Normally, the pixels were square and without blurry edges when I am rendering the image without zoom
But if I zoom it in about 200,000 times, blurry edges and “rectangular” pixels were shown:
OSL code from:https://www.youtube.com/watch?v=H8nHxRO7eX0&t=6s
I have made some small change to the code and I’m sure won’t change the algorithm
Rendered in CPU and 3840*2400 resolution
I’ve tried pixel filtering, but it won’t have any effect on it
There’s the image I’ve just rendered and as you can see the pixels are square and the edge are sharp without blurry
I’d bet it’s a precision issue. Once you zoom in that far, you are dealing with very small numbers and there isn’t enough floating point precision to render it out cleanly.
Even dedicated hardcoded and heavily optimized software designed to do this stuff will eventually run into precision issues. It just happens way way WAAAAY deeper 
hmm… that’s my best guess. is there any solution to limit the domain of definition or something? so blender can only calculate the area where I need
Not that I know of. Why the need to do this in Blender and not render out images in more suitable software?
