Inverted Normals nowhere to be found in Dragon Necklace.

Discussion in 'Design and Modeling' started by 5839_deleted, Apr 18, 2009.

  1. 5839_deleted
    5839_deleted Member
    I've looked over this model's normals to ensure they were all aligned correctly, and I'm absolutely certain there are no problem areas. Looked over every polygon just to make sure. However, the validation service still tells me there are inverted/mixed/irregular normals.

    You may view the model below, as well as the attached .stl.

    [​IMG]

    Could someone please assist me? I really want to get this necklace made and see what kind of results I get. I'm really unaware of what the issue could be.
     

    Attached Files:

  2. 11460_deleted
    11460_deleted Member
    Hi murderslastcrow,

    Have you tried inverting all the normals? When I open that file with MeshLab, it looks like all the faces are turned inward.

    David
     
  3. 5839_deleted
    5839_deleted Member
    Interesting- I'll certainly try that, first of all. I think I may have noticed a few issues that might concern manifold-ness, so I'll test that after this and report back on this thread. Any other suggestions or ideas in the meantime would be appreciated.
     
  4. ArMOO
    ArMOO Member
    Hi murderslastcrow,

    I was able to get ride of the inverted normals.
    But you've still problems in the colored area's. If I try to fix them I end up in destroying the model.
    You maybe don't see it but also some of the rings of the necklace are stuck together.
    I think you should remodel it .

    Check also the other pdf, made after cleaning the inverted normals. You still have problems there.

    Greetings AO
     

    Attached Files:

  5. ArMOO
    ArMOO Member
    Murderslastcrow,

    And another pdf.

    Greetings AO
     

    Attached Files:

  6. 4219_deleted
    4219_deleted Member
    Hi! Although I'm not really part of this thread, I thought I'd mention that some types of .stl are unavoidably problematic with certain models. I had a great deal of trouble with one particular model using .stl, even though I verified it as being non-manifold and having valid edge thicknesses (it was made in Blender, so I used the internal tools of the most recent release to check it). When I exported it as .X3D instead, it worked easily first time. So, my suggestion: try all the Shapeways compatible formats you have access to first, before going looking for the more complex issues. Read the upload formats guide page to learn more about this.
     
  7. lorddarthvik
    lorddarthvik Member
    As said above, try different export formats. What is shown in MeshLab is never to be taken seriously, it´s compatibility with all the formats is pretty limited, near to useless. For example it makes a mess of mostly everything I export from 3dsMax, still those models upload and print perfectly. If you can, try to export to DXF and convert to Collada in AccuTrans, that usualy does the trick for me, especially when shapeways sais there are inverted normals although there is none in the model.

    Another way to test for where the inverted normals are, is to upload your model object by object to shapeways, this way you may be able to find the wrong part. It is a slow process, but effective most of the time. Still, this error may be caused by an export issue.
     
  8. Whystler
    Whystler Member
    *WAVES* at MurdersLastCrow!

    Good to see you here :)

    -Whystler
     
  9. ArMOO
    ArMOO Member
    Hi murderslastcrow,

    I made some better screenprints of what I can see.
    It might give you a clue where to go if you model it again.
    Maybe whistler can take a look to and give some advice?

    Greetings AO
     

    Attached Files:

  10. ArMOO
    ArMOO Member
    Murderslastcrow,

    And the last pdf.

    Greetings AO
     

    Attached Files:

  11. Whystler
    Whystler Member
    Alrighty then, I had a peek at your cool gothy necklace :)

    First of all , in Max, the piece appears too small for a necklace at 8-9 centimeters long. This might be big enough for a delicate bracelet in a perfect world. Unfortunately, at this size, the rings are pretty small to be printed with any confidence for the imperfect world in which we live ;). At best, the wire size of the rings should be 2mm in diameter. At worst, they shouldn't be any smaller than 1mm in diameter. As it is, they are not quite .5 mm, which is very thin. This might be a scaling issue in the program, so I resized it up 200% so that the rings would be at least 1mm which for me, is still to small for these rings. This makes the necklace about 17 centimeters tall (6 3/4 inches).

    So, secondly , I think the diameter of the rings should be bigger. At 1mm, they will be flexible in the SWF material, and in any other material they may break. Since a necklace has so many rings, it is inevitable that one or two rings may not make it in printing using the detail materials - perhaps more in shipping. On the safe side, I would aim for at least 1.5mm in diameter.

    Also, the rings that are part of the wings that attach to other rings below, are a little odd, because I think they were sculpted as part of the wing. Because they are odd, I don't think they are structurally sound, and will easily break. Instead, I would resculpt the wing without the rings. Then I would embed a torus in the bottom to match the other rings of the necklace, as you have done at the top of the ring.

    Next , I think you should use boolean functions to attach the horns and side rings to the head. I did this in max, followed by using the "cap holes" function to make the head watertight. This will stop you from having any manifold issues, because you won't have separate elements overlapping.

    And finally , you might want to rethink your ring-design in terms of how the necklace lays on the body when gravity is introduced. At the moment, gravity will make the wings cross each other and jumble up behind the head. So I would recommend you take the rings off of the top wing tips, and place them somewhere around the horny bits on top of the wings.

    Unfortunately, designing for 3d printing takes a bit more work because of the manifold issues. You will probably have to alter your approach to design compared to the good work I know you do in other platforms. Test out other 3d programs. Use different processes that aren't prone to making holes or manifold edges. This will take some time, but it's worth it.

    If you were starting this necklack from scratch, I would tell you to sculpt the head and the wings in your program of choice, without the rings. Then do some tests to make sure they are watertight and manifold. When you have achieved this, I would then import them into max, or another program where you would make the short tubes, or tori that create the rings for the necklace.

    Hope this helps!

    -Whystler



     
    Last edited: Apr 19, 2009
  12. ArMOO
    ArMOO Member
    Hi Whistler,

    Thanks for your advice. It also gives me insight.
    I think alot forget to use the boolean and causing the overlaps. But I still wonder why the whole piece had inverted normals.

    Thanks Again.
    Greetings AO
     
  13. Whystler
    Whystler Member
    I have no idea about why it had inverted normals, but once I started tweaking in max, I did have to invert the normals of the head and horns.

    -Whystler
     
  14. 5839_deleted
    5839_deleted Member
    Well, I did end up reading the 'problems' section of the site, and it cleared up the part of the holes and all that I was contemplating. I'm also going to fix the instable issues with the wing-rings you spoke of, and I recently became aware of some manifold issues with those rings pressing up against each other (one point shared by four faces on different objects is probably a manifold issue, as there is no way to imply the thickness of the point at which they meet.

    So thanks for the help in clearing up the issues with this. Now I have a lot more understanding to work with when I submit my next models. I suspected a lot of this beforehand, but some of it was unexpected.

    Thank you for all of your help and assistance, especially for Whystler's extensive detail. :3 Good to see you here, too. I'll come back to this post when I've got a model successfully uploaded. :3
     
  15. ArMOO
    ArMOO Member
    Hi whistler,

    When I opened the file it was only inverted normals.
    To clean them was easy. After i fixed athe 2 holes in ring at the top ( with the 2 spheres ) it had inverted normals in that area again.
    Somehow its funny. But if you can't see those things its like pushing buttons in the dark and waiting for the result.
    The darkroom of shapeways.........
    Did you refer to magic mushrooms somewhere? If you come to holland we go and find them....


    Greetings AO.
     
  16. 5839_deleted
    5839_deleted Member
    Hm, well, I ended up patching up the ring with the balls , as well as connecting the vertices of the horns and rings connecting to the wings and head, respectively. No vertices left inside, just perfectly connected and seamless.

    And still, the stl file claims to have inverted normals. (I remade the rings entirely since it seemed pointless to manually disconnect them and make them manifold since they're basic object, just redid them)

    So, no inverted normals whatsoever and still, claims inverted when I submit the stl. So, I guess when I export it to stl, it just inverts them automatically regardless of what I do, so I'm going to try a different format.

    P.S. Yeah, the scale was totally off, obviously. I have them in inches, not millimeters (same measurements, though).

    So yeah, I'll get back to you guys after I export x3d from max, blender, or sumpin. That way we shouldn't get so many weetodded issues.

    But yeah, made it completely water tight and connected (all but the separated rings, obviously), just to keep that from being an issue.
     
  17. 5839_deleted
    5839_deleted Member
    Well, I exported the fixed file from Max in .wrl format, but it still claims to have inverted normals. :eek: I'm quite befuddled with this assessment, since everything has been tended to as far as I can see. No inverted normals when I export it, whatsoever.

    So I'm going to upload a .3ds of the fixed file and ask politely if anyone can see what could be causing this error. Thank you in advance for your attention, I hope I'm not being burdensome. ^^;

    (still planning to fix the structural issue with the wing's bottom ring, but would like to see if it's going to work at all first)
     

    Attached Files:

    Last edited: Apr 20, 2009
  18. Whystler
    Whystler Member
    hey MLC,

    Here is what I was able to do for you ...

    I could not fix the horns, and they were creating problems. So I deleted them.

    I could not fix the chain. It was creating problems in certain links. So I deleted it.

    I could not fix the rings that were attached directly to the head. So I deleted them.

    I cut the ring parts off of the bottom of the wings.

    I resized it into cm to be the equivalent of the size you need in inches.

    So, what I have for you is a head (no horns), two wings, and the two large rings on either side of the head. It passes my checks and comes up watertight in Accutrans as per the Shapeways tutorial.

    Here it is in obj form. Use primitives to resculpt the horns in max. Use the tori primitives to remake the missing rings. And you're in business! I hope someone else can do better for you, but if they can't at least you've got a start :)

    -Whystler

     

    Attached Files:

    Last edited: Apr 20, 2009
  19. lorddarthvik
    lorddarthvik Member
    Well, I´v checked your model too, and I have to say altough it looks good, underneath it it looks like that some operation, like using boolean in Max have made a mess of your otherwise great design. Did you try to attach those parts ,which Whystler had to delete, together with boolean? I also tried fixing them, but remodeling them will be much faster and net you a clean result.

    I would just model the horns and all those parts separately, then move them into each other just a bit, so that they print as one piece. There is realy no need to attach them together as a continous one piece model. Especially not with boolean type modifers if your using Max :) Intersecting objects are allowed and print alright, so feel free to model everything as separate objects.

    I realy hope you get to print this model, I just love everything with dragons in them :)
     
    Last edited: Apr 20, 2009
  20. 5839_deleted
    5839_deleted Member
    Oh really?? Interecting polygons are perfectly acceptable as long as their points aren't merged? Aha, that makes things easier.

    Still kinda' confused as to why this approach wouldn't work comparitively. But I'll try again (maybe tomorrow). @,@ It'd be nifty if there were a plugin to adapt models to the needed thickness/alghorithms/smoothness or whatever. I mean, it would take forever to code, but still pretty cool. XD Maybe in the coming years.