100 Mile Visibility

First, Happy New Year! As is typical, I've been quiet on the blog because things have been insanely busy here at work. Just to give you an idea of the insanity:

• There will be a 9.63 relatively soon - the bug driving this is some Linux distros not finding the DVD. But we'll get a few new datarefs in there too.
• We have new updaters and installers to get tested, again addressing Linux DVD issues, but also with updated web download code that should give a nice speed boost.
• Chris has been working hard on Android. X-Plane for Android is pretty much the biggest APK anyone has tried to ship, and as a result we've hit a number of problems with the market that we are going to work around.
• All that's just the side show; X-Plane 10 development is of course the meat and potatoes.

Now, about visibility. X-Plane 9 restricts ground visibility to 25 nm (about 46 km) in an attempt to prevent you from seeing off the edge of scenery tiles. Many users have expressed (some more persistently than I would have liked) an interest in longer range visibility. Austin recently posted a note to X-Plane.org discussing level of detail and distance management in the new weather system, and users immediately picked up on his mention of 100 nm visibility. Here's what we're thinking; all of this is subject to change as we keep working on the product.

First, visibility: you can come up with a formula for the distance to the horizon based on height above a sphere: d = sqrt((r+h)^2 - r^2) where r is the radius of the planet and h is the height above the planet. Since the Earth is roughly 6 million meters in radius, we get a visibility to the horizon of:

100 meters: 34.6 km
500 meters: 77.4 km
1000 meters: 109.5 km
10000 meters: 346.5 km

Clearly a little bit of altitude lets you see a long way.

But there's more to it than that: X-Plane has always changed the visible distance with altitude. The 25 nm limit applies to surface observations (which is what you get from a METAR). As you move up into orbit, that distance is scaled out to the horizon distance, so that you can see the whole planet from orbit. That scaling can reveal the edge of DSFs, which are blended into the planet when volumetric fog is enabled.

So here is what I think we really need to do:

• We do need a larger 'surface level' maximum visibility, so that distant features are visible from the ground.

• We need a scaling from ground to upper atmospheric visibility that gives us more visibility sooner; one of the problems with version 9 is that the increase of visibility is slow, which gives mid-elevations a hazy look.

• In the long term, we need to load more DSFs, probably twelve instead of six. X-Plane 10 already has some improvements in how scenery shift is done, but my guess is that we can't productize this until we have a 64-bit build (since more DSFs chew more memory), so I expect this to happen in a patch.

• We need to add elevation displacement to the whole-Earth planet render, so that the blend between DSFs and the planet don't have huge height gaps at high-elevation locations. I am hoping we'll have this in 10.0, but it is not coded yet. (Usually we recut the planet textures last, since they are cut off of the DSFs.)

• We need to improve the quality of haze, fog and atmospherics. In real life, atmospheric scattering reduces the contrast of far away terrain. I believe that correct scattering could make a huge difference in the quality of the transition from DSF to planet, the required tex res (we need less if we scatter more), and generally it would be a big contribution to the realism of the image.

  I'm not sure how much of this we'll get into 10.0; I have a prototype of Sean O'Neil's atmospheric scattering shader from GPU Gems 2 running in the sim, but I don't think it's shippable. I do hope we'll get at least some scattering in place, with improvements in patches.

That's a road map, at least. If there's a take-away point, it's this: increasing visibility is complex and involves a lot of parts of the sim and there are still significant parts that need work. So I really don't know if we'll hit some kind of hitch or problem that requires us to back off visibility.

Austin's comments about 100 nm visibility reflect what the slider in the sim happens to be set to now. It's also a design goal of the new weather system - that is, we want the new weather system to handle significantly larger distances (and have better scalability) than the old one did.