PDA

View Full Version : Dimensional Precision...



Calvn_Swing
2007-03-07, 07:32 PM
At AU2006 it was mentioned to me that Revit had a higher dimensional precision than AutoCAD, thus making it more effective for exporting to fabrication devices...

I'm trying to confirm/deny this...

I have noticed that in AutoCAD most people don't take the time to be precise on the building scale because in CAD you can "fake" dimensions unlike Revit. As a result, when we take CAD plans and convert them to Revit we find it more cost effective to model in Revit from scratch and simply have the CAD drawings plotted out next to us as a reference. When we try importing/linking backgrounds we usually find so many dimensional inconsistencies that we spend several days finding and correcting these issues.

Also, since Revit is an object model rather than a representational line drawing, it makes sense that it would be more effective as an export device to fabrication tools (3D Printer).

The first issue is valid, but would rarely affect fabrication of models or assemblies. The second point can have an obvious positive affect, but it isn't really about the accuracy of either software.

Anyone know the answer?

Thanks

aaronrumple
2007-03-07, 10:23 PM
Revit can only display down to 1/256". I don't know if internally it is storing a higher level of precision. Since the dimensioning and temp. dimension go down to 1/256" it renders the use of any less mute.

Revit may use more of a Bently type system fro units. AutoCAD uses a floating point system for units. The further away from 0,0 you get - the less dimensional precision you can store. Bently uses a design cube and everything in that cube has the same level number of decimal places. This is all based on the number of digits you can store in a dword (I think).

Calvn_Swing
2007-03-08, 12:05 AM
Interesting...

I know revit goes to a higher precision... We had an odd issue in a project where a particular dimension (with measuring tape) would switch from a whole number to that number minus 1/256" depending on how you measured it.

This confused the heck out of me for about two days until I tried moving one of the objects 1/512" Magically, no more weird measurements. So, it must be more precise than 256ths...

Thanks for the reply!

phyllisr
2007-03-08, 12:17 AM
...it was mentioned to me that Revit had a higher dimensional precision than AutoCAD...
Aaron's answer was shorter but this may help you understand how this affects your work.

Generally, I would be careful accepting a statement like this at face value. I think it depends on how you define "precision" and what accuracy means to you. I would assert that the statement is patently false. My observations tell me that Revit and AutoCAD behave a bit differently with odd, teeny dimensions but that is not the same as precision.

For example, in AutoCAD and ADT, any entity or object practically exists in the "infinite" environment of space. Axioms and theorems and proofs from geometry in eighth grade back in the dark ages; between any two points, another point can be placed and there are an infinite number of points on any given line segment. If you list a line in AutoCAD and see that the length is 8'-0", it is possible that it is 8'-0.0000000004". You will not "see" this in any properties since the display in AutoCAD extends only to 8 decimal places. Just because you cannot see this does not mean it does not exist. This wreaks havoc if the "hidden" problem is in the Z axis. Try filleting one line or AEC wall that is at 0" elevation with another that is 0.0000000000000003".

In Revit, you may get messages when bringing content from AutoCAD (importing and exploding, not linking) that tell you lines are slightly off axis or that a line is too small on the screen. This does not mean the accuracy or precision in Revit is any less than AutoCAD, it just performs differently. If you draw a box with Revit walls 100'-0" x 100'-0" and divide it into eight rooms 25'-0" x 50'-0", all of the rooms will be exactly the same size @ 1,212.75 SF. If I move the center wall 1/2048", there will be no change in the plan view and no change in the dimension. Even if your project units are set to 1/256" and the dimensions reflect the project units, you cannot "see" the difference. However, you can verify that Revit will calculate this level of precision by creating a schedule and setting the units for the Area to a custom decimal rounding.

Attached is an image demonstrating this concept. From my point of view, you can be sloppy or disciplined, precise or careless and accurate or inexact in any program. The platform makes no difference. Our sloppy drafters who drove everyone crazy in AutoCAD can be just as sloppy and drive everyone equally as crazy in Revit or ADT or SketchUp or any other application.

In the real world, those teeny inaccuracies do not matter much. In the end, a .001 SF discrepancy or a wall that is 8'-0.0000000000001" instead of 8'-0" is meaningless. But they do matter when it compromises a model and things do not align correctly or function as you might expect.

Hope this helps.

Mike Sealander
2007-03-08, 01:07 AM
Phyllis:
Excellent!
I have noticed that the way Revit ties input to the display can cause problems. For instance, I have had trouble getting sections to be perfectly perpendicular to non-orthogonal grids and reference planes. Another area is with sloped objects using a slope arrow. Let's say you have a sloped roof, made by extrusion.When you go to build a ceiling parallel to the roof, you can query the roof for its slope in degrees, but the reported degrees will only be as accurate as Revit's reporting precision.
I'd like to see Revit offer more opportunities to snap graphically than I can currently get it to do. Sections are one example. Another example would be detail linework on top of non-normal model elements.

douggodfrey
2007-03-08, 10:23 AM
Actually, revit can display down to .000000000001 if you set your units to decimal inches|custom. As Aaron eluded, not practical in the real world however i have had situations where the rounding feature even at 1/256" was causing overall vs. intermediate dimension errors where this precision level helped.

Calvn_Swing
2007-03-08, 05:35 PM
Thanks everyone! I think I got my answer more or less..