Building My Own Image Engine "HakoNyans": I Beat PNG, but WebP is a Wall

Hi DEV community! 👋

I originally started developing a custom image engine called HakoNyans for a DEV Challenge. The challenge has ended, but my passion hasn't—I'm still actively building and refining it every day.

The Milestone: Beating PNG
I'm happy to report that HakoNyans has officially surpassed PNG! Getting the architecture and logic right to beat such a classic, widely-used format was a huge milestone for this project.

The Current Struggle: The WebP Wall
However, my next target is WebP, and let me tell you... WebP is incredibly strong.

Here is where HakoNyans currently stands against WebP:

Processing Speed: Currently 20% slower.

Compression Ratio: Losing by 20% (resulting in larger file sizes).

I've hit a bit of a wall. Competing with WebP's highly optimized compression logic is no easy task. Closing this 20% gap in both speed and size is my current obsession.

Let's Discuss!
I'm continually pushing the limits of my current algorithms, but I'd love to hear from the community.

Has anyone else here tried building an image codec or data compression engine from scratch?

What optimization techniques or mental models helped you break through performance plateaus?

Any advice, feedback, or shared experiences would be greatly appreciated. I'll keep pushing HakoNyans forward!