Render issues with "complex" function driven textures

[cyphyr]

I'm just starting to get a handle on the (very) basic end of function driven landscape, displacement and texturing inspired by some of the stunning canyon renders I've seen here recently (you know who you are ). I'm using the "Voronoi 3D Diff Scalar" to create rock displacements and to then texture those rocks. The problem I'm running into is that sometimes the scene wont render at all, just sits there until I choose to "stop" at which point it crashes or alternatively it is trimming off the top of some of the displacements or/and showing holes with sky showing through. Is it possible that some of the function nodes are creating mathematical loops that TG2 has difficulty in properly resolving? I've had a number of other issues with this scene, such as "Fake Stones" not showing up. Removed them and re added and they behaved normally. Another issue is I tried to add a water plane at a level a few meters below the camera, checked the camera height via its dialog box, checked the height of the terrain via mouse over, put the water in between and nothing there, its a no show. I've never had any of these issues before and the only thing I can see thats new is the function nodes.
Thanks in advance
Richard

[Oshyan]

The issue of displacement being cut off is a known problem that occurs at the edge of render tile borders. This is something that should be improved, if not entirely eliminated, in the future. The rest of the problems sound more odd and uncommon. Are any of them truly reproducible? If you can provide a .tgd project file that recreates a particular issue we'll have a lot easier time tracking down the problem. Otherwise it could be any number of issues.

One thing that does tend to create odd random issues at times is memory problems. How much system memory do you have and how much is free? I see you are on Vista, but what version, and 32 or 64 bit? Unfortunately, even with additional information, it may be hard to track down these specific problems unless they can be reliably reproduced.

- Oshyan

[mr-miley]

I cant comment on the displacement issues, but I have had the problems with the water plane before, and the same problem with object distributions. I have just finished working on a scene with a water plane, where placing the cursor on the landscape gave me a height, plug that into the water height and..... nothing. In the end I had to plug in ever decreasing (or was it increasing??) heights till the water suddenly came into view. Now I know that lots of displacement on the landscape can "alter" heights from those shown, but my landscape had no displacement applied, and also the farthur away from the camera you get the bigger the difference between the "actual" height and the reported height (isn't that something to d with the curvature of the planet???) but I had to alter the height by hundreds of metres, not just a few. Had the same thing with a distribution of trees on a hill jutting out of the water. The water went in at the correct reported height (for a change ) and I wanted my tree distribution to start a bit above the lake. placing the mouse on the preview it turned out that I wanted the distribution to start about 40 m above the water level ( the lake was set to 375m) I set up the population with a distribution shader to controll the min altitude, set that to 420m hit populate now, and got no trees at all, not a sausage  after much fiddling with the min altitude setting, I finally got the population as I wanted it, with the min altitude set to -900m. We are talking 1300m difference between  the reported height and the eventual height setting. In the words of Mr Gumby (if your'e a Python fan you'll know who he is ) "BRAIN HURTS.... BRAIN HURTS"

Ta

Miles

[cyphyr]

To be honest I've just about given up on using any kind of numeric feedback as regarding heights, the info almost never is particularly useful. I think you right about heights being thrown off by the curvature on the planet surface but displacements also seem to play a part. In the situation above the water plane was edge on to the camera so it should have only been a case of slowly nudging it down. Accurate position feedback is a must particularly when integrating between other apps which TG2 will need to do in order to take its rightful place and become a professional tool rather than a quirky hobbyist toy!

Richard

ps @ Oshyan: To answer you query, I have 2 gig ram and am running Vista 64bit Ultimate. I should be getting another 2 gig ram in the next few weeks. I'm surprised that your not planning on "entirely eliminating" the cut off problem, dose that mean that images where this occurs should just be abandoned?!

[Oshyan]

The issues of height with relation to distance from the origin point should be significantly improved in the future with better coordinate systems for planets, etc.

As far as the displacement cut-off issue, yes ideally we'd like to fix it entirely, but it's dangerous to make promises of that sort. In all likelihood we will be able to fix it for 99% or more of situations. But it's possible it may still occur in some extreme circumstances and then we'd look a bit foolish if we said it was entirely eliminated. I think it's not something you should worry too much about.

Just keep in mind that most other renderers that deal with extreme displacement have similar or other equally problematic issues in cases where displacement is very high. MojoWorld is a classic example of a renderer that depended heavily on displacement and it had similar issues, requiring very large tile overlaps in some cases (significantly increasing render time), and sometimes it just wasn't possible to fix.

When you let people go wild with displacement, not putting any restriction on how far things can be displaced, you either have the potential for absolutely insane render times to make sure you calculate everything properly, or you miss bits and the end result is wrong. But again this only happens in extreme cases. It's a thorny problem, but one which we can hopefully all but eliminate.

- Oshyan

[jo]

Hi Oshyan,

It's a thorny problem, but one which we can hopefully all but eliminate.

Yes, it is much more a problem with thorny surfaces created using extreme displacement ;-).

Regards,

Jo