Help getting rid of holes in prints of model

[krikru 1]

I have been printing this fractal pyramid (only the smallest version, pyramid4.stl, so far) from Thingiverse. As can be seen in a close up rendering of the model in Cura, there are tiny gaps between horizontally neighboring subpyramids, probably to guide the slicer into choosing an optimal path for the nozzle to traverse.

 

 

However, these gaps seem to backfire. I keep getting holes withing layers of the print, which is aesthetically unpleasing, because mathematically, there are no holes in the fractal shape (being the top half of the Sierpinski octahedron). These holes can be seen when viewing the pyramid from the top, but not when viewing it from the side, as can the seen in the following image:

 

 

Is there a way to get rid of these holes somehow?

Edited July 14, 2021 by krikru
Specified which of the pyramid I was printing

[Smithy 1,144]

Have you enabled the option "Spiralize" as the author of the model mentioned in the description?

[krikru 1]

No I haven't, but I have used single-walled without infill, which is also mentioned in the description (and one bottom layer).

 

I think the problem is that there is a gap in the model, which makes Cura leave out a gap in the printed model as well. What I would like to happen is that the walls on both sides of that gap is moved closer together so that they merge. Since I don't see how you would encode this in the stl file, the only hope is that it is possible to get Cura to do this somehow, so it is possible to make cura move those walls a little so that they overlap slightly?

 

(Alternatively I could perhaps learn gcode and write a script that modifies the resulting gcode file to make it do what I want, but I see that as a last resort and something that I shouldn't have to do if Cura was fully capable.)

[GregValiant 1,250]

In the Mesh Fixes area of the settings is "remove all holes" and you could give that a try (remember to turn it off when you are done).  You also might see a change by making Horizontal Expansion a negative number.

 

Cura can do some things to alter a model but for the most part - you're stuck with the way the artist created the model.  If a layer is offset too far from a previous layer then there will be a gap in the XY that you can only see in the Z (which is what you have).  You could try a lessor layer height. 

It's a trigonometry problem.  Given a .4 x .2 rectangle under the nozzle - if the angle of a surface is less than (about) 27° from horizontal there is a gap between the extrusions as seen in the Z.  It's the "step" distance from one layer to the next.  The printer will bridge across the gap if it's small, but there will be a gap.  In a case like that, there might not be holes in the model, but there will be holes in a single wall print.  Consider the cabin roof of a Benchy if printed as a single wall model.  The lines from one layer to the next are far apart in order to describe the slope of the roof.  As the angle of a surface approaches horizontal the step distance approaches infinity.  That's a big damn gap.

I learned this from Pythagoras himself when I was a little guy (that was a couple of years after @Torgeir and @kmanstudios took his class).  Either a lower layer height or a larger nozzle will change the geometry.  Shown is a .4 nozzle at .2 layer height.

Edited July 14, 2021 by GregValiant

[kmanstudios 1,120]

Which pyramid are you printing? There are three in the zip folder.

[kmanstudios 1,120]

I took a deep dive into the files.

• They all passed the 3DS MAX stl check...so good there
• The print used in the photo on the link is Pyramid 6 You can tell by the size
• I am guessing that you printed Pyramid by the density of the iterative triangles and the size

I am now printing out pyramid 5 *just over 2 hours just to see what happened. Full disclosure is that I am printing on an S5 (r1) and not a prusa and I used Arachne engine just for kicks and giggles. I also sliced using spiralize and 0% infill as suggested.

[Torgeir 242]

Hi there,

 

Just discovered that you need to deselect "Smoot Spiralized Conturs", if selected.

The squares will be kind of strange and offset on two of the sided of the "pyramid".

 

Torgeir

 

I'll post a picture soon..

[Torgeir 242]

Here's the pictures..

 

Only using "preview" mode in Cura Arachne Beta.

 

With Sprialize mode and Smoot (Spiralized Contur):

 

 

 

With Sprialize mode And No Smooting (Spiralized Contur):

 

 

 

Torgeir

[kmanstudios 1,120]

I just pulled my first print of pyramid 5 and I see no holes as the OP suggests. It is the one that I left on the smooth spiralized contours. I am now printing one with that setting turned off.

 

The first print seemed only present an issue with the layer height.....I just slammed it through with Fast- 0.2mm layer height. I am sure it would look better on visual 0.6 layer height.

 

I will post comparison pics when second print is done. That will be just over two hours from now since the print just started.

[Torgeir 242]

I'm seeing the pictures vertically and the first is with smoothing on.

You can see the actual setting in the picture.

[Torgeir 242]

I'm wondering if this model really is made for FDM printing, seems more like to be printed with .25 mm nozzle or for SLA printers (?).

I zoomed it out 200% and it's looking more like a model for our type of printers..

 

Torgeir

 

[kmanstudios 1,120]

1 minute ago, Torgeir said:

I'm wondering if this model really is made for FDM printing, seems more like to be printed with .25 mm nozzle or for SLA printers (?).

I zoomed it out 200% and it's looking more like a model for our type of printers..

 

Torgeir

 

My print came out fine. If you take a look at the one that was printed (Pyramid 6) on Thingiverse, you can see it has much larger areas to work with. 

 

I printed Pyramid 5 and it looked fine other than the layer height. If one wanted to, one could play with line width and layer height for a more refined print.

 

[Torgeir 242]

Aha, thanks I'll try that.

[kmanstudios 1,120]

OK, so this is the print of Pyramid 5. There is very little difference between the two methods of spiralize. One with smoothing on (left side) and one with smoothing off (right Side). Not really much difference. Could have tweaked the settings to get the top to print better, but it was not the issue. Also lousy trimming of the brim.

 

However, after re-reading the OP, I noticed it did mention using Pyramid 4 and I found the issue immediately:

Here you can see that a straight slice produces the holes, or gaps. I used Cura 4.8, but have no expectation that there would be any difference. But you can see the gaps that are inherent in the Pyramid 4 model. You can see it in the slice as well as the shading of the original model. It was made that way.

 

I then applied just a 0.05mm horizontal expansion and the gap went away without destroying the aesthetic of the model.

 

Cura's version of 'Zap-a-Gap' I guess 😆

 

The thing to keep in mind is that the Thingiverse poster showed a pic of the Pyramid 6 which is a much more dense and connected model. It is impossible to compare two separate models without understanding the difference. Had I not missed the model number being printed this would have been solved much earlier. 🙂

 

But there ya have it.

[kmanstudios 1,120]

This is the 'slam it through as fast as you can' print of Pyramid 4. Again, could use some tweaking. Spiralized with smoothing turned off.

 

[Torgeir 242]

Hi there.

 

Ehm.. Well Greg, I've to admit I cheated -a little.

So I pumped the model a little up, +200%, all this in order to see the lines I love so much..

As well, this is done with the "old" UM2Echo plus..

 

Here is it..:

 

 

 

 

 

[krikru 1]

13 hours ago, GregValiant said:

In the Mesh Fixes area of the settings is "remove all holes" and you could give that a try (remember to turn it off when you are done).  You also might see a change by making Horizontal Expansion a negative number.

 

Enabling "Remove All Holes" didn't do anything for me. Besides, even though I want to remove the gaps, I still want to keep the walls on both sides of the gap (in order to join what's below with what's above for the structural integrity), or rather merge it into one wall. Will the walls be kept if that horizontal gap is closed?

 

13 hours ago, GregValiant said:

... You could try a lessor layer height. 

It's a trigonometry problem.  Given a .4 x .2 rectangle under the nozzle - if the angle of a surface is less than (about) 27° from horizontal there is a gap between the extrusions as seen in the Z. ...

 

About the able, all surfaces (except the bottom which is completely horizontal) have an angle of atan(sqrt(2)) which is about 55 degrees, and the corners have an angle of 45 degrees, all much higher than 27 degrees. This gap is build into the stl file and is not going to be closed by a smaller layer height. If you scale it up without increasing the wall width, you may even get this effect. However, I used a 0.28 mm layer height in my print, but with a 0.4 mm nozzle and a 0.4 mm line width that should still not be any problem.

 

9 hours ago, kmanstudios said:

Which pyramid are you printing? There are three in the zip folder.

 

I am printing the smallest of the pyramids for the moment, pyramid4.stl, to decrease the "iteration time" 😁 I didn't write it in my question originally but edited it in. You can seem which of the models it is by counting the number of unique "hole" sizes in the pyramid side. Pyramid 4 has four hole sizes, Pyramid 5 has 5 hole sizes and Pyramid 6 has 6 hole sizes.

 

Each larger model consists of 6 samples of the previous model (4 in the base, 1 in the top and one upside-down right under the top, hence the word "fractal" in the model name), has twice the base length and is going to take approximately 6 times longer and use 6 times the material compared to the previous model when both are printed at full scale—if we print without the bottom layer (which I don't, by the way). For example, so pyramid6.stl should take approximatelly 6 * 6 = 36 times as long time to print as pyramid4.stl. The finest details are still equally large in all models. Hence, I'm sticking to the smallest model until I have nailed that.

 

6 hours ago, Torgeir said:

Here's the pictures..

 

...

 

Wow, that is some really wonky walls you get when using Smoot Spiralized Contours! I notice that I'm also getting wonky walls when using that setting, but your walls seem to be wonkier.

 

6 hours ago, kmanstudios said:

...

 

The first print seemed only present an issue with the layer height.....I just slammed it through with Fast- 0.2mm layer height. I am sure it would look better on visual 0.6 layer height.

 

...

 

Do you mean 0.06 mm layer height? I don't think 0.6 mm layer height would look any better than 0.2 mm layer height (for any reasonably sized nozzle).

 

6 hours ago, Torgeir said:

I'm wondering if this model really is made for FDM printing, seems more like to be printed with .25 mm nozzle or for SLA printers (?).

I zoomed it out 200% and it's looking more like a model for our type of printers..

 

Torgeir

 

 

What makes you think that an SLA printer would be more suitable than an FDM printer? The model details say "It has the astonishing property that the horizontal cross sections are simply connected and change continuously with height, so can be printed perfectly with spiral mode (or single-walled without infill)," and "The original scale is to match the limit of a 0.4mm nozzle," so it seems to me like it is at least designed with FDM printers in mind, and not for a 0.25 mm nozzle but for a 0.4 mm nozzle.

 

3 hours ago, kmanstudios said:

However, after re-reading the OP, I noticed it did mention using Pyramid 4 and I found the issue immediately:

Here you can see that a straight slice produces the holes, or gaps. I used Cura 4.8, but have no expectation that there would be any difference. But you can see the gaps that are inherent in the Pyramid 4 model. You can see it in the slice as well as the shading of the original model. It was made that way.

 

  When I slice Pyramid 5, I get the exact same gaps. I think you should get those too if you get the gaps with Pyramid 4, considering that Pyramid 5 is just a repetition of Pyramid 4 at different spatial locations at 100 % scale. Here is a screenshot from when I have sliced Pyramid 5:

 

 

3 hours ago, kmanstudios said:

I then applied just a 0.05mm horizontal expansion and the gap went away without destroying the aesthetic of the model.

 

...

 

Cura's version of 'Zap-a-Gap' I guess 😆

 

Interesting! It works for me too, but seems to work better the more wall lines I use, since all the green lines are inner wall lines. You seem to use multiple wall lines, but I want to print it with only one wall line according to the model details because I think it will go much faster and will not weaken the model. And I would like to do this only with the walls that are next to a gap, because I think this will also make all holes slightly smaller, and I don't want to make the smallest holes any less visible than they already are.

 

3 hours ago, kmanstudios said:

The thing to keep in mind is that the Thingiverse poster showed a pic of the Pyramid 6 which is a much more dense and connected model.

 

No, Pyramid 6 is not any more dense than Pyramid 4. It is more detailed, but it is also bigger, so the details are equally sized as in Pyramid 4. According to the model details "The original scale is to match the limit of a 0.4mm nozzle.." I'm assuming this holds for all pyramid sizes. If you want an equally large but more detailed pyramid, you both need to go up a pyramid number, and scale it down to 50% size (and use a nozzle half the size).

 

3 hours ago, kmanstudios said:

This is the 'slam it through as fast as you can' print of Pyramid 4. Again, could use some tweaking. Spiralized with smoothing turned off.

 

 

What layer height are you using here? In my original post I showed a 0.28 mm layer height Pyramid 4.

 

2 hours ago, Torgeir said:

Hi there.

 

Ehm.. Well Greg, I've to admit I cheated -a little.

So I pumped the model a little up, +200%, all this in order to see the lines I love so much..

As well, this is done with the "old" UM2Echo plus..

 

Here is it..:

 

 

 

 

 

 

Nice! Did you use a 0.4 nozzle? And do you get any up the "pull-apart" effect shown here?

[kmanstudios 1,120]

6 hours ago, krikru said:

if you get the gaps with Pyramid 4, considering that Pyramid 5 is just a repetition of Pyramid 4 at different spatial locations at 100 % scale. Here is a screenshot from when I have sliced Pyramid 5:

I did not get gaps with the pyramid 5 or 6. I think the person just 'cut' pyramid 6 at various heights and left little gaps in the process. You can see in my prints that I had a solid base and no gaps.

 

6 hours ago, krikru said:

I want to print it with only one wall line according to the model details because I think it will go much faster and will not weaken the model.

Spiralize only prints one wall. All prints were done with spiralize.

 

6 hours ago, krikru said:

No, Pyramid 6 is not any more dense than Pyramid 4. It is more detailed

More detailed = more dense. But I reconsidered and figured that pyramids 4 and 5 are just cut versions of pyramid 6 by cutting off the bottom parts.

 

6 hours ago, krikru said:

 

What layer height are you using here? In my original post I showed a 0.28 mm layer height Pyramid 4.

Printed at .2 mm layer height in all prints I made.

 

6 hours ago, krikru said:

Do you mean 0.06 mm layer height?

Yep...bad typo on my part. I am leaving in the typo so that your comment will make sense to anyone who reads the thread. But, yes, should have been 0.06. 

 

And I used a 0.4 nozzle....nothing tricky on my prints.

 

7 hours ago, krikru said:

hence the word "fractal" in the model name

Yes, I am familiar with fractals and have designed and made several 3d printed Sierpinski pyramids on my own.

[Torgeir 242]

 

12 hours ago, krikru said:

Wow, that is some really wonky walls you get when using Smoot Spiralized Contours! I notice that I'm also getting wonky walls when using that setting, but your walls seem to be wonkier.

 

 

This might be an issue depending of the kind of graphic display card we're using..

 

12 hours ago, krikru said:

What makes you think that an SLA printer would be more suitable than an FDM printer? The model details say "It has the astonishing property that the horizontal cross sections are simply connected and change continuously with height, so can be printed perfectly with spiral mode (or single-walled without infill)," and "The original scale is to match the limit of a 0.4mm nozzle," so it seems to me like it is at least designed with FDM printers in mind, and not for a 0.25 mm nozzle but for a 0.4 mm nozzle.

 

SLA printers have higher resolution and give more details in small objects.

 

12 hours ago, krikru said:

Nice! Did you use a 0.4 nozzle? And do you get any up the "pull-apart" effect shown here?

 

Sure I've used 0.4 mm nozzle and 0.2 mm height.

And I can print it without any hole in the bottom, -but the STC file show all this opening between the pyramids and this is what it should look like. Doing this with a 0.4 mm nozzle is not a good idea and will become much worse if layer is height increased to 0.28 mm.

 

I'd just frame grabbed this "picture" from Cura Arachne Engine Beta.

I've put a line representing 1 mm on the bed just to see the problem better.

 

By the way, -if you like to close the bottom "holes", just increase the first layer width enough and all the holes will be gone.

 

Here is the picture that show why there is small holes in the bottom and why this object is to small in order to show all those details.

As this picture shows two object, the single layer path cannot be printed (shown) here as this picture is represent the resolution difference.

 

 

Hope this help.

Good luck

Torgeir

 

 

 

[Torgeir 242]

The print I did, is the expanded "pyramide 4" zoomed 200%.

 

If you like to print the "flexible" version, just decrease the first layer width by say (-0.1 to -0.2) mm and the real bottom will appear.

 

Torgeir

Edited July 15, 2021 by Torgeir
Text.

[kmanstudios 1,120]

48 minutes ago, Torgeir said:

If you like to print the "flexible" version, just decrease the first layer width by say (-0.1 to -0.2) mm and the real bottom will appear.

I did not think of that! Thanks!! 🤠

[Torgeir 242]

Thanks.

 

For the record..

It's found under "Walls": named "initial layer horizontal expansion".

 

This one is also used to avoid the "elephants feet" -and for normal sized print we'll use from (-0.4 mm to -0.5 mm).

 

Torgeir

[krikru 1]

On 7/15/2021 at 3:57 PM, Torgeir said:

By the way, -if you like to close the bottom "holes", just increase the first layer width enough and all the holes will be gone.

 

 

That's true. But those are not the only holes in the print; here you can see holes higher up than the first layer (I've put the pyramid on a hook):

 

 

On 7/15/2021 at 4:03 PM, Torgeir said:

If you like to print the "flexible" version, just decrease the first layer width by say (-0.1 to -0.2) mm and the real bottom will appear.

 

I don't think this would work, because there are gaps higher up in the model too (as just concluded) that also need to be there in order to make the print fully flexible, so you would need to do the same thing with all layers. And even if you expand all walls with a negative value, the outer wall line still merges with itself from the "inside" (i.e., the horizontal cross section of the model has very thin parts in which the outer wall line self-intersects), hence still preventing that that flexibility. I therefore think that your only option to get a fully flexible model is to print it either scaled up or with a smaller wall width (nozzle) than intended.

 

Edited July 18, 2021 by krikru

[krikru 1]

On 7/15/2021 at 9:48 AM, kmanstudios said:

I did not get gaps with the pyramid 5 or 6. I think the person just 'cut' pyramid 6 at various heights and left little gaps in the process. You can see in my prints that I had a solid base and no gaps.

 

Hm, that's surprising; since pyramids 5 and 6 are basically just more of the same compared with pyramid 4 (if you zoom in they all look the same), my guess would be that you get the same kinds of phenomena with all sizes. Maybe I should make prints of pyramids 5 and 6 as well to see how they behave.