Nightly 4b32de3 - Render Test

[another_commander]

Please take note of nightly 4b32de3. It's a special test build. A couple of things have been switched temporarily on and should give everyone interested a chance to check out the latest additions in the gane engine's rendering capabilities. The changes are:
1. Sun light source radiance has been increased. One of the complaints we frequently get is that objects in the game can be often hard to see. With this change, this is no longer the case. The sun will now generate really bright light that makes everything in a system brighter and much easier to discern. It should help a lot in combat. The tonemapping we apply on all objects ensures that a good relationship between too bright and too dark areas is maintained, resulting hopefully in lighting closer to what would be expected in space.
2. Oren-Nayar diffuse rendering is on. This is a more subtle change and many may not notice it even. For objects that are supposedly glossy this will indeed be the case. The real difference will be visible on rough objects, like asteroids for example. These should obtain a slightly "grainy" substance on their diffuse reflections and their appearance should more closely resemble natural rough surfaced objects. When you notice it, what I am trying to say here will become clear.

Both changes are applied to system objects, as well as stellar bodies.

Of course, there is a trade-off somewhere. All objects that depend on the default game shaders will be affeccted by the change and all objects that run on their own shaders will not. For example, gsagostinho's ships will take the new effects and use them, while Griff's shadered ships will have no difference whatsoever as they run on their own shaders. If you have both types of objects in your game it will appear that either some objects are painted in dark colors or that there might be something off with the lighting. It is probably best to test without OXPs first.

This is for testing purposes, so the plan is to revert these changes soon, unless there is overwhelming approval, in which case we can consider keeping them. I just want to have a build (both on Windows and Linux (and hopefully Mac)) with these settings turned on somewhere so that people can see what else the game engine can do, test it and give feedback. If interested, keep an eye out on this thread; as soon as the nightly is generated, the download link will be provided here.

Edit to add: No screenies posted here, if you want to see what it looks like before trying it out, refer to the latest shots I have posted in the Screenshots thread - they have all this turned on already.

[Milo]

I like it. I vote to keep the changes in.

[another_commander]

The nightly build has been published for Windows and Linux and is available for download from here: https://github.com/OoliteProject/nightl ... 9.0.200713

[Commander_X]

Hopefully this will contribute to the "overwhelming approval"
I'm in!

[Milo]

For now I've added all the noshader ship variants to my Ooniverse (as shipdata additions) while keeping the Griff shader variants as replacers. I see a lot more diversity now, and so far nothing I've seen has felt out of place.

[Cody]

Can't do any testing as I'm grounded, but I had a quick look at a Povray Neptunian in interstellar space - far too bright.

[cbr]

some planets looks great




objects are a bit too light/bright




some planets look interesting


I think it could use an illumination setting on the menu
perhaps differentiating between planets and objects

[another_commander]

I think it could use an illumination setting on the menu

What I am envisioning with it is that sun radiance is set inside planetinfo.plist, then this value is used as a uniform that gets passed to the shaders. I would agree that editing the shaders directly is not the best way to go about this kind of things, but it is what we have at the moment. There is definitely room for improvement. A game option is also a possibility, but it will also require passing the desired values as uniforms to the shaders. Both scenarios are doable, but I just don't have the time needed to sit down and do the deed. If there is anyone out there who is willing to give it a try,I would love to see a pull request for this at some point.

Regarding the planets you call interesting, part of their interest is a current flaw in the system. The cloud layer is presently not affected by changes in light radiance because it is not shader-controlled. Dark clouds remain dark even with radiance multipliers set to 1000. Ideally this would require that the light radiance is set within the core code with its desired values and then is applied normally on all objects, with the shader multipliers set to 1.0. This would also make the radiance apply to all objects by default, regardless of what shaders they use. I tried doing this at some point and it almost worked, but I hit another wall: there is a bug in the cloud layer code, which makes it render in a very weird way when viewed from certain angles - btw this is something you can see already on Lave if you go round the planet so that it is positioned next to the sun; you can then see that the cloud layer renders in what appears to be triangles. This caused planets to become really, really bright so this approach will have to wait until that darn cloud bug is fixed.

[another_commander]

Here is a demo of the Oren-Nayar diffuse BRDF that this build is using. The Oren-Nayar diffuse light model takes into account the roughness (or, in Oolite's case, glossiness) of a material's surface for its calculations. When a material is fully glossy, its diffuse coincides with the diffuse given by the Lambert model, which we've been using forever. The asteroid below is given different roughness (or 1.0 - gloss) values and you can see how the diffuse light model reacts as the material gets progressively rougher.

[dybal]

I just synced to the last commit (ba63810) and re-build.

The experimental sun radiance is still on, isn't it?

It feels like there is a sun flare all the time, even with gamma at 0.6, near a Kiota Reserch st, 4300km from the sun (something like 60% of the PS lane), the glare was so much that glare filter was gone in minutes.

EDIT: I was planning on taking some pictiures at 60% of the Main Planet - Sun lane to show the glare, but... the glare isn't there! I guess it was really a flare...