Liquid Mipmapping Accuracy Update (Patreon)
Content
Shoreline depth is more consistent across mipmap levels than before, but still needs fixing. What worries me is that the formula I'm using to scale the depth across mipmaps makes absolutely no sense, it is just some guesswork that I pulled out of thin air after my previous attempts failed. Still, it's far better than before.
The position and shape of the shorelines are perfectly consistent on standard turbulent mapping across all mipmap levels (see the second half of the video). To ensure this, I implemented a more accurate approach to generate sine tables, inspired by this old article from 2012: Doing Gravity Right, by Hannu Kankaanpää (The Internet Archive is a treasure trove of such arcane knowledge that can't be easily found anywhere else anymore, btw).
But with melt mapping enabled (r_water_fall, r_slime_fall, etc), the position and shape of the liquids gets inconsistent across mipmap levels (see the first half of the video). At least this narrows down the remaining big inaccuracies to a very specific path in the code.
And lastly, as mentioned earlier this month, I'm going to raise the price of the unpublished tiers to 5 dollars. I'm really sorry for this, but the only way for me to be able to make progress on the more complicated parts of the engine is to miss some shifts at my job, so I can have more uninterrupted time coding at home. I already missed a week of work this month, which is why I managed to make some good progress on Retroquad, but I can't miss more than that or I'll lose my job.
And I can't simply stop coding, because developing Retroquad is the only thing that makes me feel truly alive. It's the only thing actually helping me survive my depression for some more time.