Jump to content

Thomas Helzle

Members
  • Content count

    154
  • Joined

  • Last visited

  • Days Won

    15

Thomas Helzle last won the day on October 2

Thomas Helzle had the most liked content!

Community Reputation

168 Excellent

4 Followers

About Thomas Helzle

  • Rank
    Initiate

Contact Methods

  • Website URL
    http://www.screendream.de

Personal Information

  • Name
    Thomas Helzle
  • Location
    Berlin
  1. Toms "Learning Houdini" Gallery

    "Cubicles" Same structure as above but rendered semi-opaque in Redshift: And a detail view: Cheers, Tom
  2. Toms "Learning Houdini" Gallery

    Then I thought I should take it to the third dimension and came up with a funny solution: This is about 12 divisions from a cube. My idea was, that I would pack the initial cube, check the boundingbox in primitive mode in my loop and if it contained at least one point, "subdivide" it by scaling it down by 0.5, unpack it and place one still packed downscaled cube on each point of the resulting unpacked cube with copy to points. If the cube contains no points, it would not be in the "do" group so I would not scale or unpack it and place only one packed cube at the centroid point - redundant but works compiled where a merge does not. I guess there are better algorithms out there, but I enjoyed the compact simplicity of this one. Compiled it's still rather fast, although no longer realtime as the 2D one. Cheers, Tom
  3. Toms "Learning Houdini" Gallery

    "X" Another one in the series with 16 levels: Again rendered with Redshift using rather strong DOF for the depth blur. Cheers, Tom
  4. Toms "Learning Houdini" Gallery

    Not being convinced by the approach Entagma took for their quad-tree tutorial, I changed it from using a radial distance search to a min-max evaluation in each direction, so I'd catch all the points and compiled it. While working on this I did some test images: All rendered with Redshift. Cheers, Tom
  5. Controlling Focus Distance

    Cool - have to try it with Redshift as soon as the current job is done! :-) So far I always used this in the focus distance field: // Camera focus distance to a null called focus: vlength(vtorigin(".","../focus")) But it's a bit clumsy to navigate the null. Tom
  6. Controlling Focus Distance

    Thanks symek - have to try that. I guess that will only work with Mantra, right? Cheers, Tom
  7. Controlling Focus Distance

    What I really miss is a way to click into the rendering to set the focus distance. This is very nice for quickly finding the sweet spot for stills. Love it in Thea Render etc. Is there a way to do this as a viewport action in Houdini with some raytracing from the mouse? Cheers, Tom
  8. Toms "Learning Houdini" Gallery

    "Hex Grid" An experiment with very finely detailed hexagonal geometry for deformation. All geometry, no fat ;-) Cheers, Tom
  9. Toms "Learning Houdini" Gallery

    "The Shell of Light" An exploration of extreme DOF in Redshift. The basis is a wire-shell, but rendered with massive DOF. This creates the most fascinating and somehow otherworldly structures. Art Prints here: https://fineartamerica.com/profiles/thomas-helzle.html Cheers, Tom
  10. I ended up using this (with "@class" being the per-object attribute from the connectivity node): float searchrad = chf("Searchradius"); int maxpts = chi("MaxPoints"); int nearpnts[] = nearpoints(0, @P, searchrad, maxpts); int maxlines = chi('MaxLines'); int cnt = 0; foreach(int pnt; nearpnts){ if(@class != pointattrib(0, 'class', pnt, 1)) { if (cnt >= maxlines) { break; } int line = addprim(0, "polyline"); addvertex(0, line, @ptnum); addvertex(0, line, pnt); cnt += 1; } } While this works fine, I still wonder if it wouldn't be faster to have a fitting group expression directly in nearpoints... ? Cheers, Tom
  11. I wondered the same thing recently and wrestled with the "group" parameter in nearpoints, but somehow did not grok how to write a group expression that does this. Using pcopen feels much clearer there, although one will probably still have to have a large maxpoints number to be sure to hit something on the other curves and then maybe limit the result manually in the while loop. I was searching for only one closest point on any curve different from the current, using the "class" attribute from the connectivity node... Cheers! Tom
  12. Is AMD potentially risky? (Threadripper)

    Interesting thread, thanks everybody! Reading about those 18 core TRs or new intel cpus made my mouth water initially, but what I read here doesn't exactly make me want to jump in right away. My 6 core i7 3930 from 2013 overclocked to 4.1 Ghz (costing about 500 Euro back then) with 64 GB GDR 3 RAM @1600 takes 30 minutes for the Boolean Sphere scene - not too shabby. So while you guys get almost twice the speed, that still isn't exactly as earth shattering as I would have expected. Cinebench Single Core is 139 here, so basically nothing much has changed in that department. So it's only really worthwhile for CPU rendering (I only score 1034 in Cinebench with 6 cores / 12 threads) but I use the GPU for that anyway. I think I'll wait some more with getting a new machine and let things settle down a bit regarding drivers, prices and bioses... :-) Cheers, Tom
  13. Toms "Learning Houdini" Gallery

    "Hexa Grid" Two color-variations of an experiment where I started out with a hex-grid and went from there... As always: watch in full resolution for all the detail :-) Cheers, Tom
  14. Toms "Learning Houdini" Gallery

    "Twisted Growth" DOF in Redshift is expensive but beautiful! :-) Cheers, Tom
  15. Server downtime

    Thank you very much for all you do! Odforce is the best :-) Cheers, Tom
×