This is very weird because this “fancy” design was the only thing that I took from the internet, and all I did was tessellate the model, and put it in this model. I am very confused why this object looks corrupted in this onshape anayalsis.
It’s not necessarily corrupt, What you see here is what is in the model. The colors are OnShape’s visual queue to differentiate between different parts. It sees this entire decorative facade as a series of parts.
I’ve not completed my analysis and will share later some prints but to be blunt, you are expecting too much of this model. It was not designed to be printed at scale on an FDM printer.
Getting back to your original question. If you want to be able to print this. You need to obey the suggestions the slicer is asking and repair the defective manifold. What’s more, my early tests indicated that it is not possible or practical to expect to be able to render these facade features at 0.08 layer heights. In other words, I’m afraid you’re trying to thread a needle using a rope.
Thank you so much for trying it, to try to figure out the issue!
My only issue which I have been trying to avoid at all cost was this, repairing the manifold. This is beacuse Orca slicer add infill to the big center piece, which I can’t have beacuse I will have other items inside that center piece that we need to see.
It also looks like even after you repaired the manifolds, and brought it back into Orca, it looks like it still messed up the door in the centre on top of the stairs.
Lets see how it comes out either way, since you tried to print it 
.
Thank you!
I’m not sure if I mentioned this, but I will be printing this at a scale of around 5x4x6 feet. It would be cut into lots of pieces and printed with multiple X1 Carbon’s over the coming months. It looks like the manifolds will need to be repaired either way beacuse when I cut it up, some walls just dissapear if I don’t repair it first as one piece.
Is there any way to repair everything besides for the center piece? What I could do, is print the courtyard, and repair those manifolds, and the print the center piece as a second whole model. Would that help?
So why are you trying to slice the entire model at once?
You have a lot more designing to do. Break your model into parts that will fit on a plate, including whatever additions are needed to connect the parts, and slice each part individually.
That is a rather important piece of information.
Having that critical information changes the conversation entirely. Put simply, you can’t do what you want to do without modifying this model. As an example, what was your plan for extracting the support material from inside the pavilion?
You do realize that you will need to support the model from within for the roof to come out correctly, right?
Also, looking at a cutaway of the non corrected model, you can see that the inner walls have no dimensions which is exactly why your model cannot be sliced without creating a sealed manifold.
I do not want to draw final conclusions until I finish printing test models but as is, you’re trying to do something with this model that is not possible with FDM. Where did you get this model? Perhaps there is another version.
I don’t believe that I will need a lot of supports because this pavilion will be cut into lots of pieces, vertically and horizontally. So it will end up being cut up in so many pieces that the roof will be on the floor, or with very little supports.
What happened with the cut? Was this just to illustrate that the walls are practically 0mm in thickness.
@Olias Were you able to print the file successfully?
Success depends on how you define it. Yes, the model was printable. Here it is, but forgive the crappy filament. In my view, I would not be happy with this result but mind you that these were printed at max scale for the print bed which was 10% of actual.
Before Extraction of supports. Printed at 0.28 layer height
After extraction of supports.
Sectional print(for faster prints), printed at 0.08 layer height. No support model is on the right and models printed with supports on left.
My conclusion is the same as our friend @lkraus
My first question is; Are you sure you want to do this in the first place? I’d estimate it would take me in excess of 50-100 hours through trial and error to find out which mix of CAD, Slicer and Mesh editing tools(Blender) were best for the task and I am already very familiar with the task at hand, albeit having never done it on this scale before.
I might add on to @lkraus the suggestion that you may want to first conceptualize how you would cut these parts. If your plan is to cut your model into squares as you indicated above, I promise you that this will simply not work and is in simple terms “it is a very bad idea”. You will most certainly drive yourself crazy. Cut squares are nearly impossible to hide the seams. On the other hand, if you can conceptualize cutting along functional barriers such as walls, doorways, floors and ceilings, you can hide the seams within the structure. Mind you for the larger contiguous sections like the roof, you may have no choice and may want to come up with a means to camouflage those seems into appearing like deliberate features such as brick and mortar.
If your still determined to move forward with this model
This model is in no way ready for print and needs a lot of rehab. If you do not have access to the original CAD drawings, it is possible to resurrect this by extracting the STL and then doing refinement work on the models component parts individually. But in my view, I believe although it will work, you are looking at many hours of laborious and tedious tweaking of the model itself. If you enjoy this kind of work, which I do, then it will be a very fun journey. But if this is for either a class assignment, work related task or just about anything with a deadline, I might reconsider this if I were in your shoes. Just sayin… 
If you want to try it, here is what I might suggest as a soft entry into such an exorcise. Note that much of this can be done using the slicer so that’s the good news.
First, the model looks like some of the original work may have been done in CAD. I say that because the file shows evidence of having the Pavilion isolated as an object.
Import the model and then click on the O or P for part or object extraction.
Move the parts you want onto a different plate for ease of use and right-click and export as an STL. Note in this example I missed the ornate scrolls but I’ll leave that up to you to find which one of the 1,518 objects make up that section. A simpler method - simpler as a matter of expression not task - would be to delete the objects that you do not want and leave behind the decorations on the pavilion.
Here’s the reimported pavilion at 10% scale.
After you export the STL, reimport it into CAD or into the slicer to do pre-print cut and shaping. If you intend to do it in the slicer, you’ll have to master cutting, negative parts, boolean and assembly features. While not hard, if you aren’t already familiar there may be a learning curve.
Performing this exorcise on the pavilion is what I mean by soft entry. The Pavilion will likely prove to you, or disprove, both the concept of doing this, your skills in the matter, and whether or not you feel it’s worth it.
Alternatively, if you don’t have access to the original CAD, this is one type of object that lends itself well to open source Blender which is ideally suited for this exact type of application. If you’ve never used Blender, it has a very steep learning curve but it is the tool of choice used by game designers for complex in-game assets and model creations of everything ranging from Tanks, Space Stations and ancient buildings from antiquity. I’ve used blender in the past as a mesh repair tool for photogrammetry project I was doing some time ago. I found it daunting but one of my other hobbies was Ray Trace programming. So blender was sort of intuitive. If you don’t know what the work Phong means, blender is likely not for you unless you want to master a new vocabulary. Blender is to STL what CAD is to STEP. It has lots of tools, so much so that it can be overwhelming.
Here’s just one example of a model for sale created in Blender.
1 Like
BTW: Here’s a 3 minute video I found on how to take an STL 3D Printer file and make it hollow in blender. This is an advanced tutorial so keep that in mind. The speaker assumes you already sat through his other videos and that you’re a Blender expert. But it highlights the strength of Blenders tools to convert a dimensionless STL into one that is a hollow shell.
1 Like
@RocketSled I was wondering… I took the broken model into Bambu Slicer, and fixed the manifolds. Then I exported it and brought it back into Fusion 360. I was thinking of fixing the model, once the manifolds were fixed. Is there any way to extrude a solid model without meshing it, as this would reintroduce the issues, and takes hours, as it a huge model? Thanks!
I’m not sure what you mean by “extrude without meshing”.
If the original model was generated in Fusion, there may be export related settings you can change so that the issues I pointed out aren’t present to begin with. I’m not a Fusion user so I can’t offer much by way of more specifics.
Your walls had zero thickness, so maybe there’s a setting to make them non-zero thickness…
1 Like
I’m very confused… I have tried removing an area with a big manifold issue, put in a good piece from somewhere else, but it is still showing manifolds, and making random triangles. I have attached the STL, and some photos from what cura is giving me. Thanks!
You can’t just cut/paste geometry in a STL and have it become miraculously manifold.
Whatever you’re pasting back has to have exactly as many vertices located at exactly the same 3 dimensional coordinates as in the mating faces of the vertices of whatever you’re pasting your geometry in to. And then, once you paste that new geometry in, you have to tell the CAD program to “join” the now-overlapping-but-perfectly-aligned sets of vertices, the ones on the original geometry and the ones on the new geometry.
1 Like
The exact same thing happens to all my prints. Any ideas on how to fix that?
Fix the STL file so the geometry isn’t bad.
Fusion on my older PC here crashed trying to open the file, I also downloaded the obj version and it imported a lot better in Blender (from the last link you provided). It still shows some ‘flipped normals’ or could be open faces, and some duplicates on top of each other.
basically blue is good, red is not (face orientation function in blender).
The OBJ was scaled at 991 x 739 x 2200 meters, which is way too large, it should be scaled to the actual print size since you are not going to run simulations on it, even then.
1752 objects (after splitting to loose parts)
I moved what seemed to be the gound plane on negative Z and it appears a part of a building is attached to it, but multiple faces cross each other in the plane, could be from weird modeling or Fusion trying to convert/improve the model.
You see the bottom lane with red and blue jagged, which means theres multiple faces on top of each other, without thickness and some of them are oriented differently aka flipped normals.
There are some tools and workflows to clean this up in Blender, without splitting the object first, merge by distance in edit mode so alle overlapping points (vertex) are joined into one
Then again, looking deeper, my guess is you used photogrammetry models or downloaded exterior and modeled the interior.
The largest building looks like this, if you could model the outside like that, you would have no problems with the inside or manifold stuff.
This shows pieces of the larger building who were floating through each other.
These could have been joined in Fusion but messed up because of overlapping faces.
I would try to fix in blender and most likely remodel everything whilst thinking about how to print it at the desired scale. Make the small building separate, to print them individually, cut up larger buildings for multiple print volumes (there’s a mod for X/P to use full build volume, just keep Z-hop in mind).
A lot of the intricate details are useful as they are, the rest can be easily modeled and copied around.
This shows that your ground plane does have thickness, but is ‘open’ as others pointed out. The red should be on the inside of the model and not visible from the outside. Here it would be easier to remove the whole ground and replace it with a single cube scaled to match dimensions.