Operators - Collection

[CiroCardoso3v]

I want to use more Operators to control my scenes. Going to the next level, I have hope. For this particular situation, what I have is a massive scene, with trees, grass, forest pack, tons of stuff. Now what I need to do is adding some people, but I don't want to render the entire scene again.

I want to apply a shadow matte to everything except to those 3 or 4 models. In my mind it is easier to select everything and subtract those models and then apply the shadow matte.

My question is what parameter do I need to add to the Collection Operator for that to happen?

[Stephen.Blair]

I would do something like

*.(@node=='polymesh' and not(name== 'model1' or name =='model2'))

I'll do a quick test to verify the syntax...

[CiroCardoso3v]

Thanks @ Stephen for the quick reply. But will this cover, Alembic files? In this case, the Forest Pack, wouldn't be necessary. If make all the geometry linked to a dummy, would that help?

[Stephen.Blair]

Depending on what is in the scene, you might have to use *.(@node=='abc')

If you do a test render, check the list of lights and objects. What do you get from Forest Pack? For example:

00:00:22   702MB         | there are 2 lights and 16 objects:
00:00:22   702MB         |       2 persp_camera
00:00:22   702MB         |       2 skydome_light
00:00:22   702MB         |      14 image
00:00:22   702MB         |       1 utility
00:00:22   702MB         |       1 lambert
00:00:22   702MB         |      11 standard_surface
00:00:22   702MB         |       3 bump2d
00:00:22   702MB         |       1 sky
00:00:22   702MB         |      11 color_correct
00:00:22   702MB         |       3 color_jitter
00:00:22   702MB         |       1 layer_shader
00:00:22   702MB         |       1 user_data_rgba
00:00:22   702MB         |       1 uv_transform
00:00:22   702MB         |       1 driver_exr
00:00:22   702MB         |       1 gaussian_filter
00:00:22   702MB         |       1 closest_filter
00:00:22   702MB         |      12 polymesh
00:00:22   702MB         |       1 list_aggregate
00:00:22   702MB         |       3 bifrost_graph

[CiroCardoso3v]

I haven't forgot this. I will have a look during this weekend.

[CiroCardoso3v]

Ok, for this scene I have 5 forest packs, but I can't find the number 5 on this list, so I am not sure how Forest Pack is interpreted.

00:01:08 68666MB         | there are 36 lights and 156367 objects:
00:01:08 68666MB         |       3 persp_camera
00:01:08 68666MB         |      35 quad_light
00:01:08 68666MB         |       1 skydome_light
00:01:08 68666MB         |     501 image
00:01:08 68666MB         |       1 utility
00:01:08 68666MB         |     358 standard_surface
00:01:08 68666MB         |     712 bump2d
00:01:08 68666MB         |     107 add
00:01:08 68666MB         |     765 complement
00:01:08 68666MB         |       6 divide
00:01:08 68666MB         |    1098 float_to_int
00:01:08 68666MB         |    1057 float_to_rgb
00:01:08 68666MB         |     159 fraction
00:01:08 68666MB         |       8 matrix_multiply_vector
00:01:08 68666MB         |      96 modulo
00:01:08 68666MB         |     504 multiply
00:01:08 68666MB         |      49 normal_map
00:01:08 68666MB         |       6 ramp_float
00:01:08 68666MB         |     918 shuffle
00:01:08 68666MB         |     998 state_float
00:01:08 68666MB         |     554 subtract
00:01:08 68666MB         |      12 switch_rgba
00:01:08 68666MB         |      60 switch_shader
00:01:08 68666MB         |      48 user_data_int
00:01:08 68666MB         |     918 user_data_rgba
00:01:08 68666MB         |       2 user_data_rgb
00:01:08 68666MB         |       8 vector_to_rgb
00:01:08 68666MB         |      27 driver_exr
00:01:08 68666MB         |       1 imager_exposure
00:01:08 68666MB         |       2 gaussian_filter
00:01:08 68666MB         |       1 closest_filter
00:01:08 68666MB         |       2 variance_filter
00:01:08 68666MB         |     237 polymesh
00:01:08 68666MB         |  156129 ginstance
00:01:08 68666MB         |       1 list_aggregate
00:01:08 68666MB         |       1 cryptomatte
00:01:08 68666MB         |      37 Max_Adapter
00:01:08 68666MB         |       1 max_driver

I am going to try and run some tests today.

[madsd]

Forest pack is usually used to distribute a lot of instances.
So in this case, we need to find an entry with a relatively large amount of integers.
We find that the "ginstance" has 156129 ID's.

This is your main entry to forestpack, fetching the ginstance.

[Stephen.Blair]

Yes, as MADs wrote, Forest Pack looks to be generating instances.

There will be one polymesh, and then a bunch of ginstances of that polymesh.

So you may have to just set the shader on one polymesh, which I imagine you can select by name. You could export an ASS file and check the name (maybe just export one "forest pack")

[CiroCardoso3v]

Thanks @ Mads and @

I did try a simple scene with a Forest Pack, Alembic and geometry. The *.(@node=='polymesh') selects everything, event the ForestPack and Alembic file.

However, the

and not (name== 'Box002')

Doesn't work

[CiroCardoso3v]

found the solution

and not (name == '/Box002')

[CiroCardoso3v]

A couple of issues with this setup.

Sometimes the ARV doesn't respect the operator. The image bellow shows that

But if I render it kind of works. Box002 isn't rendered with the material, but at the same time, Forest Pack gets ignored. Forest Pack and the geometry that is been scattered have unique names.

Could this be a bug?

Here is the scene

https://we.tl/t-evhD2ukVfi

If you want to have a go.

[Stephen.Blair]

I'll take a look.

In the second screenshot, what is the expected result? The instances on the plane should be "red" ?

[CiroCardoso3v]

Thanks Stephen. On the second screenshot the result should be seeing the green box only and everything else is red. I have a gif that shows the issue. Give me a second and I will upload it

https://we.tl/t-f8E6cknTwo

[Stephen.Blair]

Hi @Ciro

You'll need to test for ginstance nodes too. They have shaders applied to them too, and that overrides whatever shader is applied to the original polymesh.

*.((@node=='polymesh' or @node=='ginstance') and not (name== '/Box002'))

[CiroCardoso3v]

Ok, good point. But the ARV not updating seems a bug right? I am going to run a couple of more tests.

Another question is related to filter things by layer, which I don't think is possible (is a shame), but if I have a couple of geometries linked to a dummy, let's say dummy_people, then will something like this work

and not (name == '/dummy_people/*')

[Stephen.Blair]

I didn't check the ARV yet...

Layers is not supported yet.

How are the geometries linked?

[CiroCardoso3v]

Hi @ Stephen,

The

@node == 'polymesh' does pick the Forest Pack, but perhaps adding the ginstance can work as a second layer of filtering

The linking that I am doing is something like this

So with the dummy I can control all those elements. It does simplify the process, because all I have to do is using /dum_test/* and everything inside that dummy is affected. I am going to run now the shadow matte material and see if this setup works.

[Stephen.Blair]

Ok, I see. MAXtoA uses the hierarchy of objects for the object names, so yes, you can do something like that.

There is some weirdness with the IPR and operators. I see different sets of objects being touched by the operator.

[CiroCardoso3v]

Just submitted an error report 302868409 and 302868738 because Arnold Operators don't like using the Shadow Matte. ARV interactive or just render completely crashes my scene.

Do you want the scene to test it? Everything works really nice until I use the Shadow Matte. The goal here is for me to mimic the render isolate option that you have in Corona, for example.

[Stephen.Blair]

@Ciro

I have problems with the scene you already sent, so I do have enough to log some bugs.

3ds Max stops responding after I try to apply a Map to Material (so far I tried with Shadow Matte and then Utility).

I haven't crashed yet though, so no CER reports

I have ForestPro Lite btw