Jump to content

symek

Members+
  • Content count

    1,877
  • Donations

    100.00 CAD 
  • Joined

  • Last visited

  • Days Won

    66

symek last won the day on October 12 2017

symek had the most liked content!

Community Reputation

298 Excellent

4 Followers

About symek

Contact Methods

  • Skype
    szymon.kapeniak

Personal Information

  • Name
    Szymon
  • Location
    Waw/Pol

Recent Profile Visitors

22,307 profile views
  1. instancepoint() is undefined - ?

    Hi and welcome on odforce planet! There are a couple of gotchas here, which is easy since lots of Houdini's functionality evolves with time and/or overlap with each other: 1) pointinstance() is a hscript expression function, not VEX one, so you can't use it in an attribute wrangler. 2) You can't also use it in any parameter in SOP level, as stipulated in docs. It is meant to be used on /obj level to vary instance parameters (mostly xfrom. 3) It's an oldish method dealing with instancing (described here), good for things like planting trees or houses with limited variations and in moderate number. If you want to explore instancing using geometry (points) attributes you might try to play with packet primitives. Here's link with general explanation and here are details about using point attributes to transform packet prims. 4) Strictly speaking the attributes described in last link should work also with the Instance Object (instances will be oriented according to point attributes), but it's less interactive than using CopytoPointSOP and packet prims. Technically instances generated by InstanceOBJ are evaluated in soho prior to render phase, and support for them in the viewport is limited. Bottom line is: packet prims == more fun. skk. ps setting transformation attribute in VEX can be done "manually" as described in last link or more in common with CopytoPointSOP with new instance() method.
  2. Git Integration with Houdini

    This is great idea. For simplicity custom scripts and commands in a menus, correct? Python git support lib would suffice then. Although more tempting is another approach. Houdini has a notion of file system handlers, which allow you to get file with http:// prefix for example. Git would naturally fit here. Last time I tried to do something similar I hit the wall, but theoretically one should be able to do it. Help link. Not sure if this is what you're after, because: how fun it would be to have git://mySuperHda.hda@d2cd1a7ca20ece9fc47be3b1949982ebfc398cfd or maybe rather git://mySuperHda.hda@2 [previous commit]
  3. Those tiffs have to of particular type not supported by Houdini, which is common (even for photoshop) because there are so many tiffs variants. AFAIK your problem has nothing to do with icp nor iconvert. Both use native Houdini image libs. Can COPs open these files? What are they? MRI scans, DEM map?
  4. Setting COPS Image resolution

    show us a screenshot
  5. What are the advantages of implementing it as Houdini plugin? Is it really necessary? My thinking is that Houdini won't play, in general, simulation (even cached) in realtime[1]. Wouldn't be more elegant to create a play list for the sequencer as a post simulation process in Houdini and pass it to the sequencer via file? There are probably a dozen of different ways of accomplish this. The easiest one doesn't even require C++ nor Python. Just select objects visible in camera's frustrum at time x and save that list to a cvs file with corresponding timing. In case that's not an option, CHOPs are nice place to pass sequence of events to different places. Making custom OBJ level object alike. SOPs can do that also with nice advantage of taking result of 'pruning' procedure as input. You would have to provide more information though. [1] - Houdini isn't realtime application and it has full civil right to snap a few miliseconds for an unknown reason from time to time... (its cousin Touch Designer is realtime application though and can read almebics - a hint!).
  6. Setting COPS Image resolution

    I can set maximum resolution to 36000 pixels and rescale image in COP for that resolution here on 17.0.376 (macos). Works fine.
  7. Lens Distortion in COP

    I literally copy/pasted code from ldpk (classic model - described in section 2.4.3 of their pdf), so naturally credits goes to science-d-visions guys. VOP distorts C and A planes, but also exports distortion to uv plane which can be used in DeformCOP (see example). Untested, I might did something stupid. hth, skk. ldpk_classic_3de_mixed_distortion.hipnc
  8. Using hou.RopNode.render()?

    HOM, like Python, is strongly object oriented, thus most methods live and operate on objects. hou.RopNode.render() is a prototype of a class method, and similarly to most cases, you can't call it directly. It's meant to be used as follows: ropnode = hou.node("/out/mantra1") ropnode.render() #btw: print(type(ropnode)) <class 'hou.RopNode'> render() is the same method you saw in hou.RopNode, but it's now bounded to the instance of a node of type hou.RopNode ps There are some methods free standing in hou module which don't belong to any objects, but that's different story.
  9. Random Scale of Spheres

    @pscale = fit01(random(@ptnum), 100.0, 300.0);
  10. setprimattrib not working properly

    pcfind returns the indices of points, not primitives.
  11. bad allocation error

    Most probably Mantra can't allocate enough RAM. How much memory has your system? How many CPU cores? What OS?
  12. $JOB/cache/Houdini_Caches/$ACTIVETAKE/${ACTIVETAKE}.bgeo.sc
  13. Random link of interest

    Physically Based Rendering - the book - is free and online as of today: http://www.pbr-book.org ...and while talking about knowledge, another good place to gather some of it: Filament docs - very comprehensive knowledge base about PBR (theoretically for realtime engine, practically doesn't matter): https://google.github.io/filament/Filament.md.html
  14. Centos 7 CUDA

    Both OpenCL and Cuda are part of Nvidia diver installation, so you don't have to install anything else. Also Cuda and OpenCL are unrelated in terms of the software.
  15. C++ Effects Programming ?

    There is always room for improvements in FX department, although SESI builtin tools raise bar high enough. Nevertheless since product vendor has to provide tools for a wide audience, particular implementation of the tool might not fit one's needs. Typical scenario are FEM for character effects (secondary motion / tissue effects), techniques for cloth and hair effects, like cloth sim driven by rig animation, low scale / high detail SPH fluids , particles' meshing techniques (capable of handing slow motion), illumination caching for volumetric effects (still very slow for high res dense volumes), cg human all together - mocap / video driven face animation + photoreal texturing. Advanced destruction with good blend of material based approaches and artist friendly directibility. Ropes' physics! Add to the blend remeshing of dense geometry (with texturing!), render denoising, example based seamless textures in 2 and 3d, photogrammetry tools, volume upsampling, and you have your weekends busy for the rest of your life! cheers! ps It's not necessary to post the same post in two forum's sections!
×