This is probably a limitation right now, but would open up a lot of possibilities.
Since the weight proximity did not work, my plan was to use the Boolean Node and oversized spheres to create islands where I want smaller spheres to be placed.
In the video example of miro channel I shared above, the modifiers didnāt had an effect on the mesh itself, only on the distributed points.
They were still instances
So indeed thereās two solution for modifiers evaluation of instances : effect on real mesh (assuming there would be a non destructive way to make the instance merged into a single mesh & effect on the distributed points
Thanks @lone_noel that is an interesting workaround, but would get convoluted very quickly if you had to do it many times. I will have a play around with that approach, as I have been mostly avoiding nested groups.
I do love a nodal workflow for operators and seeing the general flow of a procedural setup. However when it comes to expressions, nodes do drive me nuts if I have to string a ton of them together. Generally I would much rather a single expression node with written code. I guess it is similar to vex vs vops in H. I generally prefer vex, but vops has its uses, especially when using procedural noise.
I am sure we will see some expression or python nodes in geometry nodes eventually, as it is still early days. I like the idea of being able to put an expression into any of those attribute fields, in the same way that you can in other parameter fields in blender (to auto setup a driver). You already can put a driver expression into the value input node, but I donāt think that works in any other node?
Would be interesting to be able to pipe a written expression in a node into the attribute field sockets. This way would be cleaner, as then the fields themselves would only accept a single attribute (especially if they ever get dropdown menus), but more complex things could be piped into them with an expression node.
I think with instanced objects the problem might be that they need to be converted into their own data block instead of referencing data of the source object. There might be a need for a node to explicitly convert instanced copies into real objects for things like Boolean Node, or have the Boolean Node do it automatically.
Proposing a new distribution method based on a prototype I have been working on. Point child. Primarily to allow you to create hierarchical scattering for outdoors environments.https://developer.blender.org/T84816
@Kenzie why not just using a proximity mask ? We do that In the scatter addon. Might be easier and less complex/ more flexible as once the mask is created you can use it with any particles user would like
The only thing needed is to have the proximity modifier work with distributed points
Definitely a consideration, the one problem Iāve had with the thought is the geometry would have to be tesselated enough to support the gradient, which might be a no go with lower poly stuff. Would make it easier though if users were okay with having to divide the model a bunch. Yeah, probably its the way to go.
Shouldnāt it be possible to have geometry nodes to calculate a denser mesh just to know where to place the particles, and then keep the original object topology?
Isnāt the more general solution to this problem a SDF? What if I want to, eventually, scatter some floating objects around each of those trees? A distance field that allows me to scatter within it might be more useful. It could also be useful to use when wanting to change some attribute of a particle based on how close it is to another 3d mesh etc.
Hi, regarding the concept of the active modifier, I see that there are still many areas where clicking on it doesnāt make it active:
-Clicking on submenus titles, checkboxes or modifierās name
-Changing float values
-Clicking on the handle for drag-drop
These are all instances where the modifier should but currently doesnāt become active.
I still think that the drifting from the hover functionalities for hotkeys resulted in a major loss in speed, so if weāre going to have those depending on the active state, at least the aforementioned limitations should be tackled before the release of 2.92, in order to make the active state more responsive and fast to use, now it feels almost broken to use hotkeys with modifiers.
Hi!
As of my understanding and tests, you cannot treat scattered geometry as a real one, so thereās nothing you could do with that on top of Geometry Nodes modifier. However, you could scatter one geometry on top of another, and then use this for the next geometry to be scattered on top of.
See! Modifiers worked on the generated points in old builds!
Because the generated points were vertices thus compatible with modifiers, when we did the switch to a new point type, all modifiers support got lost.
As you can see, using a weld modifier on generated point cloud was extremely efficient and useful at making the point distance themselves from each others.
Out of curiosity, does it work if interleave the modifiers? So that the points are created in a Geometry Nodes modifier, then a Weld modifier, then another Geometry Nodes modifier to instance the objects
Hi, guys. So, Iām trying to instance a cylinder on another cylinder, but for some reason the instances are incredible squashed. Why?
Also, Iām trying to use a remesh modifier to āmergeā all the instances on the original cylinder but it doesnāt work. It only affects the original cylinder. What would be the ideal workflow?