Damn… And I hoped to get those elves into the game 
Well, in the real world a photo camera does crap all the time. If the PBR is P-enough you’ll get overexposures and so one which I’d love to see. What I really wonder is, why the bad shots should look bad. Is it, this kind of situations, where real photos look artificial although they are real?
For PBR to look really good, you need multiple direct light sources from multiple angles + IBL. If the Hellion is in a hangar, it will have these many varying light sources. It has many lights on its hull, there are many lights in the hangar etc. When on a planet surface, there are also many light sources, but there is one that is overbearing, the direct light from the star. The more a singular light source out-intensifies the “competing” light sources, the “flatter” the asset will look. If you play a game like Star Citizen, no matter where you fly the lighting on the ship looks very similar, like it is in a museum display, but highly unrealistic (they seem to have phantom light sources placed in their scenes from what I can tell) in space or even on a planet in full sunlight. So while our system is more realistic, it might be different than what most, if not all games, present to the player. Which is a very consistent, artist fine tuned lighting setup for each scene. This is all still a topic of discussion amongst the dev team. The solar system scaled game world prevents the artists from tweaking every possible scene scenario to lighting perfection, and the realistic stellar intensities place some very demanding challenges on the art PBR-wise.
Does light intensity vary over distance (in-engine)? If so, the ship should be fairly unlit, barring flight lights, simply traveling between stars or other low light situations.
This might also legitimize black camos/skins as a viable way to avoid visual detection.
Yes. However there is a cap to the intensity the engine can achieve. So for a given star you will reach a distance from said star where traveling closer will not yield any increase in intensity. It is not known at this time whether or not we will normalize & scale stellar intensities of real high output stars to the engine’s capabilities. As I said, there are still ongoing discussions.
Would it be possible to have an “clear-view mode” using such unrealistic-but-equal-looking light à la Star Citizen?
Fiction-wise, your ship’s computer rework what you see so you can better do your job as a pilot - like Shattered Horizon’s sound, or like a better version of a HUD overlay. And similarly, it could be shut down in some conditions…
I have no idea how much effort that would represent, so I’m curious.
Sure it’s possible, but that would likely be a last resort option for us I think, as it defeats the purpose of creating a realistic lighting model in the first place.
I’m happy to hear this. I hope you guys can find the cool in space.
That said, how much realism can the engine provide in terms of lighting? Everything on the Hellion in that last still should be fairly well lit. There seems to be lots of stuff around to scatter light, so why isn’t the ship more evenly lit?
That process of reflecting and scattering light would also help with ship lighting in space. If I were to flash a light on one part of the Hellion, the light would ideally reflect and scatter across the various bits and pieces on the hull. This is classic ray tracing stuff, but I wonder if a more primitive form has made it to the world of rasterization.
Well, if you manage to make it look realistic and gorgerous yet practical, I’ll have to retroactively pledge even more to the kickstarter 
Not necessarily. It seems like the light source is coming from the horizon (hence the… what do you call it, @Hutchings help me out… sunset? 
), so the light would come horizontally unto the ship. So the actual reflection from the ground probably won’t light the underside of the ship IMO.
Just wanted to say Thank you Hutch for the steady stream of screen shots (Say that five times fast). I’m absolutely loving the pictures. They look awesome. I’m really looking forward to the kickstarter. Count me in =D
Judging by the angle of the shadows, the star is fairly high in the sky. It is simply that the color of the light is dusky. That doesn’t really matter because we don’t know how much light is getting from the star to the surface.
I went back and looked at the larger version of the still and it looks much better at that size. There is clearly light reaching the underside, so surface reflections/scattering are implemented. They do seem a bit on the dark side, however. That may be due to the material of the ship.
In the end, the ship doesn’t look right to me. I don’t know the source of the problem, but the one before it suggests a mechanical object on an alien planet. The scene is crisp and clear, reminding me of the pictures of Apollo and the various Mars rovers. In contrast, the second one looks like… aha! That’s what it looks like! A watercolor painting. Straight off a paperback sci-fi book cover. Just slap a title on there and you’ve got a Bantam cover. It just doesn’t look like a real mechanical object. The lack of light on the underside only adds to the uncanny valley feeling for me.
If the Hellion is a low-quality model that’s being kept around for use in Kickstarter backgrounds, then I’m not concerned. If it’s a high resolution model for the game, I’m concerned.
We are currently doing rudimentary approximations of global illumination via a combination of ray and cone tracing. This is due to both performance considerations as well as the fact that we can’t just work on graphics forever. Currently we do not calculate indirect illumination caused by bounces from non-celestial objects. It came as a huge surprise to us that this actually has a much larger impact on objects in space. If you look at real-life pictures there is a huge amount of illumination, relatively speaking, on the dark/shadowed/unlit side of objects caused by inter-object indirect illumination.
Fortunately since space is filled mostly with space and planets are conveniently spherical, when you aren’t standing on them, it’s much easier to ray-trace than conventional game environments. I’ve done quite a bit of research on this and I think that, leveraging the numerous enhancements of d3d12/vulkan, we could achieve real-time ray tracing in space on high-end hardware - particularly the next generation of GPU’s coming by sometime next year. Of course I am intentionally saying “in space” as ray tracing atmosphere is extremely expensive and once you begin getting close to the surface of a planet stuff starts getting real complicated (story of our lives).
This would also of course require a heavy engineering investment that we will not be in a position to make for quite some time. Also people seem to think tree’s and clouds are more important anyway ;).
The problem you are seeing is a lack of indirect illumination from the ground combined with a rudimentary calculation of indirect illumination from the atmosphere. Obviously it’s reasonably subtle, visually, and since it’s incredibly difficult to solve for from a technology standpoint we simply aren’t bothering at the moment.
I still think that the level of detail makes this close up shot look a bit unrealistic. The texture suggests a scratched old surface while the silhouette of the ship look brand-new…
Edit: actually I’d love to see how it would look if a post-process screen-space filter would displace the contours a fraction of a pixel. If this would be based on the grain of the bump-map at that part of the contour, this could add much more photorealism
The Hellion is the “newest of the older assets” we use in the KS video. The asset was designed/concepted and modeled before we even had a mesh importer, before we had a material system, before we knew max tri counts, etc. The ship is only 11k tri’s, about as much as a Star Citizen helmet 
So yes, now that the artists have run a few assets through the pipeline, we have a better handle on the required fidelity going forward.
As stated above the mesh is only 11k tri’s, and there is no damage mesh. These would contribute to what you are seeing.
I’m a bit worried that as a small development team you’re taking on board fairly nit-picking criticism and comparing yourselves to titles with huge budgets.
What I see looks really good and is certainly “good enough”. Aiming for perfection will burn your budget and leave you with a half finished game when the money runs out.
I agree with Crayfish. People should stop comparing us to AAA games that have entire companies and millions of dollars of budget behind them. There’s no way we are going to compete with them, even if we get funding from KS.
Guys, we’re a small indy company. Look at what indies are doing, that’s a better comparison. Saying that “your game will suck compared to your competitors”, when such “competitors” are Star Citizen and its $80M budget, or Elite Dangerous and its 200 employees…
Please lower your expectations…
Hey!!! Can you please stop pulling Star Citizen out of your sleeve all the time… I don’t even know this game!
(I was just scrolling up this thread and “People” in this case are ThornEel and all people from iNovae. These are the only ones comparing this to AAA games… Mostly inovae ppl themselfes)
We are talking about Infinity Battlescape. And we are talking about feasible stuff for INovae. I would never complain about any ship, if I wouldn’t see a close-up like this. Additionally the indirect illumination is missing (which will appear visible in game) and this is an old model.
I was talking in terms of budget. It doesn’t matter which game it is. What matters is that we are indies, and that people are disapointed ( or will be ) that our stuff doesn’t look as good as a AAA game. That’s a matter of expectations.
Yeah, in terms of polycount it’s an old model. I’ve been experimenting with level of detail recently and it looks like we can easily pull of an order of magnitude more polygons and still run well. But remember that we’re going to have to display hundreds of ships on screen. They can’t each have millions, you have to stay within a realistic budget.
Indirect illumination is missing, but solving this, as Keith said, is a major technical challenge. One we cannot solve on minimum Kickstarter funding. Therefore saying that it will be solved for sure in the final game is wishful thinking at this stage.
I personally think our stuff will look really good for an indy game, and will fall short of AAA quality.
Interesting. I personally think, your stuff can look better than AAA companies stuff technically. But it will never be able to compete with them content-wise. You have the chance, to make own decisions, thats what AAA companies often don’t have I think. The first and major decision you made was developing a very own engine.
Now you are making decisions like:
…which I really like!
In other words, I get the impression that AAA games are no real competitors to your game. They are different. This is like comparing music from the Top 10 Charts with your favorite music.