Previous | Next --- Slide 31 of 69
Back to Lecture Thumbnails
tib

I'm a little confused as of which coordinate system are we using that caused this error. Would the order be as follows? 1. Project 3D coordinates (that have varying depth) onto a 2D coordinate (screen coordinates) 2. Use barycentric coordinates to interpolate on the 2D coordinates 3. Creates an incorrect interpolation?

I think I missed the part between step 2 to step 3 on why it would fail to interpret the values.

keenan

@tib After your step 2, you would just use the interpolated 2D coordinates to do a lookup into a 2D (checkerboard) texture. It wouldn't fail to interpolate the values (i.e., you would still get a piece of a checkerboard), but the rate of change of the UV coordinates will disagree across triangles, as shown in the image labeled "Affine."

tarangs

In this case, i guess it helps with the intuition if we assume we are "constructing" the checkerboard.

We construct the checkerboard by taking the 2 vectors(drawn in the image) and dividing each side in equal segments along those vectors.

The opposite vectors should be anti parallel in reality but the projection breaks their anti parallel-ness, so the constructed checkerboard in the projected space is not aligned.