A BIG thank you to Ashley!

i just tried out the new experimental tilemap seam fix and i can say with tears in my eyes that it works!!

After countless hours of trying to eliminate them , Ashley came guns blazing and killed every last one of them!

Now all it is left to do is to match the original batch rendering and we are set for life

If i find something out of the ordinary along the way i will post it.

But for now i will enjoy my seamless game!

Same here ... To be honest I'm more happy for the tilemap fix than the real-time shadow casting, but that is also a great addition.

every new feature is welcome but for me the tilemap seams thing is the most important since you just cant have a proper game without it working properly.

i truly believe that it is a core feature of C2 ,just like the editor and the visual programming , and it must be solved once and for all so everything will work in harmony once exported.

Actually, the real-time shadow casting is pretty awesome too, just look at this topic: The new Lighting & Shadowing System

I'm glad Ashley has listened our petitions.

Ashley is probably one of the few people in software development that actually listens to its customers.

But i am sure that up until now he just couldn't figure out a solution.

He is after all a programmer not Harry Potter :p

Anyway i am happy since this was bugging me since day one and if i ever see some performance gains as well i will take him out for a few drinks!

I'm glad it's working well, but remember as the release notes mentioned it may not always work in canvas2d mode. I also don't think we can get it working with the previous optimisation: when a whole rectangle of tiles is drawn in one go, it must be set to a repeating texture, but then floating point scaling and linear sampling means the edges can wrap around and sample pixels from the opposite side of the texture, causing seams. Seamless mode renders every tile one-by-one without wrapping so pixels at the very edges never wrap around to the opposite side of the texture.

Still, in the WebGL renderer the code to render a quad (e.g. a single sprite or single tile) is extremely well optimised and rendering tile-by-tile should still be pretty darn fast, even on mobiles. Modern phones can get thousands of quads on-screen and still get good framerates, so even a few hundred tiles should be OK.

I just tested the game I am working on and the CPU usage actually went down somehow... maybe because of some other optimization not related to tile maps?

Oh and the seams are gone which solved the last technical issue I was working on before starting the process of building out all the levels in the game. This update came just in time.