Leaderboard

Yes it looks different but that depends on the mesh you're using as a base for the solver of and the noise a lot Glad it worked out for you with the sweep though.

2 years later and this just made my day. I modified it to work on named prims and added a naming property to the subnet so I can drop it into a named folder. Thank you!

font_fill_particle_attract_vel.mp4 pop_font_fill_particle_attract_vel.hip thant should give you an idea !

Bumping this old thread in case anyone else finds themselves in a similar boat where they're drawing a blank on this technique and end up googling "Houdini noise transition". The trick is to first form your gradient using a value, in this case the x position along a grid... ...and then add noise to that value. For a deeper dive into this kind of thing, definitely check out this series of classes by Main Road Post and this brilliant page by Kiryha. Hope this helps.

and I made it work inheritVel_Test_v02.hip

I don't know why it doesnt work but you could just write the intrinsic to a string/int attribute before and then use the uniquevals function after. That's probably still more convenient and faster than using python. Also typename is a string, so declaring your foo array as an int would, if the function would work with intrinsics, result in an array with a bunch of -1s I think.

noise_shift_001.hipnc

Wow, ok... so I just learned something new again I apologize to @cwhite I just need to be properly educated Thanks again @julian johnson

Alternatively you can extract the hit normal and do a dot check against the the initial point normals (if dot is larger than 0, it is a backface) and if needed, you could start a new ray at this point to check for subsequent front/backfaces