- Screen Space Ambient Occlusion
- Supports Approximate Normals (50% faster)
I really wouldn’t put most of the post processing filters, apart from the basic ones (color overlay, tone map) into the feature list. Majority of the others look straight outta 2005 and are not something I would brag with on the info page.
I’m not sure about this myself, as I very much think jme-vr needs a rewrite, but might be worth mentioning somewhere that (at least experimental) VR support exists. I don’t know what the status of our iOS builds is, but might go into the same category.
Perhaps we have an experimental section on the features list then?
I can remove some, but things like water, SSAO and bloom should stay I think.
There are other and better filters available on the store such as the MipMap bloom filter.
Detailed frame profiler:
We use the profiler a lot in our engine. It is a very valuable tool that is very little known.
DetailedProfiler should get more use now that issue 794 is fixed.
Bloom is the one I’d explicitly not mention. The default one is just… no. Unless your game is cube based, it comes off as overly blocky and overal pretty ugly for anything but very small amounts of bloom where the blockiness is not noticeable.
MipMapBloom from the store is the one I’ve been using and still think it should make it into the core and replace the default bloom.
Water is a mixed bag. It can look okayish when tweaked correctly, but I see it more as a base for writing more customized fluid shaders. I guess it deserves a mention if looked at in that light.
SSAO also never produced satisfactory results for me, but I never took time to tweak it a lot, so I can’t say whether it should be included or not.
Don’t forget the material definition system
There is also a visual editor.
I actually want to try to transplant other system materials into shadernodes, so that I can inherit and improve the original materials of jme3.
I’d tend to agree - I think it should too, but @pspeed is of the opinion that we should be trying to get things out of core to make sure the engine gets the dedication it needs and not fixing things that aren’t necessarily core and leaving it to core devs to fix. I would agree with that, too. I think this is partly where the store should come in and make it easier to depend on things on the store. I have a few ideas, but I’m not 100% on how to go about that (think auto-jcenter on submission).
I’ll make a start on creating the page on the main site now and publish it live in the next day or two.
Bloom is a weird one. The built in one is faster I guess and looks good for normal settings. People’s tolerance for bloom has gone way up since “back in the day”, I guess… where even a moderate amount of bloom will give you complaints about “too much bloom!!!”
I think some of our post processors’ default settings are a little wacky maybe. And given that the post FX are in the own module, I’m personally ok with adding additional options for normal things and I think having the option of a Mipmap bloom may make sense.
…but keep in mind that the renderer refactoring that Ricc is doing may change the concept of what a post processor is. So it may not yet be time to open the flood gates.
I still wouldn’t mind having a Mipmap bloom in core if it didn’t replace the gaussian one that we already have.
Adding something like this somewhere on the website can be a good idea I guess. Maybe in get started page?
If you feel adventurous enough to get started with your networking game while utilizing the power of Entity Component System and a high-performance real-time object synching library, there is a very good example project provided by our engine leader @ pspeed :
could each be a link to dive deeper?
Like the Open Source could link to Github, the Fantastic Community could link to Discourse and Discord (and perhaps Youtube if we get that channel going again), the Software Store could link to the store
From what I understand, at least in theory, the one in the core is effectively the same as a mip map bloom with the amount of mip maps set to 1. I am not familiar with the implementations themselves, but the concept behind them is the same. So unless I’m missing something, the store filter with mip map levels set to 1 should be just as fast and the stock one, while keeping the option of it looking nicer if performance is not the concern.
Also Nifty GUI should be listed I guess
I do not believe that nifty is a supported platform anymore. Iirc it has been many years since nifty was last updated.
I’m not going to argue against adding nifty if that’s what the people want, but I’m not going to fight for it either. I’m not a big fan of it. It has a lot of quirks and isn’t updated at all.
IMO between Lemur and jfx-11 you’re all set.