Implementing antialiasing on the anycubic photon in software
Having thought for a whole 10 minutes about this problem in the shower I'm going to brain dump and hope that someone else can actually test it out.
There are only two processes I can think of at the curing point: beer-lambert style exponential drop off with distance and some kind of surface level agglomerative process. The first of these means that antialiasing doesn't really exist - all you get is a less cured exponentially decreasing layer near the LCD which will likely snap off. So if antialiasing exists, it must be at least partially agglomerative surface chemistry.
Let's assume this is true. I believe the photon file format allows for specification of a cure time for each layer. Say we have a cure time of 1s per layer. We can turn antialiasing into what was called a planar image format back in the day - for each z-movement we put lg(levels) layers in with exponentially decreasing cure times: the first pseudolayer would be 0.5s, the second 0.25s, 0.125... As each layer is mostly white, the rle compression should work quite well.
I think this would be quite easy to add to
https://github.com/Photonsters/Slicer
if someone can work out a way to generate the antialiased pixel values.
Interesting follow-up, if this process works, it works at all scales (with varying rate between the two processes - beer lambert must take over eventually) so we could invert this function in the style of interference lithography and potentially significantly increase the print speed.
Ping me if you want to give this a go and I didn't make it clear enough.
翻譯年糕
