Some time ago i meet a guy who was looking for a job, presenting himself and his work to the public, and was dissapointed because it was not sufficient enough for the standard they use for the project, and now he is more arrogant as my mother in law.
First of all, thanks for taking your time and taking this into discussion with us. I guess we both understand GI, so it should be obvious for us that sky has noticeable contribution but it's not all there is - especially locally. I definitely like there's some blue tone in shadows, but like I previously said: And that's what I think of these screenshots. While I like they have non-monochrome ambient, it ended up being more apparent than I would like it to. I think the ideal situation is to have an engine with GI that produces accurate shadow colors and then to have HDR implementation running on top of it, partially adapting to dominant color (still leaving something noticeable for our eyes to deal with ). Not having GI (I understand why) we need compromises. That's why desaturating ambient color in such lighting engine is perfectly fine - just not complete desaturation (in my opinion). Perhaps a 50/50 or 67/33 combination of monochrome/color ambient would do the trick? EDIT: Tosch beat me to it, although I don't know what kind of proportions he used Anyway, I just wanted to make you consider the subject - like I said, adding my opinion to yours. Also, thank you Tosh, for adding yours |EDIT2: Good example Lot's of orange bounce but also some blue shift on white paint on tarmac. Nice reply from Tosch here as well. I think some point was reached in this discussion. I'll just add a small reminder - I also wrote about lighting during and past sunsets. This could also be interesting playground for trying various mixtures of monochrome and color ambient. No need to post them here, I think. It was never my aim to consume your time with long discussions - only to raise some points for consideration. Thanks and have a good sunday (at least as good as mine, but it will be hard to beat ).
Personally I recognize the effective efforts that are being made in order to improve rF2 GFX. It's working. My only complain is about the "yellowness". I don't know if it's the fog colour or something but distant objects looks more yellow than should for me. Maybe it's lack of blue, but I'm speaking out of my impression.
Wow, today Unity3D released v5 also for free*. This war is becoming interesting. *Non of them are actually free. The tools are free, you have to pay something when you release the game. Still, I think the price of these tools are nothing for a company like ISI to be worried about. The problem lies in making it all work together. And that's why I really don't have hopes for that.
According to the website licensing costs are 5% of the games price. https://www.unrealengine.com/what-is-unreal-engine-4 Anyway they seem to be very confinced that they have a good set of tools not charging upfront.
It'd probably be cheaper/better/easier to just go straight to the source of Unity5's, and (and integrated into UE4's and Frostbite's) rendering engines and get Enlighten from Geomerics...if anything, that is. I'm not suggesting that ISI do so, but if they were to, it seems that these guys would be the ones to go to.
I know this isn't 'my' wishlist, but it seems to pertain to the topic at-hand. Here's a cool video from GDC this year of a large outdoor environment done in Unreal Engine, that they had to develop techniques for. I share it as source of some cool inspiration for potential graphical enhancements for the future, perhaps: https://www.youtube.com/watch?v=clakekAHQx0
Well, things move forward, but if you develop your product over years, which is typical for complex simulators, you would need tons of effort to stay up to date. These engines are excellent for game studios that work fast on "typical" games (you know, cinematics, animation, dynamic lighting, latest image processing, etc.). It may not work so well if you do a "scientific" project, like racing simulator. For example - if you want to deal with latency, or support controllers with unusually high frequencies - these engines may not give you enough control over these aspects, since they focus on other things. As for surveying, it's always good to learn from others. On the other hand, if you don't have the engine that could take advantage of it, then it may be pointless to survey it that way. Even if your engine can take advantage of some information in the future, it may turn out, that you need to repeat surveying ares, because it changed over years. Especially true for racing circuits, since I don't suppose end user would tolerate getting circuit that is out of date, while in shooters or other games it may make no difference, since areas are mostly unknown to users anyway.
@Tuttle, love the effort and the detail being put into the sky and the sim in general. It makes a big difference when using a projector and racing an open wheel car.
Fog color is set manually by modder, most rF2 tracks has it at 205 215 235 which is somewhat bluish-grey (defined in SCN file). Speaking of fog (and after reading this thread describing way more complex things with rF2 graphics engine), it would be nice if fog could be controlled in more flexible way, which was possible in rF1. Currently it limits modder to use exponential fog. The shape used, which cannot be changed, doesn't work well at all in my eyes for simulating heavy fog. Put fog to a high number and it affects objects nearby too much, making the entire scenery grey. It's not possible to simulate a "fog wall" (crude example). Also there is something else about rF2 fog that looks unrealistic, but I don't know this subject well enough to specify.
I recently came across this site when doing a little research: it seem to be worth having a look. http://skyrenderer.blogspot.com/ I think the important thing to garner from it is his latest post about the effects the ozone has on ambient lighting at twilight. You can clearly see the amount of blue cast from scattered light from the sun after it's just about gone past the horizon. I'm not necessarily suggesting that rF2 goes quite this far, but it seems that ambient light, currently, is a bit dark around this time in rF2.
Yes, that's what I talked about in my first post - about this common mistake made in many games. Thanks for the link I think the tricky part is first to realise just how blue the world is at that time and then not to make the world that much blue in game. That's because in reality our vision / brain will adapt to such blue light and won't notice it that much, but if you display something so blue on monitor it will be too apparent to us and look overdone. So ideally I think you should get your math right in the rendering engine, but then make your HDR implementation play it down a little, by adjusting not only to brightness of image but also to dominant color. I'm not sure what would work better - desaturating the final image all the time or desaturating it only when one color starts to dominate too much. That's just something to play with and see what does the job better.