Welcome to od|forum

Register now to gain access to all of our features. Once registered and logged in, you will be able to contribute to this site by submitting your own content or replying to existing content. You'll be able to customize your profile, receive reputation points as a reward for submitting content, while also communicating with other members via your own private inbox, plus much more! This message will be removed once you have signed in.

pusat

Members
  • Content count

    97
  • Joined

  • Last visited

  • Days Won

    6

Everything posted by pusat

  1. Straight VEX based volume deformation with no geometry conversion in between.
  2. I find that channel pretty useless. Just non-technical commentary on new papers.
  3. Performs PCA (Principal Component Analysis) on a point cloud.
  4. Indeed, this tool only creates connections between immediate neighbours up to N depth, so you don't need to create unnecessary connections. After that you can decide how to break up your glue network using noise, rules, etc. In your case you can easily identify the connections that belong to different material by querying the attributes as this node stores original name attributes, and which piece they originate from, etc.
  5. You can achieve it using something like this: Later on I made a more generic version called Adjacency Skeleton. Need to update the video sometime
  6. I think you need some kind of optical flow algorithm so you can certainly use PCA: https://ps.is.tue.mpg.de/research_projects/pca-flow https://github.com/jswulff/pcaflow https://asp-eurasipjournals.springeropen.com/articles/10.1155/2010/680623
  7. Before any new serious COP features though, SESI needs to fix some core bugs that are as old as the system itself. I reported many but I get the feeling that there are no plans to address any of them. There are threading issues, deadlocks, round trip (COP to SOP to COP) limitation, GPU image artifacts, anti-aliasing issues using the exact same code in VOPs vs VEX, etc. Otherwise I really like COPs too. I made an OpenCL Wrangle COP recently that proves to be very useful for image filters.
  8. You can make your own modeling tools. Check out some examples here: http://orbolt.com/search/?q=animatrix+sop
  9. +1, this is a nice RFE. I wrote shelf scripts that does this for Display flag so it's possible to script this action as well. I can send you the code if you want that you can bind to a hotkey. My RFE: In H16 if you copy some text inside Houdini such as code from inside an AttribWrangle and then select and copy a node tile, and then paste inside an AttribWrangle, Houdini will paste the path of the node inside the text editor. Before 16, it didn't do this, it preserved node tile copy clipboard separate from textual clipboard. If this is a new feature then it should have its own shortcut not alter the default copy/paste behaviour people have used to for a long time IMO. [ID# 84001]
  10. Compiling wouldn't work with Carve SOP as it's not compilable yet.
  11. It's very easy. First compute the best fit plane of these points. You can use this VEX tool: http://orbolt.com/asset/animatrix::bestFitPlane::1.00 Then create a circle in VEX that's at origin on XZ plane, then using: matrix3 m = dihedral ( {0,1,0}, bestFitPlaneNormal ); multiply your new circle points with m and then add bestFitPlaneCenter to these points.
  12. You can use Attribute From Volume SOP.
  13. This operator allows you to run an OpenCL kernel as part of your COP network. Depending on the GPU, some operators can be orders of magnitude faster than even VEX. In this case OpenCL is 192 times faster than VEX on GTX 970. Please note that the actual cooking time for the VEX node is much longer (~77s for 100 iterations vs 40s for 10000 iterations) than what the status bar is showing. The cooking timer is shown long after I set iterations to 100 as can be seen in the video.
  14. There are many ways to do this but one way I would do is delete all groups but the largest one by element count using something like this: https://www.dropbox.com/s/lkl7j2w6r65r2im/DeleteGroupsByCount_SOP_Animatrix.otl?dl=1
  15. You can fake it using point clouds:
  16. MPC is also working on this title.
  17. I made these long ago but unfortunately they are not CVEX. http://orbolt.com/asset/animatrix::inlineVexGenerator::1.00 http://orbolt.com/asset/animatrix::inlineVexFilter::1.00 I would also like to see OpenCL COP, multi-threaded cooking/generating thumbnails in the COP network like Nuke. Any new feature added should IMO be mirrored in every other context where it makes sense. It's a lot of work, that's why unifying all contexts would help with that.
  18. You can't dive inside the Wire Solver. It's an HDK operator.
  19. I would get 64GB RAM.
  20. This operator allows you to run an OpenCL kernel as part of your SOP network. Depending on the GPU, some operators can be orders of magnitude faster than even VEX. In this case the OpenCL code is 144 times faster than VEX on GTX 970.
  21. There was discussions about this started by me but the main idea was that there wouldn't be much savings and is also highly unlikely to be done any time soon coz vdbs are not supported in dops (they get converted into dense grids). Flowline does this kind of adaptive approach but i am not a Flowline expert.
  22. Volume convolution on the GPU using OpenCL. For 27M voxels using 100 iterations, OpenCL is 650 times faster than C++ and 12525 times faster than VEX.
  23. This also throw me off when i wanted to save as. SESI should have a utility function in any case IMO.