Suprised they didn't mention adaptive layers, that way you can have 0.3mm layers and transition to 0.1mm as the slope decreases. Same goes for the trapezoid shape.
Not sure if you're aware, but there are variable layer heights in almost all is not all slicers now. It allows you to set really thick layer heights for vertical surfaces while creating a gradient down to a finer say 0.05mm layer height for those rounded cases.
Another solution is variable layer heights. Use larger layer heights and only decrease when the horizontal stepover between the layers exceeds a certain size (down to a certain minimum layer height). This minimizes the amount of time spent printing at very low layer height, so the print times aren't increase as much, and you still get much better surface finish. Autodesk did something similar for their ember 3d printer. th-cam.com/video/HAmneiL5-jQ/w-d-xo.html Edit: Added a clarifying statement so that people don't think I'm there one who implemented it.
Another option is 'non-planar printing'. In this case, an app would warp the top surface to make it flat, the warped model is sliced, then the app unwarps it, resulting in extruder moves in 3 dimensions per layer--rather than the standard 2--and the result is a smoothly curved top. (Obviously, there are limitations to this method, but it works fantastically well within them.)
I would love for this to be implemented in a mainstream slicer without additional post-processing scripts. The problem is it doesn't apply universally to all models, so making the slicer do it automatically is very tricky. It'd be nice if Prusa or Ultimaker added it as a manual modifier, though, if they can't work out an algorithm cheaply enough to safely apply it automatically...
@@oasntet thats a tall order for slicer codebase. i dont think you can get away with less maths than in CAM for milling. i am into this, and will try to do something like this when my printer is rebuild. there are lots of possibilities.
@@simonschneider5913 Yeah, I'd expect it to need to be an advanced feature, probably a plugin, and it'd require a lot of hand-holding by the user. There's still a lot of rough edges in the external-to-slicer scripting approach that a slicer plugin could polish off, even without making it a mainstream feature.
Hey I just want to give you a big thank you! I enjoy your videos so much and they help us little guys print better parts. I know you could keep this information to yourself and say its trade secrets etc. but you don't. I feel you truly care about the 3d printing industry and want to see everyone be as successful as you, and put out quality parts. Kudos to you and your team. Thank you again for all your awesome videos!
The industry is so small. The more we can get people to use 3D Printing the more we all can win. When the industry is fully grown then we can hold our cards close.
In prusaslicer I add a height range modifier and choose a range that covers the curve and choose to halve the layer height for that section. You can then click the adaptive layer height icon in the toolbar and choose "smooth" to blend the transition area. I also add a low level amount of fuzzy skin - 0.03mm depth, 0.06mm distance to try hide the layer lines more.
You could try high resolution just on the top portion, getting most of the value for a fraction of the delay. I suppose a slicer could implement gradual layer height reduction to make it even nicer, but even without that, you can improve things a lot for a minor delay.
You can do adaptive layer height with Prusa/Orca/Bamu slicer and I think Cura as well. You can specify just the top parts too. It does help without adding too much time but doesn't get rid of the stepping completely.
Yup variable layer height. I use it all the time. Theres also a small layer height with an infill multiplier or different layer heights for infill compared to external surfaces. Saves time but preserves layer height. I've done this with .04 on the outside and .3 on the inside and the part is super smooth.
Well - adaptive layer height works, which works well in PrusaSlicer and friends. Also, Fuzzy skin, also in PrusaSlicer etc, does what you mentioned lastly, only without any work for the 3d model.
@@slant3d It's not. Adaptive doesn't increase resolution everywhere, only where needed. You should check it out as it's a nice compromise between fast printing and increased detail.
@@dsp4392 the whole point of the video was how to reduce this issue while keeping print time low. Adaptive is fine but it typically results in variable surface finish and quite a bit of extra time.
@@Cryous Not sure you are understanding what adaptive layer height means. In the dome example of the video if only the top 10% of the model would be printed at the 0.1 mm layer height then the total time would only be increased by about 7%. Moreover, the dome shape itself means the area of the top layers are dramatically reduced, which means the time increase is only a couple %, barely noticeable. Of course the actual impact depends entirely on the specific print, but generally adaptive layer heights increase print times by minor percentages, which I think he alluded to time increase of 50%, rather than 200%. It would have been clearer if he explicitly mentioned adaptive layer heights.
im confused, why did you add the chamfer to the bottom when it would be printing vertically? if the cylinder is printed vertically there wouldnt be any steep overhands right, just up and down. Im new and trying to learn best practices.
Adaptive layer height. It would be great if you could use a formula to determine h as a function of z For example h = min(max(0.1 , 0.4 sqrt (R^2/z^2-1) , 0.3). That would give you the necessary height to have a constant step of 0.4 mm but having 0.3 and 0.1 as upper and lower limits.
It's usually achieved using a setting called Fuzzy Skin. The slicer will simply add 3D noise to all vertical surfaces. The amount of noise is configurable too. I really wouldn't recommend doing it in CAD as it increases the polygon count and might cause unneeded detail loss.
Maybe I'm the odd one out, but I really don't mind the sort of stair-stepping appearance of layer lines towards the top of a curved surface like that. That said, I understand there's a lot of considerations that need to be made if you're producing and selling parts like this that I as a hobbyist don't really need to consider. I still really appreciate and enjoy your videos, and have incorporated some of your tips into my own designs already.
You could just print the part on its side. The other examples/suggestions are to change the geometry in its entirety which is not printing curved surfaces anymore. The scope for improvement here is null. How do you improve something but not doing it? Print in ABS and acetone vapour exposure. That is the ideal solution for mass producing curved surfaces.
I love using the stairstep approach to design. I will often go all the way down to .3mm steps, and then the part will perfectly match my designed shape and eliminate the layer effect that happens when you design and try to print curved parts!
I've found that just by adding extra walls helps to hide the effects of stepping. It obviously wont rid you of the layer stepping on a curve but more or so camouflages it instead. Giving you a cleaner looking finish. You could also just change the orientation of the line fill to be concentric rather than on the standard default 45. Consider if your part will be under stress, the latter may weaken your part slightly. Alot of people commented adaptive layer height and that would be fine for some things like this videos example but it can really make your prints look stripey. Which i dont care for at all. Id much rather have uniformity and a little bit of stepping than stripey layer height changes throughout my prints surface.
Cura has dynamic layer height exactly for this reason. While I find the stock settings somewhat lacking you can customize them to something that fits your need.
Another solution is non-planar 3D printing. Basically you can move in 3D space rather than relying on stacking a bunch of planes on top of each other. There are some experimental slicers for this, but it also will be limited by the nozzle configuration of a 3D printer. Many 3D printers would cause a collision between the hot end assembly and the print because the nozzle doesn't protrude far enough down from the hot end assembly.
Adding noise (or "fuzzy surface") is the most underestimated feature of most slicers - I use it a lot on visible faces on commercial products and it almost completely conceals the 3D-print look (=visible layers), what still a lot of customers associate with low quality. The products look more like injection molded with a nice texture. It even helps with overhangs as the lines "wobble" in x-y direction and are therefore a lot wider and can span shallower overhangs. Downside is: You have to carefully chose what surfaces you want to be treated like that or you will dramatically increase print time and overall wear of your printer because of the vibration it causes.
Great video. The concept of digital noise is interesting….is there a video on just that topic? Would be interesting to see how to add that to models in blender and or if it is just slicer settings that could be adjusted.
If you are trying to make tens of thousands just use injection moulding. Like seriously, 3D printing does scale pretty poorly. I understand it is appealing as you can gradually increase capacity as you go by adding more printers, but there is a limit to when it makes sense.
I like the solution I use of having the layer thickness adjust for the model geometry. That way, I can still print normal or extra thick layers on straight vertical surfaces, and infill. Then, as the model has angled or curved surfaces, the layer thickness adjusts according to the angle of the surface. I don't remember my exact formula used to calculate this off the top of my head, but it's something like: If (nozzle diameter * 1.1)/ ((previous stepover - current stepover requested)/2) > previous stepover Then current stepover = previous stepover * 0.75 Else If (nozzle diameter * 1.1)/ ((previous stepover - current stepover requested)/2) < previous stepover Then current stepover = previous stepover * 1.25 This isn't runnable code by any means, but you get the idea. There's also a check before this that spans multiple layers to see if it is even necessary to go into the modified layer height sub routine.
The slicer needs to be able to adjust player height for different parts of the print. With matterhackers matter control you could reslice it with shorter during the print as it gets towards the top
Depending on the print, it may be preferable to have the lines though. Like, when 3d printing wheels, the lines may provide extra grip, like tracks on the tires.
Non planar printing! It's a real pain to set up given the lack of support in most slicers, but if you're doing the same part again and again then it makes more sense. Absolutely no layer lines.
I was going to mention variable layer heights, but it looks like everyone else already beat me to it... I suppose I shouldn't really be surprised, but it would be nice to be (at least one of) the first on something 🤷🏿♀️
Great video! Another option that works for certain materials is to smooth out the entire surface with a chemical process, for example leaving an ABS part in acetone fumes for a little bit can give nice and smooth results.
I build an outer wrap with the layer lines going in the æsthetically optimal direction and slide the bulk of the part inside that wrap, that bulk having the layers in the direction optimal for strength. Made a toilet roll holder like this. Print outside cylinder vertically; print inside cylinder horizontally. Either that or I just remove the layer lines from the PLA with clove oil.
Care to explain what you mean about the clove oil? Are you just sanding it using clove oil as a lubricant? How does clove oil help with removing the layer lines?
It would be nice if i could do 3d ironing and be done with it. sort of like 3D printing, not 2.5 D printing if you know what i mean. OR, there should be a setting where if im printing the entire model with .8 layers, the outer layers can be printed at .4 layer height just to get the surface to look ok, but it wouldn't take as much time as doing the entire model in .4
I had a sphere to print on my Bambu Carbon, I haven't checked if the slicer can do variable layer height, will check! As well as using concentric infill for the topmost layers I used a bit of post processing using primer and sanding...
thx for your videos. i am also very interesting how to make noise or a texture on a surface using cad or maybe other tools? fuzzy skin is not always the right thing, because its difficult to disable it in specific surfaces like holes and so on
Seems like the most obvious solution should be software. Maybe eventually most slicers will have a factory default enabled that detects a certain slope angle and automatically starts increasing resolution the closer it gets to the top. So you could print the whole thing at a big layer height then the last few be small.
found your channel based on your videos about filament and found it super interesting. You said you make your own filament in house, but do you have any suggestions for purchasing filament until you guys are making your own? great video.
I wish I could find a good option for over-hangs in Cura. I end up with the worst possible surface finish imaginable anytime I have a near flat overhang. On some cases I can tell that the coarse crisscross support layer is fused to the part, but even if that layer comes free, there's a layer of a mess of coarse, partially fused, not flattened "lines" fused to the part... and I can't tell if that's a layer that should be wasted or if it's the actual first payer of the part. It's very rough and not dimensionally accurate too, so any flat over-hang parts can't be used in mechanically functional parts. They're just too thick and too rough. (...and sometimes the very coarse lines aren't fully fused to the part. They can be picket off with a finger nail and/or are already broken free in some places). Forgive the long explanation. Just trying to accurately describe the problem. I can say I've been Very lucky though. Got an Ender 3 just a month ago and it's printed almost everything I've tried to print very good. This has just been the one problem I couldn't easily find a solution for).
How do you add noise in CAD? Does Fusion 360 have this option? Is it just applying a texture/material in the CAD software? If so, could you elaborate on what method you use for this?? Thanks in advance!
Adaptiv layer hight is made for this. Who says that infill needs to the same layer hight as perimiter (Cura can do this). The outer perimeter could also be finer than the inner perimiter. (not sure if this feature exist, but I have tested it myself, with source code editing the g-code file.)
Can you do a video about interfacing with already built items? A lot of these design patterns don’t seem to apply when I have to work around an already existing item. I know it’s very broad but would be helpful.
I have been designing my screw hole parts with hexagon holes instead of circles because the slicer always messes them up and makes them flat at the top.
It's not really the slicer, the upper layers are sagging because they are unsupported, so it appears flat at the top. You can counter that by exaggerating the upper side of the circle and make a teardrop instead of a circle. But hexagons are probably a better solution because they don't have that problem, their sides are 45° and support themselves, topped by a bridge.
@@jeffreygordon7194 oops you're right, octagon is that I meant, hexagon would work when oriented with the flat side up, the side would have an overhang angle of 30°. But even with the tip up, 60° should give you better results than a circle, 60 is still quite printable.
I think there was a DIY 3d printer video where the author made a 5-axis 3d printer that print along a non-planar contour. So even spherical surface is not an issue with minimum layer line without needing to sacrifice print speed. But that machine does require custom 3d model slicing instead just regular planar slicing. I think i saw it 8 years ago, not sure why that 5-axis design didnt spread beyond university research project to the general public.
I don't understand why FDM printers don't able to rotate they extruder around an object in the first place. It won't add much more struggle to the printers we already have and it would be an nice alternative option. Only slicers would need to get updated for this. It would allow us to use smoothing layer option that we can see for ex. in Cura, while also it would allow to make parts way stronger and lightweight at the same time as we could combine different print orientations of lines on one object. While also getting rid of support needed for many situations where the head just move diagonal. Also you may then would able to avoid that you could kick a already other print from your print surface because a detection camera or something could calculate to print around it.
Everyone commenting variable layer heights, but my take is that often there is a texture difference on one or more sides which can look worse than the low angle steps.
Variable layer heights may help a lot while not increasing print times drastically. Or there are options (at least in Cura that I know) to define the wall and infill layer thicknesses separately. Print the walls with 0.1mm height and the infill with 0.3mm!
I appreciate the info you provide. I’ve been designing a box for some fans with using your methods. What methods do you use to reduce costs? The box takes 1.2kg of filament at 2 walls and 10% infill since it’s a fairly big box.
The easiest option is to use lightning infill though the part will be somewhat weak. Another option is to skeletonize the part by removing material in the cad model where it isn't needed. If you decide to skeletonize the part you can also reduce the thickness of the walls to a point where you don't use infill at all and just use a couple millimeters thick of solid material.
Some matte surface filaments make layer lines hard to see. Some glittery one will as well, probably for the same reason you mentioned marbled filament. For example, Prusa's Galaxy Black PLA prints both matte and with small glitter flecks. It can be really difficult to see layer lines when printing with that filament.
Idk why but for me those layer lines are fine, I hate visible tesselation and regular layer lines but I just don't see the problem with the top of curves. I'm 3D printing a quadcopter and there's a part of the arms where there's a top curve and I'm just fine with it.
Changing the design that someone has put lot of effort is the last think i would attempt to do. Instead of fixing the problem at print end you always ask customer to redesing 🤣
You mention printing on its side, but I'm surprised at how often printing at any angle will remove stairstepping rather than rotating a full 90 degrees.
Big thanks for this - great discussion. Sometimes ironing is necessary to smooth flat tops - I wondered if you had thoughts on how to optimise slicer settings to achieve best surface quality when ironing?
Not sure. But I think you mic is backwards bro. Depends on mic type, if its directional. Then you'll have a lot worse voice recorded quality, when the main input is on the control side.
you're going to hate this answer, but you can oversize the wall by a little bit and then sand it back down, not perfect, but it's stupid simple for small batches, or for parts that need a really smooth surface
One solution is just to flatten the top. Yes, it not the greatest but good design is about figuring out the best set of compromises. And 3D printing FDM is already a comprise on aesthetics, you're making for the multitude of benefits of FDM.
Would be nice to see variable layer heights and infill with thick layers. I can see an issue with adhesion as there can be air gaps, but adaptive wall thickness can also fix that.
it would be nice if someone made a slicer that would do perimeters at .1 and then infill then infill at .2 or .3 I don't see why this would not work but I'm not an expert in this.
I think you already showed another solution - to use splines instead of circles. So where the stepover becomes too wide, you put spline with greater curvature, so it mitigates that.
Suprised they didn't mention adaptive layers, that way you can have 0.3mm layers and transition to 0.1mm as the slope decreases. Same goes for the trapezoid shape.
came to make sure this was said
😂👍
I use this a TON! Great comment
Exactly
But the point was to have consistent layer lines, right?
Not sure if you're aware, but there are variable layer heights in almost all is not all slicers now. It allows you to set really thick layer heights for vertical surfaces while creating a gradient down to a finer say 0.05mm layer height for those rounded cases.
Another solution is variable layer heights. Use larger layer heights and only decrease when the horizontal stepover between the layers exceeds a certain size (down to a certain minimum layer height). This minimizes the amount of time spent printing at very low layer height, so the print times aren't increase as much, and you still get much better surface finish.
Autodesk did something similar for their ember 3d printer.
th-cam.com/video/HAmneiL5-jQ/w-d-xo.html
Edit: Added a clarifying statement so that people don't think I'm there one who implemented it.
Another option is 'non-planar printing'. In this case, an app would warp the top surface to make it flat, the warped model is sliced, then the app unwarps it, resulting in extruder moves in 3 dimensions per layer--rather than the standard 2--and the result is a smoothly curved top. (Obviously, there are limitations to this method, but it works fantastically well within them.)
I would love for this to be implemented in a mainstream slicer without additional post-processing scripts. The problem is it doesn't apply universally to all models, so making the slicer do it automatically is very tricky. It'd be nice if Prusa or Ultimaker added it as a manual modifier, though, if they can't work out an algorithm cheaply enough to safely apply it automatically...
@@oasntet thats a tall order for slicer codebase. i dont think you can get away with less maths than in CAM for milling. i am into this, and will try to do something like this when my printer is rebuild. there are lots of possibilities.
@@simonschneider5913 Yeah, I'd expect it to need to be an advanced feature, probably a plugin, and it'd require a lot of hand-holding by the user. There's still a lot of rough edges in the external-to-slicer scripting approach that a slicer plugin could polish off, even without making it a mainstream feature.
@@oasntetunfortunately, Autodesk patended non-planar printing in 2015 (and never used it for anything).
At 2:06 was there a reason to put a chamfer on it even though that profile was flat already? (not sarcastic)
Hey I just want to give you a big thank you! I enjoy your videos so much and they help us little guys print better parts. I know you could keep this information to yourself and say its trade secrets etc. but you don't. I feel you truly care about the 3d printing industry and want to see everyone be as successful as you, and put out quality parts. Kudos to you and your team. Thank you again for all your awesome videos!
The industry is so small. The more we can get people to use 3D Printing the more we all can win. When the industry is fully grown then we can hold our cards close.
In prusaslicer I add a height range modifier and choose a range that covers the curve and choose to halve the layer height for that section. You can then click the adaptive layer height icon in the toolbar and choose "smooth" to blend the transition area. I also add a low level amount of fuzzy skin - 0.03mm depth, 0.06mm distance to try hide the layer lines more.
if he says to just print it on the side im gonna scream.
e: lmao
Yeah, most of those tips were cop outs. "How to improve the curve: Get rid of it!"
You could try high resolution just on the top portion, getting most of the value for a fraction of the delay. I suppose a slicer could implement gradual layer height reduction to make it even nicer, but even without that, you can improve things a lot for a minor delay.
You can do adaptive layer height with Prusa/Orca/Bamu slicer and I think Cura as well. You can specify just the top parts too. It does help without adding too much time but doesn't get rid of the stepping completely.
Yup. It is part of Prusa slicer for some time now.
Yup variable layer height. I use it all the time.
Theres also a small layer height with an infill multiplier or different layer heights for infill compared to external surfaces. Saves time but preserves layer height. I've done this with .04 on the outside and .3 on the inside and the part is super smooth.
Well - adaptive layer height works, which works well in PrusaSlicer and friends. Also, Fuzzy skin, also in PrusaSlicer etc, does what you mentioned lastly, only without any work for the 3d model.
adaptive is the same as high resolution. The tradeoff is is slows down print time adding cost in mass production
@@slant3d It's not. Adaptive doesn't increase resolution everywhere, only where needed. You should check it out as it's a nice compromise between fast printing and increased detail.
@@dsp4392 the whole point of the video was how to reduce this issue while keeping print time low. Adaptive is fine but it typically results in variable surface finish and quite a bit of extra time.
@@Cryous Not sure you are understanding what adaptive layer height means. In the dome example of the video if only the top 10% of the model would be printed at the 0.1 mm layer height then the total time would only be increased by about 7%. Moreover, the dome shape itself means the area of the top layers are dramatically reduced, which means the time increase is only a couple %, barely noticeable.
Of course the actual impact depends entirely on the specific print, but generally adaptive layer heights increase print times by minor percentages, which I think he alluded to time increase of 50%, rather than 200%. It would have been clearer if he explicitly mentioned adaptive layer heights.
@@Cryous That "pixelated" "solution" also adds quite a bit of time as you are now introducing extra top layers to the part which need to be printed.
A lot of great advice here. The Pixelate tip really resonated with me, was exactly what I was looking for. Thanks for the vid
im confused, why did you add the chamfer to the bottom when it would be printing vertically? if the cylinder is printed vertically there wouldnt be any steep overhands right, just up and down. Im new and trying to learn best practices.
Adaptive layer height. It would be great if you could use a formula to determine h as a function of z
For example h = min(max(0.1 , 0.4 sqrt (R^2/z^2-1) , 0.3).
That would give you the necessary height to have a constant step of 0.4 mm but having 0.3 and 0.1 as upper and lower limits.
Can you explain the digital noise technique? That looks interesting.
It's usually achieved using a setting called Fuzzy Skin. The slicer will simply add 3D noise to all vertical surfaces. The amount of noise is configurable too. I really wouldn't recommend doing it in CAD as it increases the polygon count and might cause unneeded detail loss.
Nope, fuzzy skin is hard to apply selectively like they seem to do
@@dsp4392 thanks for actually explaining that. that helped a lot
Maybe I'm the odd one out, but I really don't mind the sort of stair-stepping appearance of layer lines towards the top of a curved surface like that. That said, I understand there's a lot of considerations that need to be made if you're producing and selling parts like this that I as a hobbyist don't really need to consider. I still really appreciate and enjoy your videos, and have incorporated some of your tips into my own designs already.
Thanks. Every descision always depends on the context of the product being make.
Embrace the layer line, the layer line is art
Um,,,,,Your print quality is exceptional.
You could just print the part on its side.
The other examples/suggestions are to change the geometry in its entirety which is not printing curved surfaces anymore. The scope for improvement here is null. How do you improve something but not doing it?
Print in ABS and acetone vapour exposure. That is the ideal solution for mass producing curved surfaces.
I love using the stairstep approach to design. I will often go all the way down to .3mm steps, and then the part will perfectly match my designed shape and eliminate the layer effect that happens when you design and try to print curved parts!
I've found that just by adding extra walls helps to hide the effects of stepping. It obviously wont rid you of the layer stepping on a curve but more or so camouflages it instead. Giving you a cleaner looking finish. You could also just change the orientation of the line fill to be concentric rather than on the standard default 45. Consider if your part will be under stress, the latter may weaken your part slightly.
Alot of people commented adaptive layer height and that would be fine for some things like this videos example but it can really make your prints look stripey. Which i dont care for at all. Id much rather have uniformity and a little bit of stepping than stripey layer height changes throughout my prints surface.
Can you show how you do the CAD Fusion 360 Noise Texture. You mentioned that previously, And I'm not a fan of Fuzzy Surface in Cura. Thanks.
Cura has dynamic layer height exactly for this reason.
While I find the stock settings somewhat lacking you can customize them to something that fits your need.
Another solution is non-planar 3D printing. Basically you can move in 3D space rather than relying on stacking a bunch of planes on top of each other.
There are some experimental slicers for this, but it also will be limited by the nozzle configuration of a 3D printer. Many 3D printers would cause a collision between the hot end assembly and the print because the nozzle doesn't protrude far enough down from the hot end assembly.
PLEASE cover how to introduce "noise" to the part! That effect looked absolutely awesome
Using the adaptive layer function of the slicer is good too. It only reduces the layer height over fine areas like the top. Saves on printing time.
Adding noise (or "fuzzy surface") is the most underestimated feature of most slicers - I use it a lot on visible faces on commercial products and it almost completely conceals the 3D-print look (=visible layers), what still a lot of customers associate with low quality.
The products look more like injection molded with a nice texture.
It even helps with overhangs as the lines "wobble" in x-y direction and are therefore a lot wider and can span shallower overhangs.
Downside is: You have to carefully chose what surfaces you want to be treated like that or you will dramatically increase print time and overall wear of your printer because of the vibration it causes.
So you add the noise in the slicer program itself and not in the CAD model? If so, I am assuming you can do this in CURA
Great video. The concept of digital noise is interesting….is there a video on just that topic? Would be interesting to see how to add that to models in blender and or if it is just slicer settings that could be adjusted.
The stair steps are simple and something I never considered. Thanks for the suggestions :).
I'd like to know how you create the "noise" you mention.
you can also try vapor smoothing with some plastics, batch processing is possible with large containers as well.
If you are trying to make tens of thousands just use injection moulding. Like seriously, 3D printing does scale pretty poorly. I understand it is appealing as you can gradually increase capacity as you go by adding more printers, but there is a limit to when it makes sense.
Sand it down smooth as a post process
Now do that 1 million more times
Can you explain in some video what kind of settings you use to add the digital noise to the surface?
In Cura you can search for "fuzzy skin". It's a feature they've had for seven years or so and I'm sure other slicers have an equivalent as well.
I like the solution I use of having the layer thickness adjust for the model geometry. That way, I can still print normal or extra thick layers on straight vertical surfaces, and infill. Then, as the model has angled or curved surfaces, the layer thickness adjusts according to the angle of the surface. I don't remember my exact formula used to calculate this off the top of my head, but it's something like:
If (nozzle diameter * 1.1)/ ((previous stepover - current stepover requested)/2) > previous stepover
Then current stepover = previous stepover * 0.75
Else If (nozzle diameter * 1.1)/ ((previous stepover - current stepover requested)/2) < previous stepover
Then current stepover = previous stepover * 1.25
This isn't runnable code by any means, but you get the idea. There's also a check before this that spans multiple layers to see if it is even necessary to go into the modified layer height sub routine.
Slicers could have an outter wall layer hight option. So it could print the wall first in a higher resolution and the rest in a normal height.
“How to prevent stepping”… “make a different part” 😑
Why no mention of variable layer heights?
The slicer needs to be able to adjust player height for different parts of the print. With matterhackers matter control you could reslice it with shorter during the print as it gets towards the top
Depending on the print, it may be preferable to have the lines though. Like, when 3d printing wheels, the lines may provide extra grip, like tracks on the tires.
Non planar printing! It's a real pain to set up given the lack of support in most slicers, but if you're doing the same part again and again then it makes more sense. Absolutely no layer lines.
How do you do that ?
It should be possible to print the outer layer with say 0.05mm 4 times first, and then the infill 0.2mm one time, so that the layer height match.
Im a new printer here, and this was a very well made video
Can't beat Epoxy Resin + Sanding
Non-planar printing might be an option in the most edges of cases. True 3d printing.
I was going to mention variable layer heights, but it looks like everyone else already beat me to it... I suppose I shouldn't really be surprised, but it would be nice to be (at least one of) the first on something 🤷🏿♀️
Great video! Another option that works for certain materials is to smooth out the entire surface with a chemical process, for example leaving an ABS part in acetone fumes for a little bit can give nice and smooth results.
I build an outer wrap with the layer lines going in the æsthetically optimal direction and slide the bulk of the part inside that wrap, that bulk having the layers in the direction optimal for strength. Made a toilet roll holder like this. Print outside cylinder vertically; print inside cylinder horizontally.
Either that or I just remove the layer lines from the PLA with clove oil.
Care to explain what you mean about the clove oil? Are you just sanding it using clove oil as a lubricant? How does clove oil help with removing the layer lines?
This was a helpful video, but not exactly what I was looking for. But thanks for the ideas
It would be nice if i could do 3d ironing and be done with it. sort of like 3D printing, not 2.5 D printing if you know what i mean. OR, there should be a setting where if im printing the entire model with .8 layers, the outer layers can be printed at .4 layer height just to get the surface to look ok, but it wouldn't take as much time as doing the entire model in .4
Can you cover, or provide links, or even terms we can google to learn how to utilize textures? I love that idea, but struggling on where to start.
I had a sphere to print on my Bambu Carbon, I haven't checked if the slicer can do variable layer height, will check!
As well as using concentric infill for the topmost layers I used a bit of post processing using primer and sanding...
you can always slice it on prusa and send the gcode to your bambu
thx for your videos. i am also very interesting how to make noise or a texture on a surface using cad or maybe other tools? fuzzy skin is not always the right thing, because its difficult to disable it in specific surfaces like holes and so on
Seems like the most obvious solution should be software. Maybe eventually most slicers will have a factory default enabled that detects a certain slope angle and automatically starts increasing resolution the closer it gets to the top. So you could print the whole thing at a big layer height then the last few be small.
This already exists
found your channel based on your videos about filament and found it super interesting. You said you make your own filament in house, but do you have any suggestions for purchasing filament until you guys are making your own?
great video.
Very helpful, thank you. I am new to the 3d printing world and would like to ask what software you are using?
I wish I could find a good option for over-hangs in Cura. I end up with the worst possible surface finish imaginable anytime I have a near flat overhang. On some cases I can tell that the coarse crisscross support layer is fused to the part, but even if that layer comes free, there's a layer of a mess of coarse, partially fused, not flattened "lines" fused to the part... and I can't tell if that's a layer that should be wasted or if it's the actual first payer of the part. It's very rough and not dimensionally accurate too, so any flat over-hang parts can't be used in mechanically functional parts. They're just too thick and too rough. (...and sometimes the very coarse lines aren't fully fused to the part. They can be picket off with a finger nail and/or are already broken free in some places). Forgive the long explanation. Just trying to accurately describe the problem. I can say I've been Very lucky though. Got an Ender 3 just a month ago and it's printed almost everything I've tried to print very good. This has just been the one problem I couldn't easily find a solution for).
You could probably do an entire video entirely on digital noise and fuzzy skin
How do you add noise in CAD? Does Fusion 360 have this option? Is it just applying a texture/material in the CAD software?
If so, could you elaborate on what method you use for this?? Thanks in advance!
I must be the only one but I find laters lines like this quite appealing
Adaptiv layer hight is made for this.
Who says that infill needs to the same layer hight as perimiter (Cura can do this).
The outer perimeter could also be finer than the inner perimiter. (not sure if this feature exist, but I have tested it myself, with source code editing the g-code file.)
Can you do a video about interfacing with already built items? A lot of these design patterns don’t seem to apply when I have to work around an already existing item. I know it’s very broad but would be helpful.
How do you add noise to the surface of your parts? I am really interested in learning more about this!
I have been designing my screw hole parts with hexagon holes instead of circles because the slicer always messes them up and makes them flat at the top.
It's not really the slicer, the upper layers are sagging because they are unsupported, so it appears flat at the top. You can counter that by exaggerating the upper side of the circle and make a teardrop instead of a circle. But hexagons are probably a better solution because they don't have that problem, their sides are 45° and support themselves, topped by a bridge.
The angles of a hexagon are 60 degrees. 45 degrees is an octagon.
@@jeffreygordon7194 oops you're right, octagon is that I meant, hexagon would work when oriented with the flat side up, the side would have an overhang angle of 30°. But even with the tip up, 60° should give you better results than a circle, 60 is still quite printable.
I think there was a DIY 3d printer video where the author made a 5-axis 3d printer that print along a non-planar contour. So even spherical surface is not an issue with minimum layer line without needing to sacrifice print speed. But that machine does require custom 3d model slicing instead just regular planar slicing. I think i saw it 8 years ago, not sure why that 5-axis design didnt spread beyond university research project to the general public.
I don't understand why FDM printers don't able to rotate they extruder around an object in the first place. It won't add much more struggle to the printers we already have and it would be an nice alternative option. Only slicers would need to get updated for this. It would allow us to use smoothing layer option that we can see for ex. in Cura, while also it would allow to make parts way stronger and lightweight at the same time as we could combine different print orientations of lines on one object. While also getting rid of support needed for many situations where the head just move diagonal. Also you may then would able to avoid that you could kick a already other print from your print surface because a detection camera or something could calculate to print around it.
Everyone commenting variable layer heights, but my take is that often there is a texture difference on one or more sides which can look worse than the low angle steps.
great content as always
Variable layer heights may help a lot while not increasing print times drastically.
Or there are options (at least in Cura that I know) to define the wall and infill layer thicknesses separately. Print the walls with 0.1mm height and the infill with 0.3mm!
in prusaslicer this is: Print settings > Infill > "Combine infill every [ 3 ] layers"
I appreciate the info you provide. I’ve been designing a box for some fans with using your methods. What methods do you use to reduce costs? The box takes 1.2kg of filament at 2 walls and 10% infill since it’s a fairly big box.
The easiest option is to use lightning infill though the part will be somewhat weak. Another option is to skeletonize the part by removing material in the cad model where it isn't needed. If you decide to skeletonize the part you can also reduce the thickness of the walls to a point where you don't use infill at all and just use a couple millimeters thick of solid material.
use cardboard for the walls and only print the mating surfaces
@@TheRealPlato well the idea is this is a product I wanna try to sell
Also use marbled filament, the pattern takes the eye off the layer lines, and makes the whole thing look better.
Some matte surface filaments make layer lines hard to see. Some glittery one will as well, probably for the same reason you mentioned marbled filament. For example, Prusa's Galaxy Black PLA prints both matte and with small glitter flecks. It can be really difficult to see layer lines when printing with that filament.
Idk why but for me those layer lines are fine, I hate visible tesselation and regular layer lines but I just don't see the problem with the top of curves. I'm 3D printing a quadcopter and there's a part of the arms where there's a top curve and I'm just fine with it.
Thank you sir obviously!
Ty for the information:)
No problem!
How do you do the noise in the last option? 🤔
How about variability layerhight like Bambu Studio? It‘s Genius!
How did you Texturize or add digital noise to your surface? I thought features such as textures are lost when put into a slicer and printed.
Changing the design that someone has put lot of effort is the last think i would attempt to do. Instead of fixing the problem at print end you always ask customer to redesing 🤣
You mention printing on its side, but I'm surprised at how often printing at any angle will remove stairstepping rather than rotating a full 90 degrees.
Big thanks for this - great discussion. Sometimes ironing is necessary to smooth flat tops - I wondered if you had thoughts on how to optimise slicer settings to achieve best surface quality when ironing?
Great video!
What about smoothing ABS with acetone?
awesome, once more
whi
hellow. Thanks for your help!! Which 3d software do you use at thiw video??
When is your mug 3d print file going to "in stock" again? The one with the regular handle says out of stock which is odd for a print file.
My favorite solution: Print in ABS and simply do acetone vapor smoothing later.
Just use adaptive layers and non planar printing for the top, perfectly smooth.
Not sure. But I think you mic is backwards bro. Depends on mic type, if its directional. Then you'll have a lot worse voice recorded quality, when the main input is on the control side.
Another solution is to setup a slicer that support non-planar printing. It is challenging but doable.
you're going to hate this answer, but you can oversize the wall by a little bit and then sand it back down, not perfect, but it's stupid simple for small batches, or for parts that need a really smooth surface
One solution is just to flatten the top. Yes, it not the greatest but good design is about figuring out the best set of compromises. And 3D printing FDM is already a comprise on aesthetics, you're making for the multitude of benefits of FDM.
Would be nice to see variable layer heights and infill with thick layers. I can see an issue with adhesion as there can be air gaps, but adaptive wall thickness can also fix that.
What if you could use thiner layers on the outter wall only? like 0.24 on the inside and 0.12 on the outter wall...
it would be nice if someone made a slicer that would do perimeters at .1 and then infill then infill at .2 or .3 I don't see why this would not work but I'm not an expert in this.
Adaptive layer height seems like the best option when you aren't able to change orientation
I think you already showed another solution - to use splines instead of circles. So where the stepover becomes too wide, you put spline with greater curvature, so it mitigates that.
Could do adaptive layer height, followed by vapor smoothing, no layer lines anywhere.❤👍
How do you apply physical surface textures in CAD?
0:52 "getting around this" I saw what you did there!!
I don’t understand why there was put a bevel on the part when printing it on the side. I looks like a perfect flat surface in the side wall
What about variable layer height?
Everything is an electrical enclosure to this guy 😂 Excellent tips though, thanks!!
would a resin 3d printer make curves a little easier