I created a 3D portrait bust in Blender cut it in half to expose the interior shell (similar to a mask) then exported it out as an STL… I have an issue slicing the model to create a smooth surface.
I really could use some assistance as to how to properly slice this. I am printing with the outer surface down (with supports) and the inner surface up.
I am uploading three images for clarification:
BlenderModel - This is the view of the interior ‘mask’
InteriorSliced - These are the weird artifacts I see when sliced
PhotoInterior - This is an actual photo of the 3D print.
I am trying to achieve a smooth surface. I have tried dozens of slicer settings, I have also tried thickening the model in Blender using the solidify option.
I think I am just missing something major. If anyone could tell me how to fix it I would be appreciated.
Unfortunately there is no magic pill to solve this basic law of physics. The way you have your model set up, you’re relaying on layers being deposited in the z-Axis. There really are only two tools in the “printer/slicer” tool box to address smoothness and both are blunt instruments. Then there is post-processing of the model itself. I’ll cover the first.
Category 1: Slicer settings
Tool #1: Layer height.
The lowest layer height supported is 0.08 as you probably already know. That will give you the best possible resolution for a 0.4 hotend nozzle.
Tool#2: Optional 0.2 Hotend nozzle
Whereas this will effectively double your resolution, it will not make it perfect.
Category 2: Mechanical post-processing of the model.
Sanding - Needs no explanation.
Filler and sanding - Just like performing body work on a car. Use putty glaze(which is typically red) to fill in the holes. Then sand smooth and coat with paint.
Chemical smoothing. For ABS filament, acetone is the solvent. For PVA filament, Isopropyl Alcohol is the solvent.
There are literally hundreds of YouTube videos on this topic. Virtually every single one in my view is 80% clickbait and 20% substance. Why do I say that? Because they show a nice very glossy thumbnail of a shiny object, only to reveal that all they did was apply a surface smoothing finish of some type. This is hardly what we are looking for, isn’t it. These videos in my view are a big nothing burger. Here’s the search term for them but they may give you ideas.
This one video is probably the most honest in that this person who is a model maker doesn’t hide the fact that he just uses paint, sanding and fillers to achieve his look. https://www.youtube.com/watch?v=xxdjFREJpKs
Here are some of the thumbnails that I was referring to that are more clickbait than substance.
My best cheat method that I’ve used. Glaze Spray Paint.
I use a simple technique for my smoothing. Purchase any triple-glaze spray paint, they make it in both gloss and matte. After printing at 0.08 layer height, sand down the surfaces. I recommend printing at 4 walls which will allow for enough extra material so that you don’t accidentally sand-through. Then spray paint and you will have the best solution.
You will always get the best results if you tils the model at an angle and use supports. This is a very, very, very did I say very? Time consuming, long and trial and error process to find the optimal angle.
Note however, by changing the default angle and using tree support I can get some relieve from the contouring affect that you desire to remove but not 100%.
Then there is the line width Parameters which can also help - to a small degree. If you are already printing at a 0.8 layer height, this parameter won’t make a difference but I note it because it can help you if you want to speed up your print and only slow it down for the outside lines.
The limits of what’s possible with FDM technology.
If you want to see what is possible using a combination of slower print speeds, post processing and finally polishing and glazing. There is no better example than the optically clear thread that was posted last year.
Although I was able to get some startling results. It was really in my opinion a parlor trick that relied on as much post processing as it relied upon profile tuning. Here is one example of the possible results. a 25mm cube with obvious clear but not perfect, optical properties.
While this will give you some measure of smoothness, it will be at the sacrifice of detail and may only give you some relief. In the end, you are limited in making fat filament skinny.
Thank you Olias for your very detailed and informative explanation of the issue and options available. I was hoping to minimize post production manual surfacing. The exterior surface would be simple to sand and finish but I really need the interior smooth. I could definitely finish the interior with the techniques you recommended but I am trying to reduce the labor.
When I built this model I exported only the surface polygons of the face. Now polygons are just planes with no thickness or depth so I am not sure how the slicer handles that. I was trying to achieve a single layer thickness for the shell and ideally something like a spiral vase print to minimize any artifacting. A spiral vase print did not work because I needed supports.
There are two different types of problems with the model that I see, one is the squiggly layer lines which I understand, and as you mentioned I may try switching to a finer nozzle and reducing the layer height to help minimize those lines. The other issue is the messy what looks like second layer, and I don’t understand where that comes from. I included an image of the slice with the two areas highlighted.
What I have been doing to minimize the condition is first painting the surface with a clear epoxy and then when dry painting with a matte paint.
If you watch the video till the end (about 12 seconds) you will get an idea of the effect I am going for.
I believe there’s a setting to ensure wall/ceiling thicknesses and that might be what’s triggering the second layer? I mention this since you mentioned exporting zero thickness polygons.
Printing a model that has what sounds like no thickness, it could be that setting is what is allowing it to print. Frequently when parts of a model get to be on the order of the line width, the slicer will just leave them out.
I don’t know that’s what’s going on with your print but maybe?
I will look that up. What I just tried was creating a cube in Blender and then using a bool subtraction of the positive of the mask. I made a prlnt and it looks great. I still see the layer lines but that doesn’t bother me. Those weird messy double layer artifacts are gone. I am going refine down the cube and I can play with a higher resolution nozzel. I am using Bambu Lab’s new white PETG-HF and I am printing in ludicrous mode. The semi matte surface is great.
@phodara quote=“MZip, post:4, topic:99755”]
Printing a model that has what sounds like no thickness, it could be that setting is what is allowing it to print. Frequently when parts of a model get to be on the order of the line width, the slicer will just leave them out.
[/quote]
This is my theory too. Without access to the model, I can only infer that the slicer might be trying its best to render a dimensionless model, i.e., one that has only a shell without solidity, and is therefore only able to approximate the geometry. There are tools such as Meshmixer that have the ability to make this solid. Fusion360 as an example, as a function that will convert a mesh into a solid.
@phodara : Without seeing a cross-section of this model, it’s difficult to provide specific help. We can only speculate on what the slicer is interpreting and the suggestion above is only a guess.
If, for example, your STL is a hollow mesh, meaning a shell with no solid inner surface, making the mesh solid might resolve this issue. However, without more information, we can only make educated guesses.
Here’s an example: On the left, we have a solid model, and on the right, we have a hollow model. You can see that the slicer treats these two models differently.
I’m assuming that this is a proprietary model that you don’t want to share. However, it would be extremely helpful if you could at least cut out the problematic section and upload that, so we can directly see what the slicer is interpreting. Of course, you’ll want to do a test print to ensure the model hasn’t already been made solid. If you encounter a “Non-Manifold” error, don’t fix it—try the print anyway if possible, or just upload the 3MF file with the open mesh error.
Also, where did this model originate from? Is it from a 3D scan?
It is a 3D photogrammetry scan done with an IPhone. My last print came out pretty good. As I mentioned I did a boolean subtraction from a cube. I am trying to reduce the back of the cube with another boolean subtraction, I want as a final result a shell with 1-2mm of thickness. I am doing another print now. My current model started as a solid cube, not a 0 thickness polygon shell. I don’t mind sharing the model, nothing proprietary except my image. Let me finish this print first and see how it works out. I can send you something tomorrow, I’m in the NY U.S. time zone.
Using boolean subtractions from a mesh I was able model it with about 3mm between front and rear surfaces. This worked perfectly and gave me a very nice surface finish on the inside surface. Here is a pic: Final Result
That my friend, was an ingenious workaround. I wouldn’t have thought to us a second shape with Boolean mesh. Absolutely Brilliant!!! Thanks for sharing.
The results are fantastic as well. I don’t think you would have achieved this level of quality had you gone the route I was thinking by using a mesh editing tool. I just learned something. Thank you.
Thank you for pointing me in the right direction and giving me more insight into all the options available.
One interesting issue I had was when creating the shell for the outside shape I had to scale it up, since it had to be larger than the inside surface. My first attempts failed, cutting off the tip of the nose and not properly forming the sides. I first tried by increasing each dimension (X,Y,Z) by 10%, but found I had to adjust each axis separately in addition to moving the outer shell away for the inner shell. I haven’t figured out how to calculate it programmatically yet, I just eye-balled it and experimented till I got the result I wanted. Assuming the nose is on the Y axis and moving the outer shell out (away from the inner shell) would give me additional depth or thickness for the final structure. But by moving it out on the Y axis the outer shell no longer covered the entire mesh cube on the Y axis and If i didn’t move it far enough on the Y axis it would cut the tip of the nose off. I had to increase the scale on the Y axis more than the X, Z axis. I also had to experiment with each axis independently. I still don’t understand how the topology affects or changes the percentages. I am afraid I could spend a week researching this, so for now I will just play with it, till it looks good.
Thanks for sharing.