"I hope to make a new rendering system one day, that can make accurate shadows under objects, unlike this system we use now, which often misses shadows where there should be, and puts shadows where there shouldn't be one. But that is not the subject of this patch."
It's vertex prelight. Shadows are simply baked into vertex color.
This is an old technique, where quality of lighting depends on detail of the mesh. vitaly_m, baked lightmaps was not possible because old tech limitations. I hope it will be fixed with Shader Model 3.0 and pixel shading.
I like some of the changes but some not so much. For one, the flawless tarmac in comparison to the patchy/old concrete type. Thats not really an improvement and rather boring even just to look at. I'd love to see maybe old road parts still existent. Like Blackwood with old and new tarmac sections. This type of detail helps a lot with immersion of the speed.
The new chicane is interesting, but the old one was a lot of fun too. I've seen a lot of people fly off there, so difficulty was never an issue yet super fun if you nailed it
The interesting bit is the new infield layout, we believe there has to be a right hand hairpin in the middle. This is based on image 7, you can see where the track splits, you can see the next left sweeper at the bottom of the hill but right above the white building, you can see another piece of track, then above that you can see the backstretch and grandstand. I believe that image 12 and 21 are from the new track, after this hairpin and in the run back towards the backstretch...this next image is pure speculation.
I think the piece of pavement that can be seen right above the white building is located on my map right where my speculated track is drawn over my #5 marker...if that makes any sense.
As others have mentioned, LFS' shadows are vertex based and that's really a hand me down from DX8. Pre-baked Lightmaps are complicated in the LFS arrangement, although are possible... but they would curtail any hope of a day/night cycle.
Pixel based lighting has some possibilities now that LFS is in DX9 - but there are some issues which Scawen will need to take a view on.
I doubt he will do deferred rendering because I get the impression that if Scawen attempted that from his current position he'll probably have to disappear for a few more years and that won't pay his mortgage.
That leaves us with a couple of options... Firstly stencil shadows, they are a bit old school now but they work quite well if the geometry has no holes in it... As odd as that statement might sound, you'd be surprised how hard it is not to end up with this disaster from my own experiments [pictured below] with this technique - caused in this case by my games swappable mesh pieces on the main character. This technique is slow in high geometry environments and is not normally used on scenery - it would blow any chance of South City running on an entry level system.
One approach is to do it as a post-processing effect, from what I can tell this was the method used in Assassins Creed 3. You render the view from the light and compare the depth of each point in 3d space to your camera - it works really well until you render the world on a 3D screen... Whereupon the shadow overlay simply doesn't work because it is drawn after the scene at a depth of 0. LFS' new 3D mode would be broken by this approach.
By raising the system requirements to shader model 3 (perhaps leaving the old vertex lighting as a fall back) it becomes possible to use some alternative techniques. I have no direct experience with these and don't know which if any would be good with LFS' 3D support, PSSM looks promising at a quick glance.
The thing about shadows in modern games is that you usually design your game levels around the lighting architecture of the game. LFS pre-dates all of the modern techniques so the shadows are being added in afterwards to a 3D world architecture that will not necessarily be optimised for the technique that we end up with - updating the lighting is therefore going to cause some headaches!
My advise to Scawen would be to decide between one of two options: A full structural 3D overhaul with a switch to deferred rendering - which should yield the best results but might have an impact on the current entry level systems, or more likely a keep it simple approach with hacks for given entity types or areas to fake a good shadow system (and you would not believe how many of todays cutting edge games are FAKING their effects - it's probably not the right word to use when talking to end users, but essentially games programming is itself the art of how to fake reality).
For example he could add some worldwide light level map textures, maybe with flexible u/v scaling based upon common view positions/angles with 1 being under full sunlight and 0 being in shadow at all times of day with 0.25 being in shadow in the morning and 0.5 at midday etc, maybe using RGBA channels for sunlight arc movement. Whilst this global texture would render at a fairly low resolution (although should be higher than vertex density) it would allow to fake the sunlight across the level and could be filtered back out when under a headlight by the pixel shader. How well it works would be down to the size of the texture / divided by the scale of the map (or map area covered) multiplied by any efficiencies regained by uv scale warping... This is just a thought based upon a pre-rendered approach to try and keep the low system requirement and 3D support. Scawen knows his architecture better so will have better ideas I'm sure.
Shadows are a very complex issue, but if Scawen is anything like me with programming (and he is an oldschool nerd too) - then I believe he'll really enjoy solving it - I first discovered HLSL at the tail end of last year and it has rapidly become my favourite programming language, and is by far the most fun I've had at the keyboard since I first discovered I could make pixel perfect vertical scrolling sprites on the Spectrum computer by poking the font rendering to use my own area of memory... Oh them were the days...
EDIT: Oh my, I just remembered the memory address I used to poke: 23606 and 23607 :P I'm so happy right now...
I don't think it's likely Scawen would attempt a major overhaul before the physics are finished, I'm actually kind of surprised about the move to dx9 but I think that's because he got excited over 3D. Although it still amazes me that he did it without a 3D monitor or headset of any kind...
Actually no. There's some trick with lightmaps. You can bake only skylight high quality GI with white color and multiple it however you want in shader. Combine it with realtime directional light/shadows and you'll get a nice look. Of course there's no sunlight GI, but making some low-res bounces in realtime is not so costly than whole GI
But thats a static light position, which admittedly leaves 3 input channels with global uv mapping, sweet... Surface water could go on one ;p. but it must be possible to use those channels to pre bake lighting at different times of day - but that would result in shadow artefacts of the shadows... (That is t going to read well is it!).
It's why I was thinking of storing an angle on each channel rather than a light level, this could then be transposed through time and applied against the normal map as the light direction - I admit I thought of it whilst typing and haven't proved that I'm talking rubbish - but in my head it seems to work*
Really, all of these graphical things are cool, but I think the majority would prefer physics and new content first. Once that is all out of the system new graphics would be a huge bonus, but not presently to me.
Having the 3D patch released just a couple of weeks before the Rift DK2 ships to customers is not a bad move at all.
Eric is the one doing most of the work for the Westhill update, and probably Scawen didn't have much to do until the track was almost done, so he could improve the major performance hogs.
Anyway, I'm exited to try the new track, I never liked WE on single seaters (99% of the time pedal to the metal was not appealing to me...), and the new version looks like a lot of fun in all the cars. Plus the new layout (and access roads if I'n not mistaken) are going to be a key part on the calendar of the leagues that are to come in the second half of the year
I think Scawen is doing the right things these last months, despite some people not seeing the whole image. I think he is actually buying back the time LFS have lost, so people will be more pleasent to wait for a late 2014 new content release (not WE, something really new).
I've never got to know Alex (though his name has been already mentioned some times), who seems to be the person who did the graphic engine of LFS. I wouldn't mind having another person developing a new engine (along with multi-threading), but Rockingham and Westhill already looks amazing to me (except the shadows and no realtime ilumination).
well i tried yesterday to run lfs in a stereoscopic monitor(true 3d without glasses and hippy thingies)
(its a planar monitor that uses 2 lcd's one above you and one in front of you)
while at first its going nicely and the motion of fluid like you expect (all of that in a cruise server) when you start to floor it down the background of the track goes into some sort of "Bloom" effect (i dont even know how should i even describe it..)its like the car is going into a stargate and reaching light speed sort of thing
i dont know if it is incopartibility/drivers/not properly configured (which probably is but im not willing to spend so much time to try and find what is going on)
Thanks to the development team, not only for the update but also for the progress report!
We know it takes some courage to post anything related to future / plans / next releases in here...
I am pleased with the overall positive and constructive attitude of my fellow LFSers in this thread - it's been a long time since the silent majority was not so silent, and I'm looking forward to driving the new Westhill