This is visibly broken on http://romaxa.bolshe.net/css3d/morphing/morphing-cubes.html

Comment on attachment 566116 [details] [diff] [review] Check the Z scale of the inverse instead Review of attachment 566116 [details] [diff] [review]: ----------------------------------------------------------------- r+ with a couple minor suggestions. ::: gfx/thebes/gfx3DMatrix.cpp @@ +836,5 @@ > + gfxFloat det = Determinant(); > + float _33 = _12*_24*_41 - _14*_22*_41 + > + _14*_21*_42 - _11*_24*_42 - > + _12*_21*_44 + _11*_22*_44; > + return (_33 < 0 && det >= 0) || (_33 >= 0 && det < 0); I'm a little confused by the usage of >= 0 here, as that will make things pass the wouldn't have passed the Inverse()._33 < 0 test. But, I'd propose a much simpler test: _33*det < 0. It will have the same sign as _33/det, even if the magnitude is completely wrong. ::: gfx/thebes/gfx3DMatrix.h @@ +298,5 @@ > } > > gfx3DMatrix& Transpose(); > gfx3DMatrix Transposed() const; > + Whitespace-only change. ::: layout/base/nsDisplayList.cpp @@ +2549,5 @@ > > +/* If the matrix is singular, or a hidden backface is shown, the frame won't be visible or hit. */ > +static bool IsFrameVisible(nsIFrame* aFrame, const gfx3DMatrix& aMatrix) > +{ > + if (aMatrix.IsSingular()) { It's a little sad that we can't combine this test with the backface test, since they both compute the determinant, and that's the bulk of the work here.

(In reply to Timothy B. Terriberry (:derf) from comment #1) > > It's a little sad that we can't combine this test with the backface test, > since they both compute the determinant, and that's the bulk of the work > here. I think that if matrix maths ever starts showing up in profiles, then we can implement a dirty flag and start caching calculations like this.