The hottest one is coming to town

[PsichiX]

He found easy trick to outcompete other 2D engines - Phaser users hate him!
[WATCH MEMES]

[b10b]

I'm always interested to see new game engines, but this post (and the current github links) is too cryptic for me ...

https://github.com/PsichiX/Oxygen

What problem(s) does Oxygen solve, and why is it better than the alternatives?  What are your goals for the engine and how would you like others to contribute?

[PsichiX]

http://creative.psichix.io/lights/
(watch on desktop, it uses typical desktop features and not all mobile gpus may support them)

This is dynamic lighting made with deferred rendering pipeline - everything you see here i composed within one small JSON scene file, and almost none code. it's just one thing of many, that other 2d engines lack - they are years behind

I will gradually unravel next cool Oxygen features in next days, stay tuned 

[b10b]

It doesn't work on mobile, and the demo effect (requiring desktop) looks similar to what can be achieved cheaply with vanilla canvas (on any device).  Is "deferred rendering using parsed JSON" the primary objective of Oxygen?  If so, what problem does that solve for 2D games in particular?

[PsichiX]

Quote

looks similar to what can be achieved cheaply with vanilla canvas (on any device)

okey, so can you provide such example of vanilla canvas with dynamic lighting with normalmaps, also maybe second example with phaser? if not, then you have answered your question - Oxygen gives you different pipelines to achieve different graphics problems easly solved withing few components, what cannot be currently done with vanilla canvas and glorified phaser and this is just beginning! When i was designing features, i was looking for what other engines cannot do or can do but with too much effort put into making them. Oxygen architecture is designed to easy solve those problems, and lights are the common one among 2D engines. Engine currently gives you 4 rendering pipelines: forward (you can call it default one seen in other engines), simple deferred (with very simple lighting features, no normal maps, but you can easly light up some part of your scene), default deferred (normalmaps included to give your scene some visual details) and PBR deferred (your material can have not only normalmaps, but also metalness and smoothness and emissive features).
Your ideas can be easly achieved within very minimal (or even none) code, it gives you better graphics opportunities than other 2D engines and graphics is not the only one. As i said, more cool features i'll unravel later within next days, and now i'm waiting for your examples to prove, that you really can do what you're saying - challenge!

[rich]

None of the Playground examples work for me (lots of WebGL errors in the console), which is a shame as I like the 'json' approach - although to say it eliminates coding is wrong, it just moves the knowledge required to create a scene away from your JavaScript API and moves it instead to knowledge required to learn the data structure of your JSON.

The lights demo worked. Very pretty. But at 624 gl ops and 40 draw calls I'd bloody hope so!

[b10b]

@PsichiX I could spend an hour replicating some colorful circles moving on a brick wall using additive for the lights and an alpha for the relief based on angle of lights (all using canvas).  But I'd rather spend my time more wisely because nobody else cares and I think you might be missing my well-intentioned point?  Your opening post failed to communicate anything of substance.  Your subsequent demo failed to demonstrate the use case of why such lighting may be significant.  Features are only "cool" if potential adopters can identify with the problems they solve.

[PsichiX]

Thanks guys for your kind words! I've got information i wanted to get, now i'm going back to fixing stuff you both mentioned and see you soon! ^^

[AndreasLoew]

I'm impressed!

 

 

 

 

 

 

[PsichiX]

Tutorials are updated to latest engine version (last days i've changed engine default shaders and moved them from plauground to module and tutorials was not following that, and that took me longer besause i had to introduce web fetch engine from remote hosts, for engine assets this is unpkg service - now they works fine)

still have to provide overlay on labs preview to tell user that assets are loading (like on playground)

[vamos]

I think you have a good idea there, but maybe you'll need to work a bit more on its presentation.

Also your claim that these things cannot be done with other 2d engines is just not true. Just the other day I was reading this blog by the Wade guys that explained how to make this exact same normal mapping thing with their engine: 

A dynamic lighting shader for your 2D sprites

I copied their implementation and had it up and running in my own framework in about half an hour. So it's cool but not unique.

Like b10b said above, if you want people to use your engine you may want to explain a bit better why they would want to switch from their current pipeline to yours.

The concept is good, it sounds like you have some good ideas so I'd like to check it out when you have some nice tech demos - but those need to show something interesting and unique like the blog I linked above or kevs3d etc

[PsichiX]

@vamos on weekend i'll drop next demo, with proper game-like settings and explanation!