How does SketchUp handle a scene with thousands of objects in it?

Every few years the idea of using an existing modeling package for a game editor pops up at the places I work. I personally enjoy SketchUp more than Max or Maya, but I believe all three of them would ultimately collapse under the weight of handling thousands of objects in a scene.

I know metadata (attributes) can be associated to unique components, so that takes care of baseline game needs. Plus the component architecture is really nice at mimicking game prefabs. I'd love a reason to spend more time inside of SketchUp if I think it can deal with a variety of primitive objects that represent game logic objects and the intertwined game data they hold.

OR, on the flipside -- has anyone had success in mimicking the simplicity of SketchUp model creation inside of Max or Maya (as a plugin)?

I'm totally screeching in to the last page without reading previous ones. Having said that, has anyone looked at: http://playuptools.blogspot.com/?

I don't have a video of it, but it's fairly simple.

It takes whatever object(s) you have selected and moves them to the nearest "grid" setting. The grid settings are found in: Model Info: "Units". To turn on the grid also check "Enable Length Snapping" in Model Info.

The action uses the origin point (local axis) of the object as the snap-to-location.

This is kinda like an idea I had a couple months ago (but not savvy enough with Ruby yet to try out) --> create an generic_menu.rb which passes in menu variables. Add this (require generic_menu.rb) to any existing plugin (yours or other peoples), then comment out the original plugin's menu. Now call the other script and send over the plugins menu commands and how you'd like to group it (submenu or even a separator).

For example, 5 different plugins from all places on the web. Re-route them where ever you like:

plugins
... Frequent
...... FredoScale
...... Mover2
... Utility
...... Weld
...... Cleanup
... SketchyPhysics2

You'd have to tweak existing plugins, but it'd only use one external script to do the routing. It could be expanded to also influence context menu options or even toolbar or cursor options.

Has anyone found a way to have certain models not affected by inference-snapping?

I'm finding a couple situations where I wish the inference-snapping would ignore a model in the scene. For example, I have a large sub-divided plane and I'm trying to extrude a cube in the foreground, the motion keeps snapping at sub-division way off in the background.

@thomthom said:

@alz said:

• include the sandbox "Flip Edge" in your toolset

Why not just use the Sandbox function? I'd like to keep the toolset as small as possible to void cluttering it. And this is an edge function rather than a vertex - and it exist already.

Yeah, that's true. I was looking for a way to not have the whole toolbar eating up space for that one (more frequently used) function.

Thanks for the reply and thoughts!
(and effort of course!)

Wow -- very cool thomthom!

Here's a few ideas (which might've been mentioned already):

• include the sandbox "Flip Edge" in your toolset

• have a (variable) tapered radius that pulls or pushes vertex perpendicular to the camera (what yer facing) or axis-aligned (ie. only along the X-axis). Now you can create hills and valleys. (this would also need a "strength" value for the distance of each execution). *** I think this is what you mean by soft-selection, ya?

• Then have a vertex XYZ min/max limit:

It's like you're slowly pushing crazy piece of clay into a glass wall.

For example, you have a 100, 100, 100 room with a subdivided 30,30,30 cube in the middle. Use jitter or a tapered radius to push n' pull parts of each side so it looks like a crazy star fish. Now set the min limit to z=10 and the max limit to z=90. Enlarge your tapered radius to cover a whole side. Pull the warped top side vertex. It will slowly pull upwards. When the peaks of it reach z:90, they stop moving upwards. Continued pulling will cause neighboring vertex to also stop at z:90. Stop somewhere before all vertex are pulled to z:90 and you'll have something looking like Monumental Valley -- a series of flat-top plateaus. They same could be done with the bottom side of the cube. But go all the way -- push them down until they are all flattened at z:10 to form a flat surface once again.

• Now have a way to quickly capture a XYZ limit based on whatever vertex is under the cursor (or selected) -- like an "eyedropper" tool for position info.

• How about an "average" tool -- this takes the affected vertex and brings them closer along 1-axis to reduce "spikes". Use it too much and it will flatten all vertex at the average height.

That's right! I remember that plugin now (but forgot what it was called).

Thanks TIG!

That's what I just figured out. I'm working with some people that are use to having a visible grid, so I created a grid object (group). I've locked it to prevent it from "opening", but it'd be best if it wasn't even selectable (but allowed snapping to).

(I wish SketchUp had a grid option like Banzai or other editors -- I know people frown on that here, but it's very handy for certain developments and quickly knowing what the current unit snap is).

Hi TIG!

Yeah, I use the VCB all the time. I was looking for something that was local to the shape. It'd be a variant of the "Dimensions" tool. It'd show the dimensions while being created, but go away after you unselected it. It'd be nice if when you reselected it, the dimensions showed up again, but not critical.