Home Artists Posts Import Register
Join the new SimpleX Chat Group!

Content

Some progress on my perspective correction studies. This screenshot shows the results of different algorithms.

The pair of columns on the left shows the results of the subdivided perspective correction algorithm used in Quake and (afaik) all other software renderers in the world. It has an extremely small deviation, which means it is extremely accurate.

The pair of columns on the middle shows one of the perspective correction algorithms I've created. It has a deviation of 0.69%, which seems small but can result in visible texture misalignments at the edges of surfaces. I had some high hopes for this algorithm, but it's not good enough.

The pair of columns on the right shows the results of the latest perspective correction algorithm I've created. It has a deviation of -0.033%, which is over 20 times more accurate than my previous algorithm. It's still nowhere nearly as accurate as the standard algorithm, but it seems to be accurate enough to not cause any misalignments at the edges.

The plus side: my latest algorithm should be massively faster to compute. I'm still going to work out the implementation, but it feels like the end result may be revolutionary.

Files

Comments

No comments found for this post.