Just noting that this has always been the case afaict [at least in 2.79 this was also a limitation...]

@Philipp Oeser (lichtwerk) Thanks for confirming. I guess it would be nice if this limitation is considered a bug at this point given that it really undermines the power of metaballs and the collections used together.

I am also guessing that this must be the same reason why metaballs can't be used with particle systems as object instances.

This was reported before as T71242, which was incorrectly merged into T69753.

@Sybren A. Stüvel (sybren) In https://developer.blender.org/D7478#179533 you said,

The patch could be extended to support other instancing methods as well, but these are relevant to T69753. Once this fix is accepted by the reviewers, we can look at extending.

Is this one of the instancing methods you were thinking about extending?

I am also guessing that this must be the same reason why metaballs can't be used with particle systems as object instances.

@kursad k (kursadk) I am not sure what you mean by this.

@Evan Wilson (EAW) It does not when metaballs are instanced from another scene. I.e. create two scenes one with collection of metaballs, other with a particle system and attempt to render metaballs from the particle system. It does not work on my machine ( Blender 2.92.0 )

In T76759#932685, @kursad k (kursadk) wrote:

@Philipp Oeser (lichtwerk) Thanks for confirming. I guess it would be nice if this limitation is considered a bug at this point

Having a limitation is not considered a bug, even when that limtation has been in Blender for a long time. It's just a fact of life that it's not possible to implement everything at once.

In T76759#932698, @Evan Wilson (EAW) wrote:

Is this one of the instancing methods you were thinking about extending?

I wasn't thinking about any specific case, just that in general I wanted to fix only what was reported/considered as broken.

@Sybren A. Stüvel (sybren)

Is this the same instancing issue why the metaballs cant be used inside a geometry nodes setup via the object infor or the collection info nodes?

Metaballs have their own implementation, and are quite a special case (there is all sorts of special handling of metaballs in Blender's code). Even though in the end they are converted to meshes in order to be rendered, they are very different from meshes in nature, and thus cannot be produced by a system (like geometry nodes) that outputs meshes.

We're currently facing this issue again. We have created a small library of Meta Objects and need a way to handle these, collections come to mind, to make sure that changes in the original also propagate to the rest of the new derivates.

Metaballs and impicit surfaces are a very valid tool for visual designers to create organic and soft shapes.

Beeing not able to instance these basic geometric (and parametric) shapes is a real turndown. (Especially with rounded caps for curve objects nowhere near the horizon?)

I understand that Metaballs and meta objects use really old code and are no priority for highend game asset or movie production, but Blender being for anyone these objects deserve some love as an amazing level of entry for visual thinkers.

In T76759#1177017, @Sybren A. Stüvel (sybren) wrote:

Metaballs have their own implementation, and are quite a special case (there is all sorts of special handling of metaballs in Blender's code). Even though in the end they are converted to meshes in order to be rendered, they are very different from meshes in nature, and thus cannot be produced by a system (like geometry nodes) that outputs meshes.

And of course we'd only need some form of static mesh representation of the meta object in a collection.