Nice! That sounds like a great extension!

Thanks! The addon is so powerful!

Thank you

Thanks Duncan!

Thank you I’m glad to hear that!

@@Erindale Just wanted to add that a lot of your other topics do touch on things I haven't thought of (or from a fresh perspective), although I have done years of work as a graphics programmer. So thank you 😎

Thank you tt! I appreciate it

As long as you don’t have emulate number pad turned on! I have a full size keyboard now but when I learned way back I was just on a laptop so used the number row for camera views. Old habit

@@Erindale, that makes sense ofc.

Since geometry nodes works backwards, it'll collect the context for the position node from the Set Position node that it's (eventually) plugged into. Since we plug the cube geometry into the set position, it'll reference the cube vertex positions.

@@crisishotl1ne Thank you, that explains it

A very useful thing to know when working procedurally!

Thanks so much 😁

I mean that like “here’s one I prepared earlier” but it’s an example of what we’re actually making for the first half of the tutorial

My next tutorial which is being edited now is about doing arrays deformed to splines so you’ll get to see how you can work with instances in that one. It will result in identical instances though so you can’t deform them to the curve, just instance them along

@@Erindale Omg, will it have rotation, start position offset and tilt? Cause this is exactly what I was needing help with rn!! An array with deformation along a curve and one without. :P

Great question! In 99% of cases the field is evaluated on the context of the geometry it connects with. This is slightly different for sample nodes and raycast and geometry proximity. For these nodes you have something evaluated on the target geometry and then some way you want your main geometry (in this case the cube) to retrieve it from the sampled geometry. For example the Geometry Proximity node has a “Sample Position” socket. This is what position you want your main geometry to sample from. By default it’s the position of the geometry but you can specifically set it. Same on the raycast etc. With the Sample Curve node, it can sample a specific attribute or it also passes through position, tangent etc by default. With our use case here we want each point of our cube to “know” where on the curve to sample from. The factor is a predictable 0..1 range so we just need a float value between 0 and 1 on our main geometry to sample that curve. Same idea with the Sample Index node. You use an integer on the main geometry to correspond to specific indices of the sampled geometry. If you’re familiar with coding, these are like a dictionary. We have a key:value pair which is spline factor: position. If we want to retrieve the position from the spline, we need to give each element of our cube a key to go and get the right thing. Hopefully this will become more obvious in the future. There’s been discussions around how to visualise things more clearly. So rule of thumb: the field is evaluated on the context where it meets geometry EXCEPT from the sample nodes where you’re telling them how you want to map data between two different geometries. Does that help?

@@Erindale yes that seems very logical since now that you mention that the goal of sampling nodes is to pass the value or attribs to next target geometry. Thanks🫡!

Of course I found the solution right after I asked. I was applying all this AFTER I converted the curve to mesh, if I do it before it works.

Yeah definitely going to add one of these to ETK that would be very useful

At least since 3.6, not sure when exactly it was implemented

Yes instances are just referencing the same mesh data which is why they need realising to be modified independently

@@Erindale You are an amazing tutor. Thank you for deepening my understanding of this incredible software.

If you invert the normal from the sample curve node, does that solve it?

@@Erindale It did for me. Thank you!

You’ll get there eventually! This is not a basic thing though and it’s using a bunch of geo nodes more abstract concepts of data flow so don’t be hard on yourself not understanding immediately!

You're not alone. Keep going.