Jump to content


  • Content count

  • Donations

    0.00 CAD 
  • Joined

  • Last visited

  • Days Won


Everything posted by Atom

  1. FLIP Divergence

    That is a nice tip, unlocking the solver and changing the drop down to Set Always. Most of the time when I work with FLIP I also change the Grid Scale to 1.0 on the FLIP Object and then head over to the Reseeding tab of the FLIP Solver. I typically add one more particle per voxel to the Particle Per Voxel expression. ceil(pow(ch("../flipfluidobject/gridscale"), 3))+1 Then I increase the surface over sampling to somewhere like 6-18. this gif shows an over sampling of 9. Over sampling helps smooth things out, so you can use a higher particle separation, in some cases. ap_animated_FLIP_Divergence_v0100.hiplc
  2. Converting Hard Constraints into Glue Constraints Help

    You do have a misspelling of the attribute, @strength.
  3. Quads instead of triangles

    The simplest trick is to drop down a VDBFromPolygons then drop down a ConvertVDB (defaults to volume, change to polygons) to turn the volume back into polygons. You will get all quads, but not necessarily flowing loop cuts.
  4. UV out of range

    Make sure you use SPACE-5 in the viewport to visualize the layout of your UVs. Mari NC only supports one UDIM tile so you have to fit all your UVs within that first box.
  5. Apply force on certain rbd objects

    The force to a group may be working, but the members of that group might still be bound by constraints to members that are not in the group. If you have constraints to the windows, that could be it. Try turning the windows into a static collision object, instead...?
  6. Central gravity force

    It may seem odd, but you can reference the location of a moving NULL, and tie it to the goal vector of a pop steer seek, using a relative reference. Then all the points in the RBD sim, that are in range of the field, will be attracted to it. The force is activated by animation, which causes the RBD chunks to attract to the green NULL. ap_particles_attract_to_point.hiplc
  7. Houdini 18 Wishlist

    Once again, I request talking lip synch characters. Here is the way I want it to work. I supply a wave file that is used to activated the mouth. I supply a text file that contains the words that the wave file says. I press a button and I get a talking head that I can render. The head should be shapeable, using controls, similar to make human or other character generating tools. I think that's it, two files, one button.
  8. Dynamically add wires into wire solver

    You're on the right track with time shift. Shift to the end of the timeline, then reveal the wires, over time, by deleting the ones who's @id does not yet exist.
  9. Vellum Aerodynamic Lift

    Check out the cloth+pyro.hipnc, at this link. It demonstrates how to achieve lift in cloth (not vellum) by making it responsive to the temperature field of a pyro simulation.
  10. keep uvs on flip anim

    You need to unpack Alembic before you begin processing. Here is a revised setup using a .abc file. There were some errors in your file. There are some UV anomalies, but if you increase the FLIP resolution (which is matched to the pointsfromvolume resolution) the anomalies become less obvious. mb_african_femal_01.abc ap_flip_UV_transfer_001.hiplc
  11. rotate flip pump source

    It is common practice to not animate DOP sources at the /obj level. I moved your spinning animation into the SOP level and cleared the /obj level animation. Also, you don't need that rest. It seems to be locking your vel source to frame #1. AP_01 pump-rotate-test 03.hipnc
  12. Emit random object in RBD Packed object

    Check out Matt's setups. http://www.tokeru.com/cgwiki/index.php?title=HoudiniDops#Emit_packed_prims_into_RBD_sim
  13. You could use the actual NASA data.
  14. I was playing around with an RBD domino setup and extended it to accept a graphics based image (i.e. logo) for input. The Trace node converts the image to the outline of the logo, which is connected to a CopyToPoints, to setup the dominoes. ap_logo_dominoes_071819.hiplc
  15. Mesh from Lines (leaf)

    Triangulate2D node can get you started, then you can play around with remeshing or poly reduce. ap_Leaf_v0102.hiplc
  16. I think the importer might need to support that feature. It looks like Houdini and Blender do not. They convert the instanced geometry to actual geometry. What is interesting, is if you load the FBX file using the File node, you get the associated @fbx_location, @fbx_scale, @fbx_rotation of each instance as a point attribute. This means you could locate the named instance items and reconstruct them. Or repack them to get their centers and redeploy. ap_restablish_FBX_instance.hiplc
  17. Crowd agent transition issue

    After playing around with the scene, the problem seems to be substeps. Your collider is moving very fast so it is actually passing through the agents, sometimes. The crowdsolver defaults to 5 min and 5 max substeps. Set the crowdsolver min max substeps to 1,1. Jump one level,and set the crowdsim dop network substeps to 5. This will cause all nodes, inside the dop network, to run five times, for every frame. A little slower to sim, but this seems to solve the RBD transition problem. ap_crowd_ragdoll_Collision_RBD_071519.hiplc
  18. Crowd agent transition issue

    In your Collision RBD file try setting the Transform Group of the duration CrowdTrigger to 'all'. That may solve your ragdoll transition problem.
  19. Crowd agent transition issue

    Read the help card on CrowdTrigger, Playback Time is in seconds, not frame. So you have 5 seconds, which is frame #120 (5*24fps).
  20. I generally add pscale at the end of the loop. As long as pscale exists on the line, you'll have your problem. ap_carve_q.hipnc
  21. That is correct, the CopyToPoint is interpolating that random pscale assigned at the end of each line. When you copy to the line, you inherit that interpolated pscale value. You can use an attributedelete, placed before the CopyToPoints, to remove pscale and the spheres won't change size anymore.
  22. Converting integer to string atoi

    You have to parse the string, of course. Fortunately, your piece name all have the same length prefix, so you can get away with an array manipulation, using a constant. i@nameInt= atoi(s@name[5:]); https://www.sidefx.com/docs/houdini/vex/strings.html
  23. How to connect these pieces by procedural ?

    Check out this tire thread. Or use the search features of this forum. Here is your modeled tread section switched into the Tyre creator. Tyre_Tread_1.03_lianxi.hipnc
  24. Smoke keeps expanding

    There are a lot of little problems in your scene (excessive twiddling of parameters to non-default values). Your best bet is to throw away the smoke and debris setups and recreate them with the shelf tools. Once I did this, the basic smoke and debris effect started working as expected. Your smoke isn't fading because you have changed the control field from temperature to velocity. Because you use excessive velocity, on the breaking sphere, the velocity seems to be carrying through into the velocity field, which is too strong to reduce the dissipation. This display also highlights that there are some problems with the internal boards, which kick out a bunch of high velocity particles half way through the simulation. You might want to just delete any of the debris particles that are generated inside the box of the main tower, providing only the external particles to the smoke sim. ap_Tower_19_03_maybe_05.zip
  25. Switch from 16.5 to 17.5 breaks my python code.

    I set line # 213 to mushheight = int(np.floor(mushheight0*len(vertices))+1) And line # 136 to a constant. hairEnd=0.5 #[[zs[np.minimum(np.random.choice(np.arange(hairStart[i]+3*skip//2,hairStart[i]+2*skip)),len(zs)-1)]] for i in np.arange(len(iis))] This caused your code to work on H17.5, for me. My guess is that some of the np (numpy) functions are returning floats instead of ints, now...?