Title: Alternative rendering Algorithm? Post by: tomot on March 23, 2011, 10:02:58 PM I'm not a programmer, so I make the following observations based on previous rendering methods, used by other rendering programs.
The line by line rendering method is very very time consuming, since it reevaluates each line individually, without any consideration as to similar pixel color combination's that the scene may contain. Some programs use a smaller window, or windows (some dedicate one window for each CPU) to help evaluate similar pixel colors. Having arrived at the final color they then concentrate only on other pixel colors in that scene that have not been determined yet, to produce the final image. I'm making these observations based on a question posed on the subject of pixel output. http://www.fractalforums.com/fractal-related-links/how-to-render-large-images/msg30066/#new the final 5000 x 4000 pixel image took 1HR 20min to produce on a Quad core 2.4 CPU machine, with 4 GB Ram, running Win 7 Is there a possibliity to engage an alternative rendering method, of the type I described, in the future? curious! Title: Re: Alternative rendering Algorithm? Post by: David Makin on March 24, 2011, 01:15:02 AM Short answer is no, not as far as fractals are concerned and certainly not for 2D projections of 3D+ fractals. Any attempt to do this will result in loss of detail in almost any type of render but in the case of fractals the loss of detail is completely disproportionate to the advantage of using such methods - they have a use for quick previews (e.g. the periodicity checking and guessing options in Ultra Fractal) but should not be used for final renders, especially those intended for hardcopy purposes. |