It is time that we write down an update. We have been working a lot, and some bad luck was on our way, (involving a bicycle, bus and a scull) but we are still alive and kicking.
Currently we are in BoostVC in CA,USA, working on SpaceSys still, trying to get funded by some US VCs, we cant push it alone anymore, it is too big for the few of us, and other ppl have started working on similar things, with way more resources and copying our stuff, literally, even our logo!!!???
Well..
What to say but we are sorry, guys that turn 50 and copy our ideas are not to be afraid of, but it pisses us out.
On the development side, we have done a lot, almost finished with the what we would call a beta, but without windows 10 support i dont know if we will dare call it beta.
Working on input back towards applications inside the engine.. It is a bit tricky and difficult, sending input to app that is in the background, without loosing focus to spacesys is hard, not to use bad words here
but we sorted it out, the mechanics on the windows side, we can send the input, keep the headtracking free, and keyboard is allready working, while working on mouse, we found out that the translation of 2D mouse coordinates dont play along with the 3rd dimension we are opening.. So we are taking another approach, which works better but it will take another 10-15 days to sort out, so in april we will have the build ready for push out.
Also steamworks integration is almost done, so we will be able to launch early access but we still have to decide if that is any good without windows 10 and input working flawlessly.. We will see about that.
That is about it, you can check the youtube pages for new videos, there are some, but as usual, we are not experts of promotion, and the vids are just plain and simple..
When we are ready to push out the input, you will be notified, as that will be kind of a beta
Keep your fingers crossed.
We made huge progress and thanks to Greg and the whole team here, now we have a very bright future.
There is no time to rest, currently we are working on application rendering optimizations and their functionality / mechanics, sorting out few missing shell features, cleaning out bugs and we have a clean road to beta till new year.
We are going to add input for apps in 3D, add windows 10 support and optimize, add some new graphics and more controls for 3D interactions and camera.
I cant tell 100% if we will manage to do it all till new year, but that is our beta goal.
We are also interested in new oculus SDK, so if Esenthel finds the time for that, we will have the latest missing features like direct to rift and HMD releasing which bugs our users now. Go Greg!
What we would love is to get to Steam with more content, i am wondering if any of you would like to do some environments for us, we would gladly compensate for your work if you are interested. If so, please PM me.
Here is a short video of how spacesys works right now.
Very nice! Your models have a wonderful realistic feel to them. Add some wind motion to the vegetation and people will be trying to book vacations there!
If you wouldn't mind sharing, a lot of people on this forum(myself included) could benefit from knowing what and how you planned and optimized.
There are a lots of tricks to use, i will add some tips how to trick instancing for example.. although all is sead allready on the forum many times.. the waterfall is still a placeholder ( its just there to fill the gap) as we are transfering to 2.0 editor it will be fixed because we finally can do what we want out of it..
(02-01-2013 02:46 PM)Dalmat Wrote: About the performance, there is about 10 million polys on the scene and the performance is great, even without hardware instancing or occlusion culling, (that lots of people here cry for). the only thing that we did is smart planning and OPTIMIZING. Optimize, that is the best solution.
Our hardware that pushes it in full motion together with the game mechanics that change the number of polys on the screen by 30% more on occasion is:
CORE I7 3930, 16GB RAM, GTX 680, WIN 7 X64
and we get 60fps on this.
(since the last update January b20 50fps?!)
Regards to all developers and random beholders!
Not to detriment your showcase(it looks awesome), but the example really doesn't test anything which needs instancing nor occlusion culling to perform well.
The reason I got a gripe about this is because you wrote "(that lots of people here cry for)"
60 FPS for such a great PC setup isn't an impressive feat, I hope this is because of VSync, Engines and computers has for years been able to draw millions of poly's without instancing, the thing is if you have 100 objects with 100k poly, these will draw alot faster than 1000 objects with 10k poly(not only because of update). and even faster than 10000k objects with 1k the whole point of instancing is to save calls and time spent sending data thousands of time over when you could split that to a 100th factor to the GPU, you can't trick "instancing". The whole point of instancing is for the exact purpose of rendering replicated objects for alot cheaper than sending them one by one, and having 1 mesh be 10 trees in 1 part doesn't even compare. or 100 grass leaves in 1 mesh.
besides when it comes to ease of map designs, instancing makes it easier than to edit combination of trees rather than editing multiple trees inside one mesh manually.
You can't compare instancing to traditional drawing or mesh merging.
Occlusion Culling in that scene would actually most probably make it perform a bit less, as the whole part of Occlusion culling is to remove draw calls from objects that are behind other objects. in this scene you look at them all the entire time meaning occlusion culling wouldn't be in effect at all. ; )
(This post was last modified: 02-01-2013 09:50 PM by Zervox.)
However, I must agree with Zervox about the instnacing and Occlusion Culling. OC requires a test which is an overhead. If you are testing every object and nothing is being culled then you are losing speed which would happen a lot in that level it seems (at least in the fly through).
I must admit though I think the lack of instancing and OC is the only thing standing in EE's way to being the perfect engine (for me at least)...
Anyway good job at making something look really nice. The Level design really intrigues me into thinking what is the game play like but I guess I'll just have to wait and see.
First of all thanks everybody, it feels good to finally show something and people like it.
What i was referring to is what I read on some posts that complain about the engines lack of OC and performance issues because of instancing.. "cry" was not the best word
So, on our scene shown here while its bare like this, you can see around 1500 objects, of which many of them are the same or variation of the same, but that is not the problem as is when the game mechanics kicks in there are several thousands of items shown, and that can be a bit of a problem, but we are handling it.
As for oc, in this fly trough, as Zervox pointed out we mainly see most of the scene so OC wouldn't help that much, but in game, as things start to happen, and you don't fly above the world all the time, OC would be a nice addition as we do get drops below 60fps, and that is clearly due to lack of OC..
Tricking instancing was also referred to what Zervox said - combining. Like i said before there is nothing new to tell about it, just to plan vise and prepare graphics before you import it so you combine as many objects together as possible. It is not an easy job but..
As for poly count, i made something similar with cry engine, and the results on Esenthel amazed me, results are almost the same, although it is not comparable.. I am aware of the today's poly count in game engines but we are talking engines and games developed by hundreds of people..
(This post was last modified: 02-02-2013 04:13 AM by Dalmat.)
