Announcement

Collapse
No announcement yet.

Native Imperial Vs Metric Accuracies

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Native Imperial Vs Metric Accuracies

    I've always questioned if Revit's native imperial nature meant we metric users might be getting stiffed by conversions/abstractions.

    ...and I'm starting to see more curious instances that concern me - and questioning the reliability of Autodesk's assertions that Revit is in anyway Computer Aided Drafting...

    Example?

    2017-2.PNG

    12.png

    Okay 12 decimal points of a mm is an acceptably nonsensical accuracy to use in our industry
    (perfect for measuring suits for sub-atomic particles)

    But why the indiscretion at all? I typed 5m on the copy-move of that second grid. I should get 5m flat. End of. And to have the dimensions return varying results along these "parallel" grid lines? What can we trust?


    I imagine, aside from the obligatory "let it go" assurances (that it's not the end of the world) someone might wade in with the limitations of 64bit Floating Point precision (?) or some other technological 'reason' - which I'm keen to accept - but if such is true, why the option to punch up to 12dec.pts at all? Why expose the limitation and get me, and potentially others, wound up scratching our heads?

    GARGH!?!

    :banghead:

    #2
    It gets better. Slide a dimension along the parallel lines. It will change, even after the dim is set. It's always had some wobliness at super high accuracy. Even in imperial.

    Sent from my LON-L29 using Tapatalk
    Aaron "selfish AND petulant" Maller |P A R A L L A X T E A M | Practice Technology Implementation
    @Web | @Twitter | @LinkedIn | @Email

    Comment


      #3
      Feeling your pain snowy, my OCD and me recently battled the exact same X.000000000001 thing with levels not equaling exactly as they should. It became obvious when using Dynamo to copy stuff to other levels. Even then some of my levels wouldn't move away from 11 decimal place inaccuracy like x.00000000059.
      The only thing I can relate /12 to is a bakers dozen. Which is actually 13. Lol.
      There are no stupid questions, only stupid people

      Comment


        #4
        Originally posted by Twiceroadsfool View Post
        It gets better. Slide a dimension along the parallel lines. It will change, even after the dim is set. It's always had some wobliness at super high accuracy. Even in imperial.
        Yeah been seeing that also - I imagine you've exhausted all avenues of investigation? Would love to hear Autodesk's 'excuse' about this one.

        I did the same test in 2015 thinking I had a messed up 2017.2 install - but no:

        2015.PNG

        I wouldn't mind so much if it wasn't for the fact that this 'discovery' came about when critiquing, and trying to rationalise, some fairly ropey CAD work. Suffice to say I ain't beating the Revit drum today.

        Comment


          #5
          Investigation? Over what?

          The only projects (EVER) that ive seen it in, have some bullshit brought in from AutoCAD as the root cause. Since i dont endorse bringing in trash from AutoCAD to be traced, i dont bother "investigating" or asking for "excuses" because they "dont matter" (even if that isnt an acceptable answer to some).

          Also, for what its worth, the 13th decimal point in decimal FEET, is a value that cant make a significant impact (even in radial dimensional control), unless a project is several miles long. The 13th decimal point in decimal mm, is a joke. Not even worth looking in to, or thinking about.

          Sure its *annoying to see it,* but your time would be better spent just not dragging CAD trash into Revit, and avoiding the problem alltogether.
          Aaron "selfish AND petulant" Maller |P A R A L L A X T E A M | Practice Technology Implementation
          @Web | @Twitter | @LinkedIn | @Email

          Comment


            #6
            Originally posted by Twiceroadsfool View Post
            your time would be better spent just not dragging CAD trash into Revit, and avoiding the problem alltogether.
            Well yes, quite, but these were from no-template test models with no CAD whatsoever - so unless our installer shipped with a pre-mangled "no-template" (unlikely) the "inaccuracy" is clearly evident from the get go.

            Complicated by CAD? Sure?
            Caused by CAD? Wouldn't seem so.

            Yes, working with nanometres (in the Architectural space) is stupendously pointless, certainly not my goal and not what I want to be looking into... BUT (like it or not) the fact remains we have been given CAD to work with (on this job) so I'm exhausting means to cover myself, my team and my firm.



            BTW: By 'investigation' I meant "I imagine you might have seen this and raised it with Autodesk".

            Comment


              #7
              Sure, ive raised it with them. I sent them a great video of it thats on youtube somewhere. (EDIT: Here it is: https://www.youtube.com/watch?v=saWFnhlZ0ik)

              And i think they came back with an explanation of sorts, but it wasnt something that was going to be "fixed," so it is what it is.

              For what its worth, here is what my testing just found (quick testing, since i dont think its worth messing with, honestly):

              If i make a grid in decently close proximity to the origin, and copy it and type in "5m" for the distance to copy, the first one dimensions as 5000.000000000000 if the dim is set to millimeters. If i do a second copy (from the second grid), it copies to 5000.000000000000, but moving the dimension will sometimes make it vary by .000000000005. Literally, either 5 nanometers or 0.5 nanometers. I lost count of zeroes.

              Now, if i create a grid a stupid far distance from the origin (107,000 FEET, or 20.26 miles), then drawing the first grid and copying it "5m" reconciles it to 4999.999999997894. And it isnt fixable (by moving the grid farther out and typing in a higher accuracy 5000.0000000000000 value).

              Worth mentioning, the "High Test" dimensions i keep in my file are Decimal FEET, set to 12 or 13 (whatever max is) decimal points. In terms of "covering a team and a firm" i dont know a single contractor in the WORLD who will even work to 10 decimal points of decimal FEET, let alone decimal mm. Currently all of the field layout coordinates i provide my teams goes down to the max (in decimal feet), and they still literally lop off everything after the third decimal point. Using the Decimal FEET high test accuracy, i can "fix" the dim (even 20 miles from the origin), to be back within .000000000006 decimal feet, at any value of my choosing.

              I (personally) would say you are covered. But i guess if you want to file a SR and wait for an answer that wont fix it, there is that option.
              Last edited by Twiceroadsfool; April 26, 2017, 03:19 PM.
              Aaron "selfish AND petulant" Maller |P A R A L L A X T E A M | Practice Technology Implementation
              @Web | @Twitter | @LinkedIn | @Email

              Comment


                #8
                Originally posted by snowyweston View Post
                I imagine, aside from the obligatory "let it go" assurances (that it's not the end of the world) someone might wade in with the limitations of 64bit Floating Point precision (?) or some other technological 'reason' - which I'm keen to accept - but if such is true, why the option to punch up to 12dec.pts at all? Why expose the limitation and get me, and potentially others, wound up scratching our heads?
                Did I hear floating point?
                Revit OpEd: Accuracy
                Chris Heinaranta | Architectural Technologist

                Comment


                  #9
                  Interesting you have both talked about M. Same if you type in 5000? I.e. mm in millimetre project.
                  Motorbike riding is one long bezier curve

                  Comment


                    #10
                    Originally posted by biff View Post
                    Interesting you have both talked about M. Same if you type in 5000? I.e. mm in millimetre project.
                    It happens regardless. Even happens if you switch Project units to mm,and 12 of accuracy, BEFORE you do it.
                    Aaron "selfish AND petulant" Maller |P A R A L L A X T E A M | Practice Technology Implementation
                    @Web | @Twitter | @LinkedIn | @Email

                    Comment

                    Related Topics

                    Collapse

                    Working...
                    X