How to approach modeling game asset you have no measurements for ?

Creepezel

As far as I heard game assets should be close to their real world scale but what if, the object I would like to make a model of, doesn't have any measurements in internet left ? Like pre-war tank prototypes or vintage furniture.

Ghogiel

It doesn't have to be completely exact in those cases, you're not needing precise CAD data or anything so you can mill out metal components irl. The tank can be.. tank sized. You should be able to extrapolate something approxmate surely by looking at similar objects.

Furniture can be made in proportion to a say a human character standing next to it on sitting on it etc. 

 Stuff can be very slightly fudged at times anyway. Like you might want ceilings or door ways to be a bit higher than an average real life one because of how the game character might have an (unrealistic in many cases) jump ability and hiting the ceiling every jump while indoors breaks the flow and is not very good game design. Or maybe it's fudged a bit because things have been decided to snap at a certain distances and so on.

Creepezel

Oh, that makes sense. Thanks for your response !

Klunk

does it matter ? uniform scale of an unanimated object is usually the least of your worries.

Alex_J

i just seconding what already been said, but yeah definitely test it with your other game items for scale. Doesnt matter if you got the same scale as real life if your main character looks wrong sized compared to it.

And dont get too obsessed with arbitrary accuracy that you forget to look with your artist eye to make sure you have pleasing proportions.

poopipe

It matters enough that the tiny bit of effort required to work in real world units is far outweighed by the negatives of not doing it.

Things that are almost always based on real world units:
    Physics/Cloth/Hair etc sims 
    Anything based around volumetrics (like atmospheric lighting)
    Anything distance based (like lighting)
    Things made by other people
    Real things you might want to replicate

Things that are not based on real world units:
    20 year old video games
    A lot of the shit you used to find on turbosquid

FrankPolygon

Establishing the real world scale of production and prototype military hardware can be fairly easy with a little research.

Using tanks as an example: most designs were based around existing equipment. So at a minimum, the caliber of the main armament is almost always known. Add to this all of the standardized off the shelf equipment that's added on and each identifiable item becomes another data point that helps increase the overall accuracy of the estimated rough dimensions.

Often things like tools, headlights, Jerry cans, radios, externally stowed equipment and supplemental armaments have known sizes and are visible in reference images or prototype drawings. Finding, comparing and correlating the known dimensions of these objects across multiple reference images and drawings should provide enough data to establish a reasonably accurate approximation of the real world scale.

Here's a couple of links to a detailed project breakdown that mentions this process and shows how much work went into creating the Leclerc MBT models for Armored Warfare:
https://www.artstation.com/artwork/EVGwen
https://www.artstation.com/artwork/nQX164

To add a bit to the above post in support of working in real world units: it's worth considering that Fusion 360 is becoming more popular for some hard surface workflows and 3D scanning is also becoming more accessible. If both of these trends continue then working in real world units and learning about parametric modeling will likely be beneficial.

Working to precise real world units won't always be a project requirement and it's certainly possible to create good art without using any hard measurements by just eyeballing the shape proportions in the reference images. This approach can work well for projects with unique art styles that don't have a fixed scale or projects with limited budgets where the goal is to hit minimum viable to stretch the resources and and there's absolutely nothing wrong with that.

It's just worth noting that one of the current trends places a greater emphasis on the accuracy of the model and the impact this can have on the overall visual fidelity. How complex this issue needs to be will depend entirely on the goals of each project. Though, in general, even something as simple as doing a little research and getting the overall dimensions of the block out and base mesh reasonably accurate to the (sub) millimeter can go a long way towards improving the quality of the end product and it can also help prevent costly re-work caused by inaccuracies.

Which approach will work best ultimately comes down to the project's art style and resource constraints but it's not overly difficult to work in real world units or build geometry that has a consistent scale and is fairly accurate. Once you're used to working with real world units, tolerances and tools like nearest whole unit snapping it doesn't add that much time overhead to the modeling process and it can actually save time by making it easier to get things right on the first try.