T3D1.1B3 Terrain cliping issue - LOGGED (THREED-1551)

#1 Just thinking out loud here since I haven't messed too much with cutting holes in T3D...

But does adjusting the terrain screenerror field help or worsen the effect shown in 2nd pic?

Does it seem to matter if your holes are on a slope like the above the pic or on a more level area or terrain?

#2 it happens only with slopes that I have seen. I will check teh terrain screenerror filed

#3 The problem is where its lowering the polycount for the screenerror setting applying LODs to the terrain when you get some distance from it.

#4 screenerror was set to 16. I lowered that till the issue disappeared which was setting of 4. But now, performance is worse.

#5 May have to compromise with a larger outside edifice to get around the screenerror + LOD + performance problem. I wonder if you could slightly terrace that particular area where the hole is to help combat the loss of precision at distance and lower lod...

#6 Neat! I have never considered(by design contingency), or discovered(from testing design expectations) a situation like this...Ever in the over20 years I have been in(and out)of the industry. But now that I know its a possible situation, I can foresee many situations where this will become an increasing problem by proxy of T3D's unprecedented combination of following systems; large terrain size, near infinite render frustum, aggressive and clever terrain LOD(a real piece of art in its own regard), Coupled with the fact T3D's distance fog is not needed as a FPS optimizer, nor necessary as FAR occlusion.

I postulate a manipulation of LOD on an 'patch' Static Mesh would be the quick and easy to implement solution without much time investment, just some logic based experiments(As low as 4 repetitions) to find optimal mesh LOD numbers.

An solution would be experimenting with the LOD of a very simple 'sheet' static mesh designed, by visual nature and size, to cover the 'Terrain LOD artifacting' hole. Fight LOD with LOD! (dont apply the same concept to things such as heartburn, adding acid to acid is bad!).

A 2 LOD state mesh, with a very large bounds box(the bounds will be important, make it insanely large). One mesh blank with LOD(SMALL LOD size) set to something reasonable to shut it off when near distance is sufficient to not manifest the 'Terrain LOD artifacting'. Other mesh is your illusionary 'patch' mesh set with LOD(Large LOD size) that takes over for far away view distance.

I dont foresee any plausible problems, but am fully aware of how hindsight is the only perfect sight. I can think of an possible optimization, but I expect you will discover it for yourself(assumption bias from observation of screenshots and knowing what type of cleverness is required for such meticulous terrain level design, its gunna be exciting with details and polish finished up).



Im curious to the solution of your problem, keep us posted...

#7 this seems to be an issue even when a hole is not cut in the terrain

and


it seems to show up on very steep terrain where the terrain has an almost verticle slope

#8 are you on an ATI by any chance ? On a Nvidia ive only really noticed it when theres a hole in the terrain, and the LOD stage has made the misisng face bigger and more noticable. I havent seen random missing faces like that though since i was on an ATI. Will have a play around with steep faces though to see if it replicates

#9 naa.. I am on 8600s

#10

Quote:this seems to be an issue even when a hole is not cut in the terrain

I do not believe that is a real bug(Functioning inconsistent to design expectation), in as much as excessive artifacting from LOD and extreme Z-space differences, you may take notice that it manifest inconsistent with movement and camera angles, slight indication that some form of precision is at fault. This in so way meant to excuse the probability that it is unexpected functionality, and quite possibly correctable by code evaluation.

As support to my observations, experiment with high to low LOD terrain settings and watch how it can be minimized, eliminated, and even expanded...

Another caveat you will soon find, place a mesh ontop a hill positioned so the shadow would fall 'over' the side and spill onto a terrace below. You should soon notice some very rumbustious inconsistence with how shadows are expected to function, this abnormality is amplified by position of the sun object.

#11 @Donald

If you send me your .ter file that shows these problems with gaps in terrain i can look to fix the issue. tom@sickheadgames.com

#12 @Tom,

its on its way and thanks

#13 Logged at THREED-1551.

#14 Still Present in the 1.1 preview. Also the bad edging when there is fog is still there. Just to the centre left on the horizon

#15 Is there any progress with this issue? Changing the terrain clipping helps with the holes but really effects performance. Was really hoping 1.1final would have a solution but guess it wasnt ment to be.

#16 The issue is still present and was included on the Known Issues list.

#17 Has anyone been able to fix that "bad-edging" with fog bug?

#18 The problem is still there. The quickest way to fix it is to make your slopes slightly less steep (and they look better anyway; I hate the look of long vertical slopes with the jagginess and extreme texture stretching ).