Jump to content


Popular Content

Showing most liked content on 10/11/2018 in all areas

  1. 5 points
    I can't take credit for it, but it needed to be shared. This made me cry with laughter.
  2. 1 point
    you can also just use UV Transform SOP: Transform Order: Trans Scale Rot Translate X: 0.5-$CEX Translate Y: 0.5-$CEY Scale X: 1/$SIZEX Scale Y: 1/$SIZEY Pivot XY: 0.5
  3. 1 point
    So, 3D apps get harder as you spend time with them? I must be doing it wrong
  4. 1 point
    hi, I can only repeat what I have said before. for texturing, use masks you get from heightfield erode, plus you can generate your own hi-res masks in shader using mesh masks with shader noises and position data. you can do full terrain texturing procedurally in Mantra. Of course it depends how much detail you need, how specific look you want and how much time you have. Using triplanar textures to finalize shading/displacement is a good idea. For general rock displacement detail Worley/Manhattan/Chebyshev noise patterns are good but it takes some practice to get to good results but you'll get great deal of flexibility and no tiling issues. Best for large areas. Otherwise triplanar projection of multiple sculpt maps is good too. Ideal is to use both. Painting huge textures is a bad idea in my opinion. Too much manual work, resources required for data handling, plus youre f*cked if somebody decides to make more significant changes to the terrain. Dunno what level of realism are you aiming at and whether you're pathtracing or using realtime engine. This may impact your workflow to some extent (Ive got minimal experience with realtime stuff). I personally never used LODs for this but I don't see any reason why you couldn't use it if you need it. \ cheers.
  5. 1 point
    The infinity fabric has lower latency than socket to socket. That said it's still very sensitive to memory timings and speeds. Unlike Intel processors, the Ryzen processors must have the timings dialed in and the processors respond better to higher to memory frequencies(the less memory you have the faster it needs to be). In the case of the 32 core Threadripper it can only be run in NUMA mode which windows is absolutely terrible at handling. Slow, improperly timed memory + bad NUMA + bad scheduling = performance regressions with a 32 core processor in windows. Having slow improperly timed memory will also hurt your performance in Linux, but at least you will have superior NUMA and scheduling.
  6. 1 point
    Stretching of UVs on extreme vertical changes will always be the weakness of a system using Y Projection UVs. In Alex's example, he doesn't have any vertical displacements nearly as extreme as your example, so that would be why you don't see the degradation you're seeing. The other weakness isn't just UV degradation, but also the lack of much terrain detail in highly sloped areas. I recently completed a project that had very large terrains, with a few extremely vertical mountains. The workflow that I found to work best for this situation is to separate the extreme mountains and map them individually (or sculpt extra detail into them as we did), while keeping as much of the remaining terrain as possible as heightfields. You can use some of the heightfield masks to isolate the large elevation changes, but I was manually selecting areas where the mountains were for separation. This worked in my case as the output was highly driven by concept art, so it needed to be very manually directed. As for texturing, I found exporting the heightfield masks to be very useful. I did a small amount of processing in COPs, but not much, as I preferred to use Nuke to do essentially the same processing. There wasn't much direct texturing work we did with the masks, as we mostly used them to drive scattering of trees, grass, etc. We did use painted texture maps for the rocks and such, and masked them in via shaders. Let me know if you have any specific questions about workflow etc.!
  7. 1 point
    hi, here are some answers: as far as I am understand it heightfield will always have the same resolution across its entire area since it is a volume. so you shouldn't need to care about that (at least I've never noticed any resolution issues). when it comes to texturing, I wouldn't even think about using uvs and traditional approach of texturing things as that's very painful process for things like terrains (unles you have a very good reason to do so). what I am doing is just using various masks generated by erosion sim, combined with position based procedural patterns and running it through Mantra. to me, this procedural approach seem to be quite convenient and fast. if you need to, you can always bake into uv textures and export do different renderer/realtime engine. cheers.
  8. 1 point
    Maybe a little too inside baseball, but some interesting peeks at the internals of Houdini here:
  9. 1 point
    If you don't have the original geometry and therefore no access to that initial resample. This should work. vu_tubecentre_end_bit.hipnc