The results are really good however performance impact is too high to warrant it's use on anything except nVidia 580s. Might consider using it as a hidden high-end only option. There was unusually high loss of performance when using weapons but I need to confirm that this is relating to the number of cannon shells I put in the air and all the tree collision going on.
Here is a sample of test screens over the firing range.
If you wish to read more about FXAA then head over to his blog here:
http://timothylottes.blogspot.com/2011/04/nvidia-fxaa-ii-for-console.html
More range progress
Still missing 3D grass, not sure if we'll have time for that, my proposed technique for streaming in those details has yet to be proved and there are more pressing features. AD constructed some dirt tracks and small features to represent 'Smalltown USA'.
Just in: near completed firing ranges complete with lanes and distance markers (which are a bit hard to see here).
Trying out the range (standard FSAA and HDR), looking a bit better than what I had in mind circa 2002.
'Sway' shaders normally used for tree and grass motion to simulate wind effects have been used to good effect on ADs flag model. Vertices sway according to the amount of red (through to black which equals no moment). This flag appears to be animated but is done all in a stock shader. How cool is that?
All your flags are belong to us?
Hi Flex.
ReplyDeleteThat looks GREAT!!! Sorry to hear about that impact on framerate, but still....very nice. I love the shadows in cockpit so much. Great feeling. I cant believe how cool the flag looks. It brings real life to the environment IMO. Beautiful. Is there some bug on that picture with Apache from left side during low level flight? There are only wheels visible from the undercarriage. Sorry...I know its all in quick testing conditions. Just checking. :-) Anyway...I already love this sim, almost like Im already playing it for months. :-) Great results from You and Your team.
Hats off to Dave for his continuous efforts in advancing what's possible.
ReplyDeleteI was working on wheel suspension, strut compression and trying some different things, some of which require removal of bits of model. The whole thing is put together like a model kit.
Still not really happy with it. It's undergoing some structural changes. On the ground it taxis well, to a point.
There's 1 slight error in the blog post. You mention that flag isn't animated, it's not quite true. In order to get the initial 'waved' deformation I used a series of 3d studio max modifiers to add wind/gravity and a simulation of cloth.
ReplyDeleteThen it auto animates (or simulates as it calls it) and I collapsed the modifier stack just at the point when the mesh was deformed to the correct extent.
The sway shader takes the already waved/deformed mesh adds a bit of life to that deformed shape.
The things you have to do to make a 2011 compatible flag model. :)
Cheers
Hi. Thanks for explaining that thing about wheels and struts. I really would like to understand You, when You are talking about shaders, modifiers etc. :-) Again...those pics are so niiiice. :-o I have one silly question....what is that "camera-like" circular thing on top of the dashboard? And will the explosions affect the trees somehow? Not really important for me,...just asking.
ReplyDeleteAD cheers. I understand you set the initial conditions through MAX modifiers. The setup is the hard part, the payoff is the smart application of the shader, much in the same way you built the camo netting tents.
ReplyDeleteI'll update the entry today. I know we did the tents a long time ago but somehow this seems more visually engaging, probably because you made it look so darn good.
Anyway the next step is to use the green and blue colour component to specify the flags UV offset so one flag will fit all and you can simply choose the flag by setting a key value.
ReplyDeleteVery nice! Hope you can keep the systemrequirements below or around "average"
ReplyDeleteYes I do hope so. We've experience some serious drops in performance relating to how I think the engine is culling and attempting to optimise vegetation. Before we were delirious about performance, it wasn't unusual to have 3 figure fps rates occupationally on my computer.
ReplyDeleteVeg draw times have become erratic for no obviously discernible reason so profile time real soon, get to the bottom of that one.
If you didn't want tree collision things would be much simpler :) and faster. But I do, going without isn't really an option.
In BS my biggest gripe was tree collision. Being seen through and attacked through trees was enough to stop me playing it in the end. Frustration had set in.
ReplyDeleteSo... glad to hear you think the same way. After all one of the Apaches strengths was the ability to hide behind trees and do pop-up attacks.
The situation isn't so bleak. The range doesn't have any enemy units so although it would be nice, it doesn't specifically need tree collision. Herat doesn't suffer from the vegetation slow down (200fps+ in the editor).
ReplyDeleteOn a separate note. I made a test version of the range and cleared all the trees out of the eastern hills. Dat footprint is down by about 35%. Load times are better (down to just over a minute) but veg slow down is the same.
Cheers
Anyhow, there's a quick fix solution to the veg culling problem that'll need to be tested in game but in editor it's solving the issue.
ReplyDeleteCheers
..."looking a bit better than what I had in mind circa 2002."
ReplyDeleteSooo nice to see isnt it? Jeez the pit is really looking great.
I was wondering flex if its possible for the community to pitch in and make maps or enhance maps to help save you time and let you focus on the other things the sim needs?
Flex isn't involved in terrain development, that's occurs on my side of the virtual office. If we were going to out-source terrain development to the community we'd first need to develop a freely distributable/comprehensive mapping tool and that would chew away on Flex's development time rather a lot.
ReplyDeleteHopefully the community will support us through buying the initial CH:Range taster/teaser (we really need to decide on a name). We can't ask for much more than that.
Cheers
I think we'd wait until Combat-Helo 2 which will have a larger terrain system before pushing out a content tool. Assuming we get to make a follow up of course.
ReplyDeleteahhh OK thanks guys.
ReplyDeleteI just dont know how so few of you can make this work... simply amazing to me.