How do professionals handle scale for a large world scene?

I have been watching a lot of panel talks from professionals lately. A common theme that keeps coming up is “scale”.

I guess a lot of 3D artist do not have correct scale in there scenes. So lately I have been trying to model everything to life size (the best I can) and then adjust from there.

I am getting ready to work on a big landscape scene with some vehicles and I would like them to look to scale to my rocky landscape. What is the best way to approach this without modeling a 20 mile life scale scene? I will be using ANT landscape.

In my experience a model is provided to represent scale, generally a proxy character, everything is built in relation to it, this model is typically designed with ‘standard’ units from the originating program in mind.

Your specific situation depends on how the landscape is viewed, for example it is often the case that environments have several layers, starting with detailed geometry and eventually moving in the direction of matte landscapes, in such a case perspective is a constant consideration.

1 Like

you can scale down the objects to a scale that is appropriate for your project.
The first thing to think of is the smallest object that you’ll have in your scene and the biggest.
Then find an appropriate scale so these objects don’t get outside the bounds of what blender can display.
ex : if your animation is starting from someone inside a car, then zooming out too a full city.
if the car is getting a 0.0001 dimension, you may have a lot of issues. Especially when rendering.
Sames goes for the city , try to find something that is appropriate.

The most important thing especially on big projects, is to model assets with a good scale between each-other. This is also important if you’re modeling a big asset database. This saves a lot of time and guesswork when these objects are used.
At some point you may need to choose another scale, ex : 1 blender unit = 1meter is kind of ok for character, cars, buildings. But if you need to model a very detailed small insect, then this scale won’t fit well.
you can then use 1bu = 1mm , and if you need a human with the insect , then you can scale the human by 1000.

If you have an animation that goes from very small to the biggest, (a shot that go from a car, to a city, to a country, then to space , then to the solar system) . You can break it down in several shots with seamless crossfade between them.
First the car, it can be scale 1 and the city is very big.
then , the city inside the country, you can have the the objects at scale 0.001 so the city is still visible.
you can parent the camera to an empty and scale it in the first shot by 0.001 to get the same camera move between them.

Another cheat to avoid very small objects in a big space when doing for instance the solar system is to make the sun 10 times nearer the earth and 10 times smaller at the same time.
When viewed near the earth with the distance you can’t really figure out the difference.

So to sum-up, you need a convention so every objects you model look good next to each-other. Especially on big projects. When that is done , it’s up to you to scale them all by the same ratio depending on the need of your project.

Hope that all this help a little bit.


Awesome, that helps a lot thank you.

But another thing to consider, specifically with regard to “real-world settings,” is the effect of f-stop in (virtual …) photographic lenses: the “foreshortening” effect of a zoom lens vs. normal vs. wide-angle. Beyond a certain functional distance, an object is simply “in the distance,” and you can’t really tell how far away it is, and it really doesn’t matter. The object is never going to be interacted-with by any of the actors in the scene. It will be judged only against the distant objects that are “in the distance” with it, because there really are no visual cues that can link it to the foreground action-scene.

You’re actually free to indulge in classic photographic tricks such as forced perspective, as spaghetti-western moviemakers routinely did.

In any case, the “scale” of any object is always relative to the scene in which it appears. So, you might well model an object to different scales as well as different levels of detail. Nevertheless, always consider the eventual impact of editing, where you begin to assemble the shots into a finished movie. Here is where, most of all, the shots have to look as though they really belong together. The physical size and placement of the object needs to be comparable between shots that are to follow one another.

Scale is most important in ordinary close-up scenes, where a “two-foot” footstool needs to look right in a “twelve-by-fifteen room with nine foot ceilings,” when a “five-and-a-half foot tall” actor rests her tired feet upon it. These are probably settings in which the default lens is “normal,” with close-up and wide-angle being used for impact and not for any sort of foreshortening effect. (Although it certainly can be used for dramatic reasons, e.g. when two actors are shouting at each other across a room, and you use a long lens to bring them much closer together than they actually are.)