[CinCV TNG] [glitch] zoom curve tangent & disjoint

Sam cinelerra at posteo.de
Fri Apr 27 00:21:09 CEST 2018


Thanks for the explanation. Making an additional data point shouldn't be 
a problem.

Sam

On 26.04.2018 19:55, Good Guy wrote:
> Hi,
>
> The automation curves in the timeline are evaluated in floatauto.C
> as bezier curves, which are cubic polynormals of the form:
> y = x**3*(-y0 + 3*y1 - 3*y2 + y3) + x**2*(3*y0 - 6*y1 + 3*y2) + 
> x*(-3*y0 + 3*y1) + y0
> y0, y3 = prev / next control values
> y1, y2 = prev / next derivitive values
>
> This operation is a cubic "curve fit".
> It does not fit an exponential function except in a few
> degenerate cases, like zero lengths, zero slopes.
>
> To use a curve fit which has a high effective error,
> *USE MORE DATA POINTS.*
> There is a great powerpoint presentation:
> https://ece.uwaterloo.ca/~dwharder/Maplesque/Bezier/BestFittingBezierCurves.ppt 
> <https://ece.uwaterloo.ca/%7Edwharder/Maplesque/Bezier/BestFittingBezierCurves.ppt>
> in slide 14, it shows a great fit with very low error.
>
> I suspect just one more point may be enough to get good results.
>
>
>
> On Mon, Apr 16, 2018 at 1:42 PM Sam <cinelerra at posteo.de 
> <mailto:cinelerra at posteo.de>> wrote:
>
>     I'm about to recreate the font effects from an Adobe Premiere YouTube
>     demo for my tutorial.  Unfortunately, I have problems adjusting
>     the zoom
>     so that I can make a zoom curve looks like an exponential curve.
>     My zoom
>     curve always has a downward curvature. I have attached some photos. I
>     remember that it was possible earlier, but something has changed
>     in the
>     behavior of the tagent and disjoint. Shortening the line would be
>     important to get a more natural zoom. Especially when I fly
>     towards an
>     object, the zoom behavior is more like an exponential curve.
>     According
>     to the current state (see screenshot zoom curve) I zoom out and
>     then a
>     little bit later I zoom in. I don't want to zoom out, but that
>     makes the
>     zoom curve current. At the moment this point is not very important
>     because I was able to recreate the desired effect with Natron
>     Compositing, but it would be nice if we could find a long-term
>     solution.
>
>     https://youtu.be/ma9P4nTVtDs?t=19m19s
>
>     -- 
>     Sam
>
>     _______________________________________________
>     Cinelerra mailing list
>     Cinelerra at lists.cinelerra-cv.org
>     <mailto:Cinelerra at lists.cinelerra-cv.org>
>     http://lists.cinelerra-cv.org/cgi-bin/mailman/listinfo/cinelerra
>
>
>
> _______________________________________________
> Cinelerra mailing list
> Cinelerra at lists.cinelerra-cv.org
> http://lists.cinelerra-cv.org/cgi-bin/mailman/listinfo/cinelerra

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cinelerra-cv.org/pipermail/cinelerra/attachments/20180427/41f8f454/attachment.html>


More information about the Cinelerra mailing list