Today, I finally fixe my last (I hope) issue to generate proper MSSAO. Now I need to clean and optimize (the last fix was to use RGBA32F instead of RGBA16F to store normal + depth with enough precision and avoid crappy artifact.
very nice David
Hey I always love seeing your code and WIP stuff, but I honestly habe no idea what’s going on here. Maybe an explanation for a layman like me?
There are several variation of SSAO (I listed 10+ and and a survey, if you’re insterested). The one included in jME is too blur and blocky for my need. Last year, I tried to implement a the paper of Mc Guire, but I failed to produce the expected result. Multi-Resolution Screen-Space Ambient Occlusion (MSSAO) was the other one that I noted, and Nehon +1 my choice, so I try to implement it (into my deferred pipeline). And yesterday I finally have the first results. In fact, the implementation is incomplete and I guess some optimization are possible.
My code is available at https://github.com/davidB/jme3_ext_deferred/tree/mssao
Your pipeline is really becoming something great
Let me post my humble project.
I’ve been working on my spare time on an octree editor. It’s still very early stages, basically just roughed out the algorithms to see if it’s actually usable to create worlds.
It’s my first Jme project.
The worst headache was making the textures seamless among voxels of different size in different positions. Unfortunately I have to stop this project for a while to focus on the things that actually pay the bills.
Here’s a youtube video:
And the source files here (they are still quite messy ;P)
That looks amazing! Just turn down the noise filter a bit maybe.
wow thanks that helps a ton!!! now i can actually appreciate what youre doing
After seeing what’s new this month my project immediately started to look super shitty lol however, this morning I played my first offline match with it and everything worked just as expected, so I decided to celebrate the event posting a screenshot anyways
Looks interesting… what is it?
Yah, it’s a selection of mods I believe together are incredibly powerful.
- Remain in Motion for moving blocks
- Project Red for wiring, micro blocks and switchable lights
- OpenComputers for well, computers (and lua)
- BuildCraft for generating the power needed by the computers (optional)
My unsubstantiated opinion is that as long as people don’t try to beat Minecraft on it’s own turf, the market is looking for lots of what Minecraft does badly, such as the survival part taking less than an hour to be done with and the AI is just awful (not that I’d do better) and so the player spends most of it’s time inventing needs instead of reacting to needs. Also, not story driven, and then there’s a thirst I believe for space stuff and so on. Also, there is still a great weakness in Minecraft linked to the copy pasting stuff between maps (or inside the same map) being impossible if you aren’t the kind that uses only basic (uninteresting) blocks. This just stinks.
For me, the biggest reason Minecraft still dominates lies in it’s mods… Captain Obvious to the rescue!
Eloraam is currently trying to beat Minecraft on it’s own turf so maybe even that is doable.
Anyway, I’ll gladly check up your game when you finish it ^^.
Is she still around? I thought she was dominated by real world issues and abandoned hobby of coding things for fun.
A 1v1 board game, inspired by an installment of the Might & Magic series. I started with aiming to a 1:1 emulation, time constraints made me change my plans
Maybe someone could lure her to JME side? :chimpanzee_closedlaugh:
@pspeed why no ssd?
Somewhat irrational but latent distrust of “limited number of writes”. Still, I considered it. The RAID0 option was going to be a cheaper + more space version that would be pretty fast. It’s unfortunate that I didn’t get it working… I might as well have gotten an SSD + SATA drive anyway.
Having RAID0 you have 2x greater chance of data loss. You need only 1 damaged HDD to lose everything. I recommend RAID1, it saved me two times in last 3 years.