cad data corrupted caused by compressing-tool?

Discussion in 'SolidWorks' started by JOJO, Mar 8, 2007.

  1. JOJO

    JOJO Guest

    Hi there,

    I recently stumbled upon the following problem:

    I exported a SolidWorks-file as Parasolid, compressed it at normal rate with
    WinRar (ZIP). The part consisted of some freeform geometry and a bore with
    a plain diameter of 20 mm.

    Our customer unpacked the part with WinZip, opened the part in SolidWorks.
    No changes in geometry or shape can be measured up to this point. Then he
    exports the part as STLwith fine quality settings for rapid prototyping. Now
    the STL file shows some differeces in geometry. The scale and especially the
    bore diameter is not round but somehow oval and the meassure is stretched to
    20,5 mm!!!

    When I send the exactly same Parasolid not compressed, the part also opens
    correctly in SolidWorks at our customers. He then exports with the exactly
    same STL export options as before and the STL part is generated correctly as
    it should be.

    ??? This behavior is reproducable but a reasonable explanation is still
    lacking.

    So what could be the conclusions for this weired behavior? Could WinRar or
    WinZip have caused the screwed STLs?

    I hope I was able to describe my problem. Any opinion on this subject is
    very much aprechiated.

    Thanx & regards from Germany
     
    JOJO, Mar 8, 2007
    #1
  2. JOJO

    JOJO Guest

    Hello Paul,

    thanks for the fast reply. Yes, I tried to open the compressed Parasolid
    and I generated a STL. Sofar with bare eyes it looks quite good to me,
    but I have no possibillity to measure an STL within SolidWorks (do I?).
    My customer uses a slicing software (magix?) to prepare data for SLS
    and with this software he can compare incoming and outgoing data.
    I would guess that my data gets corrupted by either one of the two
    different compression tools. Winrar (compress) -> WinZip (decompress),
    but I would like to proof.

    I am also quite confident that my data leaves without degression and gets
    crumpled by my prototyper. ;-)

    Of cause there are many ways to avoid this problem. Best solution
    seems to me just to send uncompressed parasolids (the size is quite
    as compact as compressed, so zipping is not really a big gain)

    The most important point for me would be to find out, if data
    compression could generally affect 3d geometry somehow.
    E.g if data compression worked compareable to jpeg compression,
    a data change or even loss of information could be explainable.

    I will run some further tests at my customers tomorrow and post
    the results, (if there are any)

    Good night

    JoJo
     
    JOJO, Mar 8, 2007
    #2
  3. JOJO

    Dale Dunn Guest

    Try running an MD5 hash before compression and after extraction. (Search
    Google for free tools.) If the hash checks ok, then it is very unlikely
    that RAR or ZIP have altered the data. That might be something you could
    tell the customer and say "you must have accidentally changed a setting".

    Anyhow, it's incredibly unlikely that a file compression tool could corrupt
    a file in such a way that a BREP model like a Parasolid would show up with
    a non-uniform scale applied to all the numbers which represent geometry,
    but not all the headers and whatnot that make the file readable. Data might
    get corrupted or otherwise go missing, but it shouldn't be multiplied by
    some factor that is remarkably similar to a shrinkage factor that your
    customer's software might routinely apply.
     
    Dale Dunn, Mar 8, 2007
    #3
  4. JOJO

    Cliff Guest

    Short answer: no, not with binary data compressions such a Zip
    file.
    Such are not lossless compressions as the binary ones are and
    the exact original data is somewhat lost.
     
    Cliff, Mar 8, 2007
    #4
  5. JOJO

    Cliff Guest

    LOL .. it's now a polygon mesh; inexact.
    AND they probably have a canted plane they are measuring across
    or something.
    Operator error IMHO.
     
    Cliff, Mar 8, 2007
    #5
  6. JOJO

    JOJO Guest

    Cliff,

    I am aware that stl is a poligonal mesh but that´s no reason why the part
    shows a divergence of almost 3% percent. Not to mention that stl export
    conditions were set to high. The whole part is only about 60 mm in height
    and therefore the divergence is really cosiderable!

    Regarding your concern about the method of measuring, you can be ashured
    that they are professionals. Especially in measuring small parts as yours,
    mentioned some threads below.
     
    JOJO, Mar 9, 2007
    #6
  7. JOJO

    JOJO Guest

    Hi Dale,
    thanks for the MD5 hint. I will test that asap.

    Regards

    Jojo
     
    JOJO, Mar 9, 2007
    #7
  8. JOJO

    TOP Guest

    It is my experience with rapid prototyping that scale factors can be
    introduced by the slicing software. This is done intentionally because
    many RP processes have shrinkages that need compensation. This depends
    on the orientation of the part. Send your contact the same part
    oriented to the front, top and right planes and see if they get
    differenct scaling on the hole.

    TOP
     
    TOP, Mar 9, 2007
    #8
  9. JOJO

    JOJO Guest

    Hi,

    our first suspicion was also the internal scaling applied by the slicing
    software. We doublechecked that the data degression already
    occured before the scaling was applied for sls processing.

    I know, it sounds like magic. Now I know why the software for
    slicing cad data is named Magics from Materialise.
    This causes me headache.

    Regards

    JoJo
     
    JOJO, Mar 9, 2007
    #9
  10. JOJO

    Cliff Guest

    Then you are just imagining the problem <G>.
     
    Cliff, Mar 9, 2007
    #10
  11. JOJO

    Bill Cain Guest

    Use a program like Beyond Compare to compare the 2 files in question. First
    file would be one made and never compressed. Second would be the one
    compressed with WinRAR and un compressed with WinZIP. If you find the files
    to be exactly the same, then I would think there was a slightly different
    setting when your customer generated the STL.

    Bill Cain
    www.partmaker.com
     
    Bill Cain, Mar 10, 2007
    #11
  12. JOJO

    Cliff Guest

    Lossless compression programs (such as PKZip) should error out
    if the compressed file has even one bit wrong (can happen due
    to several reasons from radiation to media or transmission problems).
     
    Cliff, Mar 11, 2007
    #12
Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.