. Touch-up as necessary. 5) Then apply as a spherical map in Displace.pxp & adjust surface detail height as necessary. Obviously, making the map is a lot of work and can never be perfect. But since the object is lit by the scanner face-on, it produces a correct bump or displacement map. You might also try using the undistorted/unscaled 6 views colored/shaded in a paint program, and box-mapped onto a simple "brain object" which initially is more angular. Then deform it more to a rounded shape after the mapping is applied. The above scanner digitizing technique of course can be used more easily & successfully with a more hemispherical object, like a ceramic mask or whatever. Unfortunately, every color scanner I've used will create all sorts of color fringing if in color mode & if the subject is not very flat artwork. Otherwise, a second color pass will allow simulation of a "poor man's Cyberware Scanner" <g>. However, with a little more work and the required equipment, you can: 1) Photograph the subject in color with very diffuse/indirect light to avoid specular highlights & other reflections. 2) Compose and print for a 1:1 scale print. 3) Use any registration scheme, scan the greyscale displacement map, then scan the photograph. Using a paint program like Painter X/2 with a translucent overlay layer, adjust the color scan as close to possible to the greyscale scan, & touch-up as necessary. 4) Displace with the greyscale map & apply mapping from Displace.pxp. This can be a LOT of work, has to be precise, and is rarely perfect. But I did have pretty good results 'way back when I was foolin' around with this.

(Was that pronounced "ow", or "ooooh"?) We bought their "R" brain mesh (R stands for "ridiculously-high" vertex/poly count, according to one of their anatomical modellers), for $400 bucks, last year. I just talked to Viewpoint, and that's the same mesh that Casey was talking about … which is not really very detailed in the interior. RE: mesh "upgrades" – Casey said Viewpoint is working out an upgrade policy "even as we speak". Stay tuned.