3D Editor Plans

I’ve spent some time looking for how an editor could work to move from ideas and speculation to facts and logic. The answer is meatballs. (Metaballs, 3d objects that connect to each other smoothly) In this post I will explain the core idea of creating a mesh. Rigging, weights and uvs will be discussed another time.

In this editing tehnique everything is made out of metaballs. If possible, there will be more shapes included like sharp ending claw mesh and a long part useful for bones, limbs and more sticking out parts. Implementing objects other than spheres would require different calculations but is very beneficial. There is also an option to be able to extend a metaball but it’s way more complicated than having two separate shapes. Location of every placed metaball will be saved and used in the editor, during gamplay they should be converted into a mesh.
Here is some helpful information about metaballs.


Every creature will consist of different tissues. The default tissue is the one which is the most common, in the simplest and most efficient scenario it would symbolize muscle, fat, skin, wood, whatever organism’s style is so the overall shape and internal organs would be inside of it. Different tissue types will not stick to each other, they will still be connected but they will interact on a vertical level only with it’s own tissue type.
Every piece of tissue is a metaball with a center point and a ray distance with the value same as it’s center to surface distance. This would let other metaballs move only in the ray distance to not create any mess. Scaling metaballs will also be restricted to how big the connected parent metaball is but it will still leave a big size variety.


The first metaball will be a parent to any metaballs placed on it, that means children metaballs glued to it and will change position right with it. The amount of child pieces glued to one ball should be restricted too to maybe 4-6. This rule includes only children and not pieces connected to children themselves.
Player will gradually grow and shape his creature to become more complex looking. Placement options for objects would be similar to what is implemented in the microbe editor. One metaball will always be in the center of X and Y dimensions and the player can choose if the creature will have 2 sided, 4 sided symmetry etc. (though more complicated symmetries will be tough to program). For example on a 2 sided creature one sphere would copy the position value of the other sphere but with X set to negative. Non spherical meta objects will require some degree of rotation control so they would copy rotation with Y and Z mirrored. Their origin point would always be on the Edge and facing towards the parent object for the best rotation results.


A very helpful building tool are negative metaballs which can carve holes in the tissue for mouth or anything the player wants it to be.


If the main tissue has a hidden group of negative metaballs it will allow placing tissue over it
without clipping to create shells, armor, claws.

-Summary of how it all works

  • Each object is a child of a metaball from which it was duplicated from. That creates branches which lead to the “origin metaball” making it a main control point (moves entire organism up, down, forward, back in the editor).

  • Parents should have a child limit of around 4, mirrored objects count as 1.

  • When a parent is moved, every child copies the position shift (XYZ).

  • If the parent is a non spherical object, children have to rotate with it using parent’s origin.

  • Every object has a limited scale relative to it’s parent and an absolute size limit to not fill the entire scene with blobs of tissue or tiny invisible points.

  • Non spherical objects can only rotate on X Y and not further than 90° from it’s original position. This rule doesn’t apply to internal organs.

  • When non spherical objects are created on a parent their original position is always facing away from parent’s center with it’s origin point close to parent’s surface.

  • Every object except the “origin metaball” has It’s own X Y Z value set that can’t go outside the distance limit set by it’s parent’s surface.

  • “Origin metaball” is always X=0 Y=0 but can be moved on Z (up down).

  • Probably only one side of symmetrical creatures would be editable, selecting the uneditable part will send you to an editable one. Metaballs on 2-sided organisms (like vertebrates) will copy a location from the editable part but change X to -X. Non spherical objects do the same thing but also copy rotation changing YZ to -Y -Z.

  • Z<0 value is floor and nothing can have a world space value equal to it.

  • There is a negative meta surface on -Z which flattens every metaball making it interact with the ground.

  • Negative metaballs work like regular metaballs but can be moved further from the parent.

  • Negative metaballs have a tissue type assigned but maybe the player will be able to selectwhich tissue group it affects and which not.

  • Main tissue has hidden negative metaballs that only work on other surface tissues.

  • Each tissue has a metaball group which only merges with itself but can still be parented to other tissue spheres and has to follow their parent location limits.

  • Internal organ tissues and bones can only be placed inside external tissues so that they wouln’t have be rendered.

  • Internal tissues can be placed anywhere but are still parented by the “origin metaball” and are not allowed to go outside the external surface.

  • All the data about size, location and rotation of objects is saved and converted into a mesh during the playthrough.

For the end, here is a process of creating a creature from scratch (only used spheres).

+ Meatball Disturbance and some other prototypes


I like this idea. It’s good that it’s something that’s been made before, meaning it is feasible to make it.

The other idea, sculpting, has a bit more freedom in making different things, so some people might want that to have the best game possible. Though maybe the different metaball shapes, or allowing sculpting the surfaces of the balls, could fill that gap. Metaballs definitely look easier to use to make something, that looks like not-completely terrible, than a freeform sculpting or direct vertex based editing.

Out of two, I had more fun using metaballs than sculpting. I get that sculpting sounds more free but it’s extremely complicated to pull off well. Metaballs provide a good control over verts unlike sculpting which can mess up a lot about the topology. Even if that’s solved by programming a dynamic topology and remeshing it’s very uncomfortable to sculpt a creature. Some shapes are just very hard to create with that method so I’d stick fully to meatballs. It is possible to create good shapes from them while keeping verts well distributed (not perfect but works well with animation). One thing that would help is smoothing out a mesh just like in blender to get rid of weird face shapes.

That’s really cool, it’s clear how much work you’ve put into it and all the stuff looks great!

One thing I really like about it is how, on a data level, each creature is organised like a tree. There is a root metaball and then children of that ball and then children of those balls etc all the way to the external limbs.

I think we might be able to do some sort of tree modification algorithm in order to have smoothly varying creatures. Maybe if each metaball is inheriting properties from it’s parent then just by going in and changing one value that change will propagate downstream in an interesting way.

For example if you make the root metaball a bit bigger that scales the whole creature up, however if you take the root of the tail (which is maybe a child of the overall root) and then make that a little bigger then the whole tail might get scaled up a bit while the rest of the creature stays the same.

I also like that it’s a simple place to start we can build on. As you say hh we could add non-spheres and complex sculpting tools to it later if people wanted more control. I think there is a lot to be said for making it easy to make something reasonable and hard to make something bad, that’s worth a lot.


One thought I had about this. Would it be possible to use the metaball tree structure as a way of rigging and animating the creatures?

For example consider a worm which is made of a metaball head and then a tail where each ball is a child of the ball infront of it. To animate it would it be sufficient to alter the angles at which the balls are attached to their parents? That could make a wriggling motion, that’s how these work, a pulse passes down the creature from the front to the back.

Even doing IK for having a hoof on the ground the problem could be stated as determinind the angles the metaballs would need to be connected at to the creature connects it’s body position with the hoof position on the ground.

Maybe it would look odd and the creatures might end up quite squishy, however one improvement might be to set some connections as very stiff and others as very bendy.

1 Like

The first metaball would probably be created from the first biggest colony of cells and we can’t tell if it will evolve into being a head. My speculation is that most of the time it will be somewhere in the middle. If we include a copy paste feature you would be able to make a segmented creature which could move using a chain and the first segment would become a the head.
With animating, the amount of metaballs in creatures will be big so I would rather convert them into a mesh because animating them in metaball state would require big amounts of calculations from every single sphere in every visible organism. Converting them to mesh would allow regular animation but rigging could use child data. Limbs without bones would be wobbly but if you put a hard bonelike structure inside the limbs an armature would be created from these bones making the movement arranged and stiff.

1 Like

Some information about uv mapping: Seams will have to be marked vertically on every symmetry division and possibly three times horizontally. Every bump would have a seam marked from the nearest division to it’s furthest edge + some more edges to unwarp better.

Symmetry center would have to be placed in the middle of a texture space (on vertical axis for 2 sided creatures, there is no center point for them). So the texture would be mirrored allowing symmetrical and asymmetrical coloration.

Different tissues could have different looping normal/ specular textures to make them look more smooth, dry, soft etc. Early multicel organisms would be translucent, similar to how cells look and possibly have bright dots on them just like on microorganisms we can observe in real life.