That just moves the model around on the build platform. This moves the X and Y surfaces of the object in or out slightly, allowing you to fine tune the tolerance on fitted parts or compensate for die swell, material shrinkage, etc. It also differs from simply scaling the object in model settings because it operates on the surfaces and not the whole model. For example, if you shrunk a part by 0.5% to reduce the outer perimeter, any holes in the part would also shrink by 0.5% and the relationship between features would change because the whole part is scaled. This setting will keep the model intact but bring the X and Y surfaces in by 0.5%.
Trying to understand the new feature. Let's say have a simple cube with a hole in the center running top to bottom. Then scale down the x-y planes by 1%. Would this basically shave off only the outside of the sides (not the top/bottom), leaving the hole the same size?
I created a 20mm test cube with a 5mm hole cut through it from top top to bottom. If I increase the "Horizontal size compensation" value, the outer dimensions of the cube get bigger and the hole gets smaller.
For example, if you set the value to 5mm (admittedly not a realistic value), the cube is now 30mm on the X-Y sides, and the hole is completely closed.
Ran some tests on this as well for part fitting. This is a single STL that has both shapes in it. I used exaggerated values to portray settings better, could prove to be very useful:
Original:
scaled NOT using the feature. Notice how everything gets scaled.
using the new feature:
I think this will be especially useful when you can't actually modify the source to compensate for your printers tolerances say for example you download a whole buildplate off of thingiverse.
BaudR8 has the idea. I can't imagine many cases where you would use a positive value. To me, it's all about tweaking dimensional tolerance of parts, especially when you need to create parts that fit together. Most of us are probably used to building this into the model. My process is to design things to exact tolerance (i.e. surface to surface) and then create cutting geometry to cut in the tolerances. This feature doesn't eliminate that step since there are many instances where I need difference tolerance values across a single object, but this setting will let me iterate more quickly to get just the right fit. I used it last night for exactly this purpose and found the fit I needed with fewer steps.
This comes in handy especially if you design parts meant to be molded, and want to print test them, where ABS shrinkage, amongst other things like printer accuracy, can make the part fit differently. Instead of editing the CAD file for two different processes, simply adjust on S3D, and print away.
ashleywebster wrote:BaudR8 has the idea. I can't imagine many cases where you would use a positive value.
A positive value can be very useful if you want to print an assembly (say off of thingiverse) but want to print it larger (say 2x or 3x larger). I've wanted to do exactly this on some models, but then the built in gaps would be too loose and cause a loose part fit. Some simple math and you can print a part 2x larger and use the tolerance adjustment to keep the overall gap the same.