Extreme-Scale Project Workarounds?

I love SketchUp, but the things I’m interested in making with it run into problems on the program. I want to make designs that include a vast range of sizes, from geological scales to mere inches, though I have other projects in mind that vary in scale even greater. However, when I create a project that includes objects of all these sizes, problems arise on the small end. Eventually, when you zoom in enough, small geometry becomes warped due to calculation limitations and the camera cannot get close enough to the small-scale structures, merely cutting them out of view at a certain point and below. All in all, it’s not the ideal way to handle this kind of project.

What I need is a workaround of some kind, some way to properly edit the design on any scale with none of the aforementioned problems. It’s also important that the different scaled parts are not totally separated - the piece as a whole needs to still fit together. I’m using SketchUp as an artist to create large-scale scenery to later draw in 2D, so really the only finished product I need in the end is a screenshot that respects the orientation of all objects in the scene. Any ideas?

One idea I’ve had is to model the whole scene in separate parts, each one slice of the scene based on distance from the viewer, then to render them each and put them together afterward using photoshop or whatever.

The problem with that is I don’t know how I would line up my view on each model so that the end result is seamless.

I’m still trying to understand the goal clearly, can you post an example? If you are creating a 2D screenshot that includes the “geological” scale elements, aren’t the inch objects way to small to see? How big is your target canvas? You would need one the size of a continent to paint at 1:1 scale just to make the 1inch object 1inch across. I think I must be missing something here.

How about using the component “save as” & “reload” features to work on parts of similar scale.

Example:

You can isolate a part by making it a component. After making it a component…

  1. context click on the part component and select “save as”. This will save out the part as a separate skp file
  2. open that part in a separate session of SketchUp and edit it. DO NOT MOVE the axis
  3. when done editing the part. Save it.
  4. Back in the original file, go into the Outliner and find the part component. Context-click on the part and select “Reload” and select the edited Part file.
    Result: Your changes to the part will be brought into the “master” sketchup model.

CD

1 Like

This is a great workaround for editing individual parts. But won’t solve the inevitable clipping in the compiled file which wants to incorporate such a wide scale spread. But I’m still not sure how the OP is actually using the file, it sounds impossible to see all the elements at all scales in a single 2D image?

I’ll be making a full scale environment and getting images from within the environment at regular human scale, so I’d need to see all objects, near to far. It has to be full scale because i need to show the environment from many different widespread locations within it.
And it won’t be quite continent sized, though a solution to this problem should support me in doing that if i decide to.

Ahhh, I begin to understand. Well, Clipping is going to be your nemesis here. I find that the informal usable range can include objects of up to about a mile across, if modeling is kept very simple. At that range you can still see and work on an object about 10" across before clipping. This is all subject to lots of variables, but it’s a place to start.

I think this all depends on how close you need to get to the larger objects. If the really large objects can stay a bit farther away from the camera so that they are not clipping on their own then you might be able to compile images using separate Sketchup files and photo editing. You could, for example, create a file with all the smaller stuff (keep in mind that distance from origin counts as model size so even smaller objects spread out over a mile will begin to clip.) then once you have your small objects in the foreground positioned the way you like you could make a scene saving that camera position. Then “save as” a second file that contains the same scene and camera position, erase the small stuff and build your giant object in that file so that it is correctly visible from that same scene camera position. Then you would need to export a 2D image from each file (.PNG supports transparent backgrounds) and overlap the two images in photo editing software. Since you are only using these images for reference you could include guide axis in the mutual scene to help you overlap the images. This would let you see small things in the foreground simultaneously with larger objects behind.

However, this method cannot overcome the basic problem of clipping of larger objects at close range. As you can see in the .gif, even without the smaller stuff, a mile wide piece will begin to clip on it’s own at close zoom. If you need to see a very close up view of any part of something bigger than a mile or 1.5 km, Sketchup will be a frustrating tool to use.

A potential option might be a separate rendering program. Here are is an image snapped exported from within Sketchup, and the same perspective sent to a render engine (Visualizer in this case). You can see that the renderer is happy to show the geometry that Sketchup can’t. By leaving the realtime render window open on a second screen, I was even able to orbit and navigate the Sketchup model sort of by looking at the render window, but it’s very cumbersome.


RND-0398

As endlessfix said, at about a mile across clipping will make modelling smaller detail difficult.
Another treshold is somewhere larger, tens of miles perhaps, when OpenGL display artifacts start to appear (distorted objects, “exploding” model, triangulation that looks like z-fighting, etc. (Archicad and Revit are recommending models not to exceed around 30 x 30 kilometers - Archicad uses OpenGL like SketchUp, Revit uses Direct3D).

You can make each component its own file, then arrange them in your extreme-scale file. Model components in their own file, then click reload to update your components.

For example, I want to make my 3D Oak Tree low poly, but there’s object clipping.


I put my Oak tree component into a container component, so you can reload them all at once instead of reloading them individually.
43 PM

I open the 3D oak tree file, then make it low poly.


I reload the oak tree component inside the container component, then they all reload with one command.


The same technique works with editing buildings, street signs, or whatever.

Make all components as low poly as possible, while still getting the desired level of detail. Make sure texture images are as low resolution as possible, while still getting the desired level of detail. This prevents lag.

Put Items on correct layers to help with selection.
Put items into correct groups to help with organization. Use the outliner frequently.
Use grouping to keep all alike items in a group.

Once you’re finished with your extreme-scale model make a copy & divide it into sections small enough that there’s no clipping if you want… for visualizations or whatever. Although, I’ve found that you can get some decent images at eye level without clipping 2 mi from the axes.

A bigger model means more work, but you can hopefully do it if you invest enough time and work into it. :wink:

This topic was automatically closed 91 days after the last reply. New replies are no longer allowed.