It’s possible to create clusters in Grasshopper 2.0?
Thanks,
Roy
Nope, not yet. I started on the concept with a somewhat simpler type called Chains, but clusters are not available yet.
Hi David,
ok, thanks
I have high hopes for Clusters or other methods for making Grasshopper 2 definitions more compact.
For years, it was not possible to rely on Clusters 100%. I think at least some of the issues with them have been fixed, but Clusters still have their problems from a UX perspective IMO. I would be very happy if they could be avoided in the second installment of Grasshopper. Perhaps one might look separately at the situation where a cluster appears only once in a file and more complicated situations, multiple clusters, or worse yet, one cluster nested within another. Maybe the solution to making definitions more compact should be one thing, and creating repeatable modules should be another thing. I wonder what your vision is regarding the “Chains” you mentioned.
I wish some of Cluster’s problems would stay in the past.
One problem is the feeling of separation from the rest of Grasshopper’s definition. When editing Cluster I would still like to be able to recompute the main definition or connect and disconnect new inputs without having to save and exit the Cluster edit window. Editing the Cluster could be done in a separate window (visible at the same time) or the main window could be divided into two. It would be great if creating clusters was convenient as a method of developing definitions at every stage, and not as it is now (at least that’s my impression) as a method of refactoring a working and proven script.
Even medium-length, structured, grouped, and tightly “packed” Grasshopper definitions sometimes have too many components to be able to comfortably look at them from a distance and analyze how the entire “strategy” of the definition works.
Another thing is that it would be fantastic if Clusters offered full support for components from the Rhino.Inside.Revit range and potentially other add-ons that may require specific communication with applications.
I agree mostly with your opinion, however I have this perception of the subject “Clusters” in GH1:
- 95% of users dislikes, hates or thinks they can do the same without them
- 75% of users uses them (and is quite happy that they exist)
Creating a kind of incoherence.
Personally my work is heavily based on clusters, not because of compaction or just “grouping”, but to reuse routine blocks which can be used multiple times within the same definition and adapt to be used in many definitions without breaking or requiring internal changes.
In those cases where a definition is made of a sequence of action blocks, represented as clusters, things will complicate a lot if you decide to get rid of Clusters
This is why I ask for them to be still used in GH2, with their issues solved and rethinking their design, or their name, but please @DavidRutten keep them so we can define encapsulated snippets which can be instanced, exported as user objects, referenced from a file, in other words used as user defined functions.
Thanks
Yes I am part of that 95%.
I hate clusters. I think they make the definition turn into a black box which I don’t want. It also is lazy cable management.
Also I find them quite impractical, when it comes to connections, sometimes you need a value that is calculated inside the cluster and thus you need to add an extra output… very time consuming and break my workflow.
At the same time the whole interface I never liked, with a the menu message always asking if you want to save or discard changes and recalculating the whole definition…
I deal with really big GH definition and I don’t need clusters at all.
Also, Hops does most of this.
One important aspect of clusters shouldn’t be overlooked: In GH1, they allow users without traditional programming skills to create, export, and share their own components as User Objects.
If clusters were removed without an equivalent replacement, wouldn’t this make developing and sharing custom Grasshopper components significantly harder for users who rely on visual programming?
I don’t think GH2 necessarily needs the existing cluster implementation, but this functionality should definitely be preserved.
That is what I wanted to express. I was rethinking about my post yesterday, and I came to the conclusion that the future could be better reformulated as the best of both worlds:
- “Chains” as anticipated by @DavidRutten : group sequence of actions, but without using plain grouping which still clutter the workspace. Imagine them as foldable sections in a text code or worksheet. Keep the space tidy and refactored when possible into distinct processes but without leaving it. Lose the functionality of instancing the same process to be used multiple places when data tree input is not sufficient
- “Clusters” or “User defined functions”: Primairly to allow users to build their own plug-ins without recurring to the higher level of complexity of compiling traditional code, deal with SDK, etc. Build functions but still with visual programming (possibly, but not exclusively, including Python/C++ scripts) which are accessed via files external to the active definition. Perhaps have them compiled as current binary .gha but without the and interaction limitation imposed by the current solution of Script Editor (Context parameters etc..). Let users build, version and distribute their tools in this way would be a breaktrough
@ShynnSup I understand your point of view, but probably you don’t understand mine. Your alternative solutions are valid, but in my case I tried all of them an none was fit as Clusters. Hops (and Script Editor compiling) work with Context parameters only, which is a limitation in possibilities and usability. It is great for complex, intensive tasks but becomes more of a bottleneck in case my functions are small fragments of operations which are to be used multiple times and in many cases the overhead time added for using Hops surpasses the time consumed by a simple cluster. For instance, many clusters I use have a plot/display use, taking some inputs, elaborating and constructing a proper visualization of data
Recalculation for each access/exit from the cluster is a pain, I agree.
Adding extra outputs.. I do it everyday and takes some seconds to do. Lazy cable management, is something that I didn’t understand. I used to rely on cable management plugins such as WiresRenderer (no longer functioning) and WireHopper, they solve the majority of practical problems with wires.
McNeel can correct me if I am wrong but I think this is exactly the use case for what Hops was created. @stevebaer, @AndyPayne , @scottd, @Japhy
Hops is not for clustering huge definitions but rather small snippets. I think it is quite efficient when it comes to this.
What I meant with this is that most clusters I have seen from people, they don’t care about the cable management inside of them as the cluster hides spaghetti mess and outputs clean outputs… but of course it depends from person to person.
Clusters are already available in GH2, with some nice additions from GH1 (native loops!).
To add my two cents to the discussion, I think clusters are a great way not only to reuse parts of definitions, but also to organize them.
Interestingly, I’ve found myself using them more and more lately, after years of dismissing them. I find they really bring clarity to definitions. They help thinkkng in terms of functions, logic hierarchy, variables, inputs and outputs, etc.

