Please improve the email feedback for models that don't pass

Discussion in 'Bug Reporting' started by 38494_deleted, Apr 15, 2013.

  1. 38494_deleted
    38494_deleted Member
    It would be GREAT if you guys could just put a bug report or some kind of file check log in the email you send back when a part doesn't pass. The canned email doesnt really help much.

    Hi ,

    Uh oh! Seems like your model needs a bit more love before our robots can work their magic.

    The 3D model MYMODEL you recently uploaded didn't pass all our automated checks, so it's not yet ready for 3D printing.

    Our tutorials and FAQ can be super helpful for more detailed information, or you can always ask the community a specific question on our forum

    If you're stuck or have any questions, don't hesitate to reach out at service@shapeways.com. Please include the file attached to the email, as well as any notes on your design, so we can help you solve your problem as quickly as possible.

    Kind regards,

    The Shapeways Service Team
    Ralph, Maartje, Kevin, Christel, Joost, Michon, Mitchell and Gary
    service@shapeways.com

    Shapeways
    Made in the future*
    https://www.shapeways.com
     
    Last edited: Apr 15, 2013
  2. AmLachDesigns
    AmLachDesigns Well-Known Member
    You are far from the first person to have complained about this, and over a long period of time nothing has changed.

    This email is not helpful, at all.

    As you imply, Shapeways knows why the part failed to upload, so why don't they tell us? As you have more experience you will start to have an idea as to the problem (usually thin walls/non-manifold object/non-connected parts etc) and set about fixing it yourself, but as a communication this email falls far short of what SW normally do.
     
    Last edited: Apr 15, 2013
  3. mkroeker
    mkroeker Well-Known Member
    As shapeways cannot possibly benefit from intentionally unclear error messages, I can only assume that
    the problem may be that the third-party software used for the automatic checking simply does not (no longer?)
    return specific error codes.
    Still I agree that the dude-speak "more love" phrase can be very infuriating - it may have been a cute in-joke
    for a geeky startup where everybody knew the handful of customers in person, but it could drive away would-be
    customers now.
    How about removing this sentence, and adding a more helpful "This could mean that your model
    has no thickness, or that it contains overlapping or misoriented parts or consists of more than 1 million triangles"
    between the "...not yet ready for 3D printing" and "Our tutorials..." ?

     
  4. Agreed, it's actually quite infuriating to post models and not have the slightest idea why it failed. Presumably the software that checked the integrity of the model generated a fail report, why not provide a link to that report for users interested in the specifics of the fail?

    This is still very new to a lot of people, so identifying the problems for them is going to minimize the amount of uploads, checks, and back and forth of emails.

    I'd also really encourage a notification for models uploaded which are over the 64Gb limit (or better again, refusing to upload it in the first place with a warning!). Currently, I don't get a notification saying the upload failed because of the file size, I get no feedback whatsoever, leading me to believe my upload simply failed. I uploaded the same 70MB model 3 times yesterday because of this before I realized my mistake. I'm sure I'm not the only person with this problem.
    Seán
     
  5. AmLachDesigns
    AmLachDesigns Well-Known Member
    In general I think that once you have been using SW a while you get to know how the system operates a bit and you can solve problems yourself/figure up how you ****ed up etc., but I think SW relies on this a little too much with the associated risk of scaring new users away.

    I remember when I started having these problems and getting either non-useful messages or none at all as you describe and getting very frustrated. Only through stubborness did I perservere.

    We are talking about data files here, and computers, and there is no excuse in my opinion for not getting better feedback at all stages of the process. If SW really wants to continue to be (imo) the best site it should cater for (i.e. notify of):

    - files size too big;
    - too many polygons;
    - part size too big/too small;
    - thin walls/wires;
    - too thick (ceramics);

    None of these should require guesswork or extended conversations with SW user support.


    Also, while we're are at it, we know software exists that gives wall thickness maps, so why not use this in the verification/notification of errors instead of the often misused Netfabb?
     
  6. Yeah, my stubbornness is beginning to wear thin, that's my problem. I'm using the exact same technique to hollow out two different models, and whereas the first one works, the second doesn't, and I have no way of finding out why. Especially infuriating since they've checked it, know the problem, and have just decided not to tell me the reason. Don't want to be included in those new users being scared away, just sending me the failure report would go a long way to preventing that from happening.
     
  7. AmLachDesigns
    AmLachDesigns Well-Known Member
    Well, as someone else posted somewhere else (is that vague enough?) probably the problem is that they have automated checks that do NOT tell the the reason, just pass or fail. My point is that they SHOULD have systems that give better feedback. Perhaps that 30 million could come in handy...

    For your model that's not working, what software? Have you already checked for the usual: size, non-manifold, normals?
     
  8. Yeah, I'm gonna count that as suitably vague thanks. :)
    I'd be very surprised if any software performing specific checks can't give the results of those checks. As in, failed because of found non-manifold faces, incorrect surface normals, etc. I mean, it seems inconceivable to me that anyone would perform a series of checks and not be able to tell you which of them failed.
    As for the model itself, I'm using ZBrush and relying on the Dynamesh and Shell functions to hollow out my models. It's worked on numerous models so far, and then suddenly with the latest model it fails every time. Guess I need to investigate other tools to tell me about my mesh integrity.
     
  9. AmLachDesigns
    AmLachDesigns Well-Known Member
    Netfabb works for a lot of people.

    Myself, I use Blender. To make a hollow model is simple:
    - delete a face or faces;
    - use Solidify Modifier to required thickness.

    There's some more details but that's the essence. I don't know if you have Blender (it's free) but you can import in several formats and do this if your other method doesn't pan out.

    Good luck!
     
  10. Yeah, cheers for that. Unfortunately, it's a bit more complicated than that for me. The model is quite complex, so hollowing i out in Blender, Max or Maya isn't really an option. I've just tried downloading MeshLabs, and it's a bit overwhelming without tutorials, so might check out Netfabb too, cheers!
     
  11. I am trying to upload my first model, I've sent many versions at this point, each time painstakingly trying something new, regenerating the model, and getting the same useless message.

    First it was too big, okay, they document that, I fixed it. Then it had too many polygons still, once again, that's a documented issue and I was able to discover that and fix it.

    Then I investigated more and it seemed like it might be an inverted normals problem. I fixed that, still doesn't work.

    I am out of ideas here!
     

    Attached Files:

  12. stonysmith
    stonysmith Well-Known Member Moderator
    Your file has 980k+ triangles. That's still pushing close to the limit for max triangles and may be causing a timeout in the automated process.
    For one thing, the bottom side doesn't need to have near as many facets as you have on it. The bottom _could_ have only two triangles instead of the several hundred that you have there.

    What material are you wishing to print this in?
    It's going to be more than $100 at this size in Sandstone and considerably more in any of the other materials..
    http://stonysmith.com/wired/VolumeEstimator.asp?L=127&W= 95&H=11&T=0

    Your triangle faces are approx. 0.15mm on a side. You could probably increase that to 0.2mm to 0.3mm with no major loss in fidelity of the image, while at the same time dropping the triangle count significantly.
     
  13. I did end up making it smaller and I converted back to ASCII stl instead of binary. For some reason that worked. I'm planning to print in the super high detail frosted plastic I guess, if I can afford it, but the total with the new files I uploaded is coming to 118, so I am not quite sure what I want to do yet.
     
  14. PDaniel
    PDaniel Member
    Mmmm.. in 3ds max there's a modifier useful for you cuz reduce the polygon number keeping the ecence. :)
    But at least you get a notification, I upload a model 2 ays ago and it's sill processing :rolleyes:

     
  15. stop4stuff
    stop4stuff Well-Known Member
    @sforsyth - The usual issue that Sculptris causes is internal bobbling to the model, this generally happens with very tight creases (the surface will turn inside out, cross over and bobble in the inside of the shell). When this situation occurs, Shapeways' MeshMedic checking software (and NetFabb Cloud) gets confused as to what faces should be on the outside and fail the model. The solution is to used MeshLab's remove intersecting faces function, repair the holes left (can be done in NetFabb), send the model through NetFabb Cloud service which should remove all of the left over internal parts.

    You can find out if the model has internal bobbling by slicing it open in NetFabb.

    Paul

    [hr][hr]
     
  16. Thanks Paul, it was actually ZBrush I was using, not Sculptris, but I suppose the same applies. Nettfabb seems to fix most of my problems.
    Thanks again!
    S