<br><br><div class="gmail_quote">On 27 October 2010 01:43, <span dir="ltr"><<a href="mailto:jimcp@cox.net">jimcp@cox.net</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Paul,<br>
<br>
I have no doubt we have a shared knowledge gap, and most of it is probably on my side. and to be sure I do not have a comprehensive understanding of vtk.<br>
<br>
When we started on this, you had a problem with "missing or dark triangles in some models", and a problem with inverting models that resulted in dark polygon displays. You provided an illustrative example with two generated cones. One of which displays black triangles when any one axis is reversed using an actor defined transform. this was assumed to be a simulation of the right handed vs left handed model problem and was addressed by applying the transform to the polydata instead of the actor. That was to address the problem I thought we were addressing was where "some of my models are right handed and some are left handed". in that case, My opinion is you need to deal with that difference by individual model at the polydata level. Unless you disagree, I think we have arrived at a resonably conclusion in this issue.<br>
<br></blockquote><div><br>Sorry, that wasn't the problem I was trying to address. All of my models share the same handedness, but the user may want to mirror the view.<br> </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Now focus has moved to the yet to be demonstrated problem of "Some of my models normals are inverted". I am not an expert in this area, but I think that class of problem would be addressed in vtk by reducing the contour to it's component points, and then generate a new polygon mesh with newly computed normals. If you have a readable file you would like to post as an example of this issue, I don't mind having a look.<br>
</blockquote><div><br>I now do not think that is a problem. All of my models have normals pointing all over the place, and the models should not need to be fixed, as I want to see both sides of each polygon anyway (eg planes).<br>
<br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
finally we have a problem "I need to display a mirrored world view". I am imagining a display mounted in a kiosk with a mirror reflecting the image. I will have to defer to other readers, I agree this might be a camera or renderer function. I hope they are still reading....<br>
<br></blockquote><div><br>Ideally you just add a UserMatrix transform on the camera, thereby reversing the entire scene.<br><br>The problem that you have not listed here is the one I'm trying to highlight - twosided lighting equations in the presence of UserMatrix transformations.<br>
<br></div></div>