davpe

what exactly do you need to know? just make selection by @name directly in Material SOP. Or make groups from the @name and then use those as a separator, if that's more comfortable for you.

again, depends on your specific setup. But for example, "transform by attribute SOP", "transform pieces SOP", and "getpackedtransform" and "setpackedtransform" VEX functions are the good candidates to look at. also this article might come in handy: https://vfxbrain.wordpress.com/2020/07/09/how-to-re-apply-packed-transform/

i think there is no out of the box solution for this as it is a very specific request. however, if you have packed prims, it should have it's transform matrix stored as intrinsic primitive (pos, rot, scale). you can then use a python script i suppose, to read individual packed transforms and write it into a text file with json syntax. If packed prim transform is not there in packed intrinsics (i.e. it's zero), it's been either discarded somehow in the prior processes, or haven't been generated at all. In that case you have to find a workflow that generates and retains this attribute for you. (not 100% sure if this is generated by "copy and transform SOP", but it's definitely generated by "copy to points SOP") this is of course just a rough workflow on how would I try to approach it. Detailed steps will vary depending on how you've built your scene, and how u wanna use the data next. cheers, D.

depends on renderer - for instance Arnold and Redshift have something like a "fetch VOP" that allows you to reference another VOP from a different shader network. Not sure if Renderman has something alike but it seems like that's exactly what you are looking for.

shame this functionality doesn't come as a node in Houdini out of the box. it would be really handy sometimes.

hm that still is an odd behavior. why would missing lights make cryptomatte not work?

that's right, Arnold doesn't support heaps of stuff in Houdini. with packed geo, it attempts to unpack it before rendering, which cancels out any benefits offered by it. On top of it, agents are kind of a special stuff tailored to be efficient with Mantra / Stylesheets. Stylesheets are neither fully supported by Arnold, if I'm not mistaken. In general, I can't recommend using Arnold in Houdini - it's a lot of pain all the time. We render crowds using Mantra as that still seems to be the best option with least amount of headache. In the future, we count on rendering crowds with Karma. cheers. D.

but anyways, in wrangle you can always do: v@uv.y = @P.y; this will literally take position Y coordinate and replaces the uv Y coordinate by it. if thats what u want.

well in that case that wouldn't be about simply locking it, as the building may be 100 units, but uv space operates in 0-1 range. so it depends a) do u want to keep it proportional? b.) do you want to fit it in the 0-1 range, or u dont mind if it goes across uv tiles? c) are u dealing with just a single asset, or u need to keep consistent scale across multiple buildings? d) are u modeling it in houdini too, or it comes from somewhere else?having things modeled in houdini would make things much easier as you can prepare things in advance in the modeling stage.

aha i see. check the example file if that's what u want. if you needed a completely procedural approach, it would be slightly more complicated to setup, but definitely possible too. right now i just created a group "extrudeSide" on polyextrude2 SOP, and then UV_transform SOP to offset the selection. UV_Y_reply.hip

ok is the scatter layer by itself noisy when u look at it alone? or does it only happen after you put them together?

hey, i'm not sitting in front of houdini right now, but you should be able to do that easily. after you use the uv layout SOP, use vertex wrangle or vertex VOP to copy the original v@uv.y from the node before the uv layout, keeping v@uv.x unaffected.

that should be super easy man - just use Labs UV Transfer SOP. plug Mesh 1 to first input, and Mesh 0 to second input. it will transfer uvs from Mesh 0 to Mesh 1. there are pretty much no settings, and as far as I have been using it, it always worked as a charm. The only prerequisite is that the shapes of the objects must be "the same" (or very similar). But the topology can be whatever.

aren't you using denoiser or something? i think that does not process AOV's (at least not by default).

hm, the way you describing it doesn't sound like a familiar behavior to me. if i select something - like a polygon - it remains selected until i do another selection. if i want to deselect, i just click the empty space while in selection mode (S). There's nothing more to it as far as I'm aware. selecting different nodes in network view is not altering the actual geometry selection. also, if you just beginning with Houdini, what I think may be confusing is concept of having a representation of "Displayed geometry", "Current geometry" and "Template geometry". This is not affecting your selection but it affects how objects are displayed in viewport and may be confusing to somebody who is not aware. hope that helps.