Title: Crocus image Post by: chaospro on March 03, 2010, 08:19:28 PM Hello,
hope you like it...found it when experimenting with the Mandelbulb formula. Reminds me of a flower. (http://www.fractalforums.com/gallery/1/1309_03_03_10_8_11_36.jpeg) Kind regards, Martin Title: Re: Crocus image Post by: David Makin on March 03, 2010, 08:47:12 PM Hello, hope you like it...found it when experimenting with the Mandelbulb formula. Reminds me of a flower. Kind regards, Martin Hi Martin and welcome to fractalforums ! Nice image - did you do something different to the Mandelbulb algorithm or is that just what you get from one of the non-integer powers ? Also is there a problem with the shadowcasting - in particular the shadow to the left of the "flower" doesn't look correct ? Title: Re: Crocus image Post by: bib on March 03, 2010, 09:31:15 PM Nice colors, looks like an artichoke with a radioactive heart :) Welcome !
Title: Re: Crocus image Post by: chaospro on March 04, 2010, 06:43:06 AM Nice image - did you do something different to the Mandelbulb algorithm or is that just what you get from one of the non-integer powers ? Also is there a problem with the shadowcasting - in particular the shadow to the left of the "flower" doesn't look correct ? It's a variation with the multiplication (which mathematically makes no sense)... Regarding the shadow: In order to correctly cast shadows ChaosPro would have to render the object twice: The first time as the camera actually sees it, the second time as the light sees it. The second pass is omitted, and ChaosPro uses only the information from the first pass. Thus it does not see hidden parts which cast shadows... Kind regards, Martin Title: Re: Crocus image Post by: David Makin on March 04, 2010, 12:43:16 PM Regarding the shadow: In order to correctly cast shadows ChaosPro would have to render the object twice: The first time as the camera actually sees it, the second time as the light sees it. The second pass is omitted, and ChaosPro uses only the information from the first pass. Thus it does not see hidden parts which cast shadows... Kind regards, Martin That makes sense now :) For my shadowcasting I do multiple traces per pixel but only one overall pass. For ever solid pixel found I then backtrace from there to the light source/s to test for shadows. Title: Re: Crocus image Post by: Nahee_Enterprises on March 04, 2010, 08:48:23 PM hope you like it... found it when experimenting with the Mandelbulb formula. Reminds me of a flower. Greetings, and Welcome to this particular Forum !!! :) A very nice image to share for your first posting. And I was wondering as well about your shadows, since it appears to have a lot more light coming through a solid object than it should. Title: Re: Crocus image Post by: chaospro on March 06, 2010, 10:03:09 AM Regarding the shadow: In order to correctly cast shadows ChaosPro would have to render the object twice: The first time as the camera actually sees it, the second time as the light sees it. The second pass is omitted, and ChaosPro uses only the information from the first pass. Thus it does not see hidden parts which cast shadows... That makes sense now :) For my shadowcasting I do multiple traces per pixel but only one overall pass. For ever solid pixel found I then backtrace from there to the light source/s to test for shadows. ...and of course doing that backtrace will increase rendering times by the number of light sources, and even worse, I don't know a good resolution to use for the backtrace: If the scan resolution of the back trace to the light source is too high, then the rendering time takes even longer. If too low, the shadow isn't perfect neither. Additionally using that algorithm it is not possible to render shadows from an inner view of an object, because there everything is in shadow... Such problems prevented me from implementing correct shadows. Too many difficulities, and the result won't be perfect. Would be great to render correct shadows, though. And when I compare two images, rendered with and without "my" shadow implementation I can see that the main point actually is that some kind of shadow (correct or not) simply adds a great additional 3D effect. Kind regards, Martin Title: Re: Crocus image Post by: David Makin on March 06, 2010, 02:42:11 PM ...and of course doing that backtrace will increase rendering times by the number of light sources, and even worse, I don't know a good resolution to use for the backtrace: If the scan resolution of the back trace to the light source is too high, then the rendering time takes even longer. If too low, the shadow isn't perfect neither. Additionally using that algorithm it is not possible to render shadows from an inner view of an object, because there everything is in shadow... Such problems prevented me from implementing correct shadows. Too many difficulities, and the result won't be perfect. Would be great to render correct shadows, though. And when I compare two images, rendered with and without "my" shadow implementation I can see that the main point actually is that some kind of shadow (correct or not) simply adds a great additional 3D effect. Kind regards, Martin Rendering views from the inside is going to be either very complicated and/or very slow whichever way it's done :) Using distance estimation the problem with how far to step on the backtraces to light source/s is covered so it's not an issue, though often the backtraces take considerably longer than the initial trace because the backtraces are often passing much closer to the surface for a lot of the trace hence giving very low DE values. For instance this: http://www.fractalforums.com/3d-fractal-generation/tglad%27s-mandelbox-and-using-the-delta-de-methods-for-rifs/msg13609/#msg13609 (http://www.fractalforums.com/3d-fractal-generation/tglad%27s-mandelbox-and-using-the-delta-de-methods-for-rifs/msg13609/#msg13609) Takes around 1min 3secs with shadow casting, but without it takes just 30secs so in fact although the backtraces are done just once per solid pixel they take as long as the plain render (at least in my implimentation). |