Did you find an example where the compensation for different resolutions is not perfect?
No, it is perfect
It was designed to provide the same shadings, no matter what resolution or supersampling is chosen. (Technically, the slopes are normalized with image width.)
MM calculates the slopes based on the 2-dimensional derivative of iteration count. By zooming in 1 step, the distance of the pixels doubles, hence the derivative halves. I think this is OK for zoom movies as long as the bump mapping is performed for each frame independently, not just for the key frames. This will make movie rendering a bit slower though, since all key frame scaling would have to be performed on iteration values, not color values. But this will also be necessary for the histogram coloring of movies. KFR Movie Maker already works this way, right?
But it is zoom level dependent, which you also mentioned.
KFMM converts several key frames to memory bitmaps and uses Win32 stretchblt to squeeze them and paste them on top of each other. I think the standard Win32-Stretchblt is very good for antialiasing 2D fractals, the result doesn't get as blurry as the extensive photo-resizing methods used in Photoshop etc, and therefore preserves a little more information. But that is only my opinion.
So when KFMM paste several stretched images on top of each other, the edges are very visible when slope/bump-encoding is used...
December 20, 2008, 03:57:59 AM
Dilber MC Theme by HarzeM