Challenge: 4,5 days for a 400x300 picture !

[Antoni Gual]

I feel guilty about your insomnia. Sorry...

[Jark]

I also have caught one more cold because of the weather changes: so it was not entirely your fault :lol: :wink:

[Jark]

a "holes oriented" sponge renderer: I now have a routine that returns the two intersections of the ray with the cube, and I can select at will the side to be plotted : external view or internal view of the cube. So I can start "plotting holes" with that...

[Jark]

I can now plot the holes instead of the cube: all I have to do now is to merge the main cube and the holes as "negative shapes"

[Jark]

I worked on that during lunch. Well, that's a major change in the raytracing logics. Instead of just searching, ray by ray, what it the closest hit on an object, plotting emptiness requires, to manage, for each ray, a series of intervals, each interval having 2 limits (an inbound limit and an outbound limit).

During the ray scan, each plai nobject and each hollow object will update the series of intervals. At the end, you keep either the closest inbound limit, either the absence of limit (no intersection).

Well, I will do that, but it's a heavy change...

[Antoni Gual]

Pov-Ray people are starting to seek for a new job...

[Jark]

But they will have to learn a new language : "Basic" :rotfl: :bounce:

[Jark]

I managed to render the first two ranks of the Sponge with the holes technique. It takes only a few minutes now... The third rank picture is currently rendering, it should be finished in less than one hour, against 5 hours with the cubes technique. (I'm doing tests without the Perlin colours).

The problem is that I dont understand why the picture is correct! Probably because the size of the hols is always smaller than the cube itself. In the general case, my algorithm should not work.

Then I will have to test that with the /ah option to run the 4th rank... Cause I need to handle Single type arrays of (1+20+400+8000)*3 records per array...

[Jark]

I think the basis of the prog is the wrong one !

Generating all the cubes first, storing them then plotting them requires a lot of memory.

I just realised each cube can be generated dynamically during the space probing: it may consume time, but there is almost no limit in terms of rank and therefore number of cubes...

[Antoni Gual]

But in this way you must modify the logic of the program...
An easier way would be to store the parameters in a RANDOM file.
Simply
-Save all parameters for a shape in a TYPE variable SHAPE.
-Create the file with the record length equal to the size of SHAPE
-To save the variable do PUT #1,ShapeNum,SHAPE
-To get back the variable do GET #1,ShapeNum,SHAPE
and that's all.




Do it as you want, but please don't give up with the holes, they are very interesting, not only for the Sponge.