09-05-2024, 07:17 PM
(This post was last modified: 09-05-2024, 07:18 PM by TerryRitchie.)
Ok, I finally figured out what the math is doing. However, in doing so I realized this approach only works in one certain circumstance ... the image to be mapped must be exactly twice as wide as it is high. While it's possible to supply a radius for the output sphere the overall math still uses half the height of the image as the radius when calculating.
For example, the worldmap image is 1235x617 and fits the one circumstance. Even though I can supply a radius of 100 the math still uses 308 (half the image height) when calculating points on the image to use and then simply scales 308 to 100 when outputting the pixels. This is why no matter which radius value you supply the drawing always takes the same amount of time.
It's an ingenious approach that needs some tweaking that I'm working on to speed things up.
The need for an image to have a width twice the height can't be overcome using this method however. As the code in the youtube video shows this is "fake sphere mapping" but very convincing nonetheless.
I should have some code to post shortly.
For example, the worldmap image is 1235x617 and fits the one circumstance. Even though I can supply a radius of 100 the math still uses 308 (half the image height) when calculating points on the image to use and then simply scales 308 to 100 when outputting the pixels. This is why no matter which radius value you supply the drawing always takes the same amount of time.
It's an ingenious approach that needs some tweaking that I'm working on to speed things up.
The need for an image to have a width twice the height can't be overcome using this method however. As the code in the youtube video shows this is "fake sphere mapping" but very convincing nonetheless.
I should have some code to post shortly.