All Activity

Looking good, but I want to try one more approach:

More ray samples lowers the framerate but it looks quite good. I need to try isolating the SSR result and running that through a denoise filter.

Adding randomness to the reflection vector is pretty effective at handling rough surfaces, but it introduces a lot of graininess:

I've got the lighting working consistently between SSR and environment probes now. An SSR reflection gives identical light levels to the environment probe. Next step is to blur the SSR so it matches the roughness of the environment probe sampling. You can see a few artifacts here because the SSR is perfectly sharp but the env map reflection is rough.

That should be fine without any special code. Just don't do everything every frame. Each AI doesn't need to constantly re-evaluate everything they are doing, instead expensive checks should use a delay so each one only gets triggered once over an interval of frames.

It's not very glamorous but this is important: I replaced my screen unproject functions with matrix-based functions that can handle any arbitrary projection matrices. This is important because some VR headsets use funny projections for rendering. This is the reason Leadwerks does not support post-processing effects in VR, but now it should be no problem. https://community.khronos.org/t/position-reconstruction-from-gl-fragcoord-and-depth-value-using-matrix-multiply/109115

One loop is always better but 40 is still nothing so don't bother in my opinion. Do what's more maintainable for you.

I don't know for sure - but the loop seems faster to me than heaps of scripts. LUA is slower than C++ so I would think the less Leadwerks has to go back and forth between calliing a script the better. Somone else may have a more definate answer.

I'll try to paint the scenario. Say I have a rather large city I'm constructing. I'm using Navmesh very lightly to update goto positions periodically. I have about 10 cops that patrol the city on foot. and about 30 citizens that wonder around. Here's my question regarding optimizing: Would it be best to run the Ai script on each object independently ( just attach a script to the cop/citizen ) or construct a main logic function that handles the cops/citizens within a single control loop? What is the price of running 40 character scripts VERSUS one main loop that controls them all ? I

There goes all hope for that house of mirrors horror game I queried about in January

Here are the results I am getting mixing screen space reflections with environment probes. The two blend very nicely. Of course it will look bad if you make a level out of mirrors, but if you use normals maps to break up the reflection and only make some parts shiny I think this will give good results: Also note how the sky reflection pans in reflections, instead of being basked into the probe cubemap.

Hmm a leadwerks engine competition sounds cool too. Lemme get finished with the syntaxbomb competition first though !

Id even be ready to spend 50 bucks for the winner if such contest would happen. Im still a coding noob because i was jumpy af tried this engine, this language and so on but now since i tried em all im stuck with LW and therefore i think its more than worth to push the engine and also the community. So yeah, im ready to cash in to make such contest happen anytime. Cheers.

Oh man, I miss the Leadwerks Games Tournament. Maybe after Ultra releases, we can have a new friendly game contest. Good luck Alienhead

Greetings Fellow Creatives, If you find my free music helpful, please consider making a small donation on my website to help support my efforts. I don't allow advertising on the site because I feel it would be distracting to my visitors so I pay for everything myself. Support from the community is literally what makes this possible. Thank you in advance! 🙂 That said, this week's new free music tracks are: On my Funny 8 page: "MONSTER MERRY-GO-ROUND " https://soundimage.org/funny-8/ And on my Sci-Fi 12 page: "CYBER STREET MARKET" https://soundimage.org/sci-fi-12/