3dome

i can't look at the file right now but please stay away from using Concave mode as far as possible. it's slow, unreliable and just not good. there are far better options like vdbtospheres, heightfields, etc

just merge them with the created lines?

using 16.5.323 dm_animPrims_wrangle.hipnc

there's no need to create groups with a partition SOP as the connectivity gives you an attribute (called class by default) identifying which prim belongs to which piece. so just run the loop as I mentioned in the first post (you can even do For-Each Connected Piece loop which sets up the loop to work with connectivity) EDIT: just make sure to set connectivity to Primitive so the attrib gets created on them instead of the default point level (the Connected Piece loop does all that for you if you're unfamiliar with that stuff)

they usually are. you can see that by middle mouse click it and then it says packed alembics and you only have like 25 points/prims or so EDIT: but there's nothing wrong with converting them to polys and then run connectivity before the loop

if every piece is a packed alembic you can loop through them. use a meta import to get random seed for a simple expression (like fit01(rand(detail("../metaThing", "iteration", 0)), -125, 235) ) in a transform SOP (on there maybe set pivot to $CEX,$CEY,$CEZ) if they are not packed, just use connectivity as in your picture and then loop

in case you wanna be more fancy and have the prims appear in a growing fashion, drop down a fuse SOP in unique mode before that code //primwrangle vector center = @P; int primpts[] = primpoints(0, @primnum); foreach(int i; primpts) { vector dir = center - point(0, “P”, i); float fit = fit(@primnum+1+@Frame, @primnum+1+(@Frame*(@primnum > @Frame)), (@primnum+1)*2.05, 1, 0); float ramp = chramp(“scale_ramp”, fit); vector set = dir*ramp; setpointattrib(0, “P”, i, set, “add”); } play with the ramp (i found moving right point closer to left side looks nice) and with the 2.05 in the code (make it parameter or so, jsut make sure its always > 2) to adjust "trail length" and falloff also pretty cool if you reverse the fit to (x,y, 0, 1) and have the ramp zero all the way then close to right side make it go to 1. gives a nice trail when disintegrating

group them based on color and then use that group in fuse/facet node? (might need to be converted to point group)

i guess such a point always has 4 neighbours, so EDIT: it's a bit diffuse how much control you have about the other points (are they grouped, etc)? so my code above might not be what you want. instead you could UV each quad and select the corner points based on their UVs ({0,0}, {0,1},{1,0},{1,1})

maybe like this: //primwrangle if(@primnum > @Frame) removeprim(0, @primnum, 1);

no need to set input to "Myself". leave it to none and just write f@myage += 0.1; EDIT: just built it like you show and it works aswell. are you sure you have points in the active group?

another way: //primwrangle int pt = addpoint(0, @P); //center for each primitive removeprim(0, @primnum, 1); //we just want the new created points connect adjacent pieces with type set to adj. points. set radius to correct value to not get too many connections. unit icosahedron would be 0.3 for example //pointwrangle int nb[] = neighbours(0, @ptnum); //this array is the neighours for each point. point numbers correspond to prim numbers

I rarely use the instance node so I can't really help, sorry. But I can offer one more way of instancing - via primintrinsics - but you might already know that dm_intrinsic_instancing.hipnc

i might be mistaken but I think instancefile needs an absolute path

although clustering might work, I would suggest splitting the shot into 2 simulations. 1) pyro coming out of engines 2) big pyroclastic area "on" the ground this way your containers are efficient and you have way more control. you still could use clustering for the 1) simualtion part though i still might have a look at the file in the evening if i have time