Talk:Resize

From Avisynth

You have to set "b + 2 * c = 1" for the numerically most accurate filter. (Then why does b=0,c=0 have the highest PSNR?) This gives for b = 0 the maximum value for c = 0.5, which is the Catmull-Rom spline and a good suggestion for sharpness.

Please you the corresponding talk page for questions like this. I have no answer to this. You might try asking on doom9 forums. Wilbert 12:06, 24 July 2008 (PDT)

[edit] Spline16/36/64 coefficients

The Spline16/36/64 coefficients appear to come from Panorama Tools, but I am unable to derive the same coefficients in Scilab using the given conditions (derivative=0 at edges). I'm inclined to believe that the splines were not derived correctly, because they are incorrect in several cases -- Spline16 is discontinuous in the first derivative between the two splines and Spline64 has non-zero derivative at the edge of the filter. I doubt this would have an impact on visual quality, but it might affect post filters that used gradients. -- Phaeron

Funny that you are posting about this. I contacted the author of Panorama Tools a few days ago asking about those coefficients. He gave me enough info to derive the coefficients myself. So, i'm looking at it for the last three days. When i have it all on paper i will contact you about it! Wilbert 16:20, 14 May 2009 (UTC)