Well, clearly this is something that isn't finished.
Work is still continuing. The last couple of days I've been rearranging some code so that the modeller objects can be included in the lighting render. Previously, modeller objects that were in a shadow appeared to be glowing.
Obviously we are well aware that there are big problems with shadows, but this is still an important graphical thing that Eric has waited for, for several years. It followed on from the repeating objects system, which allowed the repeating modeller objects to be included in the lighting with the segment.
It's one of those coding things where one thing leads to another. Some people like to criticise and mock this approach, because they have nothing better to do, but it's how things get done eventually.
As usual, all good but taking longer than expected, which is as expected... things are being crossed of the lists.
That's just a small progress report, for people who like them.
I'm aware there are other people who are just lurking around waiting to attack, get angry that we are still working on things and so on. They can keep their comments to themselves, thanks! If they know so much better than us how to make a racing sim, they are welcome to make one!
This might be not so much related about this Westhill Progress, but I was thinking what was the decision to make new westhill, but unable to release Rockingham or other tracks?
It has been long enough I last time looked these things seriously, so I bet I do not remember "enoughly" about what I thought before.
Sure, the thing is this is nothing compared what will come on future, I am pretty sure about it. But question is, why this way? What actually changed mind? It all started on as oculus rift, then Direct X 9 thing, I bet this new westhill is actually a test to see how things would look and go on upcoming updates.
Or is there a post claimed about the decision?
I was not sure do I make a topic, or post elsewhere, or here, I decided to put in here, because there is no new westhill released on test or any and this is actual progress report, so I decided to post here my question.
( if someone has an answer for my question, please spend a minute and quote it on here )
Anyway, I am impressed, but not enough due the knowledge about the future, or so at least I sense like this so.
EDIT: I am not going to criticize any, but I am thinking this recently. The thing what I read somewhere that Westhill needs an update on high or semi-high priority, also because well, new things should come in some point. But still, I am not convinced fully, that's why I ask, for interest. Yeah, back to others ---> ( TIA anyway )
Actually the idea was to do the quick "two configs" update on Westhill, in time for Christmas 2013. It has turned into something more.
Rockingham is nearly ready, that's true, but we don't think one track is enough to justify an upgrade to S3.
The plan was to get Westhill out as a quick update, then this year I would have done the tyre physics while Eric finished Rockingham and at least one more track, ready for S3 release.
Obviously, plans change and this time, they changed! Westhill became a much bigger thing so Eric has been on the case for an extra year on it.
A 'quick' 3D update turned into support for the Oculus Rift DK1. Then an overdue conversion to DX9 led to me finally learning how to do HLSL (for pixel and vertex shaders) so I was able to make the car draw a lot faster by using a single pass to draw each car instead of three passes. There is great potential for improvement of the car draw and the world, by updating the drawing system and taking advantage of programmable shaders. The Oculus Rift DK2 came out and that was seriously worth supporting. I don't want to hear what a waste of time that was, from anyone who hasn't tried it, because they can't imagine something better than looking at a monitor. The Rift gave us a very close look at the graphics and we were able to do the very significant update to get rid of many of the "alpha test" textures, most noticeable as an improvement to trees. It has been a productive year, though many people just forget the good things, because they still haven't got what they are waiting for, and also because they want something different, not just an improved version of what they already have.
Hope that explains why we are doing a Westhill update, until the next person asks the same question!
As I understand it, that sort of lightmap would need all the environment covered by non-shared pieces of lightmap textures. Then these textures are pre-rendered to find out the lighting at each point. Finally in real time this lightmap (which contains the effect of shadows and ambient lighting) is used to influence the pixel colours in the final draw.
For this, I would first need to complete the conversion of my shaders to HLSL (not hard) and I think the hardest part would be trying to create a mapping of the entire environment onto the set of lightmap textures. I have not done any feasibility calculations on this.
I was also reading up about shadow maps the other day, where a shadow map (Z buffer) is rendered in real time from the position of the sun. It also sounds interesting, but brings up many issues to solve. Apparently pretty easy to get a first version but becomes more complicated as you fix the problems, such as a requirement for multiple renders, at lower resolution in the distance. Maybe wouldn't work, I don't know, but it's interesting. Can deal with moving objects just as well as static objects, but perhaps not an optimal choice in an environment where most things are not moving.
Both of these depend on programmable shaders so the next thing before trying to improve shadows is to complete the transition to HLSL. The plan is to release Westhill before getting into any such complicated things.
I did have a few thoughts recently about a quick test on a simple improvement for the vertex shadows, one that I could try in a few hours and I might give that a quick go. The idea would be to split the triangles in a better way that doesn't create "shards" of shadows. This would be a quick improvement on the current system, testable in a few hours and does not affect future consideration of shadow maps or lightmaps.
I have a work in progress on my computer, which involves various changes to the default car shader. It includes a real time rendered version of the environment map, drawn from the car's position. It's not a cube map, it's just a single wide angle vertical view, distorted into a circle. In a car game this is by far the most important thing as the most visible reflections on a car are of the sky and objects that block the sky. At least it only requires one vertical draw of the environment per car, instead of 6 images of the environment per car, as required for a cube map. It's nice to see the trees and buildings zipping past the car as you drive.
I don't really want to work on the current system. It seems to me this is another thing to finish after the Westhill update. If you could tell me, on your thread, exactly which extra data you need, maybe I can make a note to see if it can be included in my planned update.
Keep up the great work, Scawen. I know that sometimes it might be difficult to please this audience, but please, keep working in the way that you have always done! You along with your team have proven to us that you guys surely can't disappoint when you release something completely fresh to this game. So please, don't give up and keep it going! I still truly believe that LFS is the greatest racing simulator of all time, but you just need to improve it once again like you used to before..I know you can!