Was about to start my most ambitious 10 hour print with gyroidal infill. Chose gyroidal on the basis that it looked cool. Decided to watch your video before I began. Changed to Cubic based on your results and saved 3 hours on my print time. Thank you sir for the tip. This was S-Tier TH-cam content.
Awesome. I did 25% honeycomb on a print and immediately vowed to never do it again. Watching my M5C blasting along wobbling about 3mm side to side at 30mm a second was like the least efficient thing I’ve ever seen.
One of the patterns you thought was just in there, I actually have a specific but incredible use for. On the concentric pattern, I actually have two uses. One is purely aesthetic, on transparent filaments sometimes you don't want the infill to show and strength isn't paramount. The second use is more functional. On TPU prints, flexibility is always needed, but it can vary in how much flex is desired in one axis. I find the concentric infill makes for a very strong vertical strength but very weak lateral resistance. This is super useful if I don't want to spend hours in CAD to do the same.
very good points. I used to use lightning infill for transparent prints and it looked pretty good. but for "full" transparency I recon concentric is better indeed
I like how the video doesn't waste time with lengthy testing recordings and just talks about the results. Also, I like cubic too though I go for grid when strength doesn't matter much as it's faster.
Yeah, but that's because we already know Thomas is good at testing from previous videos. If it were a newer TH-camr, I'd definitely want to see their methods before I trust their results. All the sciency 3D printing youtubers made a lot of testing methodology mistakes early on, and they keep learning and improving.
you talked alot about the strength and efficiency of the print head movement, require to print the pattern. but i think those dubius pattern or any fractal based pattern existsted beause they are easier to callculate and able to fill iregular shapped print volume while also maintain great contact surface with the wall of the print.
i was cheating on gyroid with 3d honeycomb until orcaslicer 2.0 update made it official with the upgraded version then crosshatch came out and i left 3d honeycomb
Concentric, Hilbert curve, archimedian chords and octagram spiral are great for top/bottom infill. They usually create a better looking surface finish than the standard linear infill. Octagram spiral with a silk filament on a smooth sheet give amazing results. Hilbert curve on top infill makes for a surface comparable to a bottom layer on a textured sheet - helps to make both sides look similar in PETG.
Agreed. The only way I can get a perfect surface on production ASA parts that cannot have defects is to use hilbert curve on a powder coat bed. Saved my life. Absolutely nothing else worked.
So glad you did this in PETG. i make a line of products for the hair industry and primarily use PETG.This video has changed how I will be utilizing infill differently now. Thanks again.
A silly problem with cubic is that you can often see the infill pattern slightly through the outside in daylight and it makes the parts look cheap where with gyroid the inside looks fuzzed out.
The amount of data and research you have done on this is amazing and very valuable for many of us who 3DP. Thank you for creating this video and helping to teach us all more about infill and strength of printed parts.
Concentric and archimedian chords have some use for flexible materials when you want a more flexible infill. It still provides support for top layers, but it doesn't make the part as rigid as something like cubic would.
Another vouch for the concentric, hilbert, archemides, and octogram is that as they're completely free of any isolated pockets and can be taken advantage of when you want to fill in the void fraction of a 3d print to further strengthen them. Such as as filling them in with resin. You can't do that with a lot of the other infill options.
This is the case with gyroid as well. This also means hot air can be trapped in those pockets which creates pillowing occasionally. Good to know the different options for different use cases.
This aligns pretty well with my theories and experience. I generally used cubic wherever I wanted maximum strength and a faster to print infill where I just needed support. I also tend to rely on more shell thickness any time I want a part to be robust. Thanks for all the testing!
One advantage that Gyroid has is that it allows the print to be filled internally to add weight or possibly with foam for strength. I've printed some tall sculptures that I've added sand in the base to make them less "tippy". The closed patterns won't allow the sand to flow beyond their individual chambers. I also like Gyroid for it's multi-directional support.
@@tuliagz5329 Is the vase used for liquids or just used with dry plants or flowers (or nothing at all). Usually, you would drill a hole in the bottom of the object and use a funnel to fill the infill portion of the print - or at least partially at the bottom - then use something like a mixture of 5-minute epoxy and sand to plug the hole. You may have to roll the object around to distribute the sand or whatever you're using. If the vase is easily accessible from the top you could put the weighted media in from the top - sand or plaster or lead shot and seal it with a layer of epoxy. There are a few ways to do that, but without knowing the structure of the vase it's hard to make sure recommendations.
It's not often you can finish watching a video and immediately start using that information. Literally my very first print after watching this video was where I got to use this knowledge. With a complete enough understanding of the subject I can quickly make good choices. How I look at that dropdown has completely changed. Before this video I would "play around" with various settings. I can still play around. But now I know what to look for. I know what to choose, what to watch out for, and what to expect. This is really good stuff. Oh, and nice shirt!
He may have only been using Prusa slicer. I don't think Prusa slicer has cross hatch but I could be wrong. It would have been cool to see him try it though.
I use the cubic pattern as it's the strongest from my experience. I had printed an extender to connect 2 monitor posts 6 inch in length. While this worked initially... after about 3 months the print snapped dropping the monitor. I had used 100% infill. When I used the cubic at 50%... it did not snap it just eventually started to lean and then show failure at the same stress point as the solid peice... thanks for your tests, helps confirm what I already know
I appreciate the thoughtfulness and thoroughness of Tom's videos. I print a lot feet for air compressors, leveler feet for freezers, dish washers, workbenches, et al. My main concern is compressibility/axial loads and my default infill is gyroid.
honestly, i think gyroid is a great pic in your case since it performs quite evenly across all directions. I feel like it therefore does very well to not collapse if twisted/bent in a particular way.
I wish a slicer would implement a feature to be able to draw internal walls in parts. You can trick the software to do it, but it doesnt always work great. You could tie features together with thick walls, and use a low infill everywhere else.
Hilbert curve on a powder coat textured bed gives a flawless first layer with ABS, ASA and some others. Straight lines on parts that catch light will show through in many circumstances. Hilbert is my go-to first layer for anything that has to be perfect.
I'm a noob to 3d printing and I'm happy to learn that my intuitive choice of adaptive cubic with less infill and thicker shells is such methodically proven 😀
Awesome to see that the pattern I've emprically has gravitated towards, aligned rectiliniar, gives the "best" top surface look. Even though I only ended up using that, since it was the combination of fastest and non-crossing that I've tried. I only print for "looks" as in terrain for wargames and such.
Agree. Tests without Cross Hatch are "incomplete at best", since it combines the pros of gyroid with the higher speeds of non-gyroid infills. It has become my default infill type on all my machines because of that reason.
Orca Slicers new Cross Hitch infill pattern is GOATED. You def have to give that a try. It's a cross between Gyroid and Aligned rectilinear and its amazing.
For the multiple nozzle sizes in same print, try upping the extrusion multiplier by 1 or 2% on the nozzle that prints the infill. I find it helps with adhesion between perimeters/shells and the infill.
I’ve been playing around with infills and shell thickness, trying to figure out what cases need which one. Thank you for saving many late nights! As always great video.
Basically 90% of the video explored the "best" infill type to use. And the last 10% told us that the best infill to use is NOT to have an infill at all (and use the material for the perimeters instead), which makes pointless the first 90% of the video. Fantastic!! Actually, the only reason to have some infill is to support the top layers and resist to compressive loads (and side walls buckling). But unfortunately the infills were NOT tested to resist compression, but bending (where clearly they do not contribute significantly, being by definition around the neutral axis of the section). And theoretically the best infills to resist compression are the 2D types (certainly not cubic!). I have always used rectilinear, ~5%. Which also prints as fast as you can go and gives a clean top surface. This video confirms I was right!!
Hilbert curve is super useful in both infill and top/bottom layer patterns when a part is prone to warping. I use it to print PP or PETG-CF parts that would otherwise require an enclosure or a massive brim.
from what ive tested in a very unscientific ooga booga bend until it breaks and crush it, its a tad less strong than gyroid in the bend and quite a bit less strong in the crush test, but it prints much faster and it doesnt shake the house, so adding one extra wall with it has worked fine for mee, but it does seem less consistent than gyroid, like sometimes it wont join well with the outerwall but that could just be me
I do not own a 3d printer but I've watched a reasonable amount of 3d printing content just because one day I will own one... This info may be out dated by the time I own one but I must -MUST- watch this video because the thumbnail is just that cool.
Another problem with the self intersection of grid, triangles, and stars is that the self intersections always happen at the same X/Y coordinates (3 lines thick for triangle!) This can lead to a noticeable bump forming that can catch on the nozzle tip much worse than cubic. It can also make a REALLY nasty rattling sound as it drags along.Found this out the hard way after building my Mk3S all those years ago and the included GCode for the Nefertiti bust failed several hours in from the buildup (and probably not perfect adhesion to the build plate). Honestly it amazes me that grid is still the default pattern in PrusaSlicer (I don't know about the other forks) after all these years. As you've shown here, and as I've seen over thousands of print hours, cubic and it's direct derivatives, perform as good as or better than grid in most situations.
I think your thumbnail was one of the main reasons you got 50K views in one single day! -It's just perfect! 😃 -Anyway, I'd like to put a note up that I've used 1.2mm wall thickness for more than 2 years and I've gotten quite good results with both PLA and PETG. I'm using and old, old version of Cura and won't change anytime soon, so I'll not have the benefit of getting cubic infill (which I'd like to use). [I never see any infill-artifacts on the surfaces, because I use so thick walls - except from of course when I use translucent filaments]. Excellent video; I could watch from start to finish without falling asleep (most YT-videos make me fall asleep nowadays). 😃
Personally I use rectilinear for its clean prints and very minimal internal wiping, it also prints nicely on both bedslingers and core xy alike. One thing I find it particularly useful for is checking that bridging and general extrusion is in point when I’m switching filaments for different prints as if it is off even slightly you see it won’t connect at walls or sags in the infill and is really easy to see. Top layers are nice due to the even square finish if you want top keep top layers to around 4 instead of going higher and using more filament :)
Never messed around much and stayed cubic - I‘m fine with that and now I know why 😁. But one thing I learned: I will not push the density over 30%. I sometimes cranked up to 50% for rigid parts.
I use rectilinear and aligned rectilinear when I need to use a really high infill percentage. Also, hilbert curve has its application with about a 95-100% infill. That is when it shrinks due to cooling, the stresses from it go into the many curves in the hilbert pattern itself instead of into lone direction that cause the layers to peel up.
Yes, I prefer gyroid to cubic for ASA so I don't have straight lines passing through the entire part. But I'd like to see objective data on that choice.
Before the video I used cubic subdivision (I believed it was the more geometrically tough pattern). Now I'll still use it but with a great video to support my choice. Thanks Tom.
For thicker infill use infill line multiplier (as named in cura) so each layer is double thick. This is for strength as it add rigidity to an otherwise flexible thin wall. Half the infill and double the multiplier. Speed and material use-age are similar, but strength is better
Great video, I like how detailed all the tests are (especially with the data). Interesting on this topic would also be, how the infill change the damping/vibration/stiffness in parts. This would be great information for many mechanics (as an example when you build a printer like a Voron). Proper test would probaly be some measurements on the eigen modulus and frequency. Or are there already some tests with proper data on that (didn't find anything like that yet).
Gyroid! The strengh to weight ratio of infills was tested in 3D Printing Academys video around 5 months back. Its the reason i use Gyroid almost always. 3 walls all around and between 15%-20% infill for max strength to weight ration.
Exactly what I am preaching in discussions forever now: use rectiliniar/triangles as it is fast, efficient and provides decent strength. For more strength use more walls. And for non mechanical parts lightning can be a valid option to support top layers/internal structures. One topic that wasn't covered: using infill for aesthetic reasons, e.g. I've used hexagon infill without top/bottom layers as radiator grilles, roof racks, etc. for my 3d printed RCs
Good info, thanks! I wish slicers would offer infills designed for strength. For example, by using thick double lines to draw the familiar old patterns. Or triple or quadruple lines.
Fascinating stuff. I'm definitely going to try aligned rectilinear next time I have a purely aesthetic print with a broad, flat surface on top. It's a niche use-case, but far from an unrealistic one.
The dubious patterns just look great when they print, getting factual data about their strength is quite nice 👍I mostly use the gyroid pattern though, because most of my parts do not need much strength.
Gyroid has been my go-to for the longest time now. Great strength to weight, reasonably quick printing, and minimal external surface artifacts. Another benefit is that if you do need to fill the print, all of the airspace is contiguous, so filler materials (resin, plaster, sand etc) flows fairly easily to fill the space.
Good Video... in aerospace we use thin honeycomb and skins. The load paths are, as you suggest, only in the skin which takes all bending and direct tension/compression. The infill is only for out of plane or transverse shear but also can prevent skin buckling in a compressive or bending load. Honeycomb core has no bending strength and flexes just like you demonstrated in the video.
I believe cubic is still the pattern that produces the most fully sealed cells / air pockets, correct? It's a pretty rare use case, but this would be important for insulating properties.
Going into this video rooting for my beloved cubic subdivision, excited to see the results! Edit: Hecc yeah, let's go cubic! Glad to see you mentioned cubic subdivision as well in the end, even if it wasn't tested. If you ever return to the subject and make another vid comparing the physical attributes of the few best infills in more detail, I'd gladly watch!
Super interesting video! Recently dove into this and my takeaway was that grid was the day to day best for me with gyroid being great for looks and if you need the same strength in all 3 dimensions But sounds like I’ll have to look into cubic again!
Here is a long description of my experience probably not many people have so I want to share in case someone finds it usefull. Recently I have been printing very large molds for someone who then makes a thick fiberglass master model in them for further finishing and polishing by hand. The final serial production negative molds get then made on this polished master model. The printed molds therefore need to survive a single wet lamination process and usually get destroyed during the extraction of the very rigid master model out of them. The designer of the molds maxed out the size of my 500x500x500 Rat Rig on some of them (he designed the sectioning of the mold arond the size of my printer). Finding settings that would produce the required parameters was interesting. Requirements are smooth surface, enought stiffness and no warping. I ended up using 4 perimeters with 0.4mm nozzle and 0.2mm layer height. This gives decent surface smoothness and some wall thickness in case imperfections need to be sanded down. For most of the parts I used 4% gyroid infill. This gives enough support to even very gently curved walls to hold their shape while having the resing squeezed out of the fabric by some rollers. In areas where M6 bolts are used to join multiple sections of the mold together I used 20% infill to better resist the compressive strength of the bolt. Gyroid got picked over cubic because in my experience the more long straight lines you have in such huge print the more it wants to lift its corners off the bed (or bend the bed off the magnet) and then some sections of the mold no longer fit together. Since PETG was requested this can be a serious issue on such big prints. Not as bad as ASA would be but noticeably warps more than PLA would. Gyroid with its waves does not pull on anything as the cooling extrusions contract. And as it turns out when they need to destroy the printed mold to get the master model out then gyroid seems easier to dig out with chisels in large chunks (not sure why). One time I used lightning infill because it was by far the fastest and did not compromise the mold where it would matter. This mold was basically almost 500x500 base with something resembling a flat top mountain coming out of the base. All the functional walls wer curved in various directions relatively sharply and just the top enclosing the volume needed support to print cleanly. With 4 perimeters and lightning infill set to 20% it saved about 24 hours of printing and 1kg of filament compared to 4% gyroid. And all the curved surfaces where fiberglass was to go were still perfectly shaped and rigid enough. It is a rare case but here lightning was the best despite this beeing a technical functional print.
The Hilbert curve is actually very useful for some specific usecases. When boring into the part in 2 locations you can be sure both holes are connected and you can suck air through them, no matter where you bore or what shape the object is
Another thorough and useful test Thomas. You are a gentlemen and a scholar my friend. I have been using the Voxel PETG + for a few months now and it has been performing great on my XMax3. I was hesitant to believe the claims it works on high speed printers but it certainly does. It sands much easier than PLA as well.
The "dubious" infills are often good for translucent materials. There are some aesthetic things I make with TPU, because you can print at almost any speed without affecting clarity. When I do this, I use concentric, because the clarity is relatively even in all x/y directions. So you can adjust your walls and infill to adjust the depth.
Fantastic. This is what I have been wondering for a while as I print mainly for structural rather than decorative use. I felt that more perimeters were the better option and defaulted to 7, but I also added more infill for parts I really wanted to be very strong and often hit 70% . I think a little experimentation similar to yours is called for on the infill density impact
CNC kitchen did a pretty deep dive on this issue a while back, at least in respect to shell thickness and strength. Infill has relatively little impact on part strength even when it's set pretty high. I usually run 4-6 perimeters with as low density possible with lightning infill to save on material since infill contributes relatively little even in different use cases. The only con of lightning infill is slow print times though.
The problem with infill in regards to bending strength is that the infill contribution to the cross section's moment of inertia around the neutral axis is extremely low. Extra top and bottom layers put material where the majority of normal bending stresses occur.
Nice review. My infill is "default" because I never give it much thought. If I need a stronger part, increase the shell and infil percentage. Otherwise, I don't mess with the pattern.
From an engineering perspective a 3D print is basically a torsion box. As such, the infill should only be strong enough to prevent the surfaces from buckling before they break in tension, and not a bit more. Your perimeter-vs-infill results reflected this accurately, with the additional note that you can go pretty low before buckling becomes the dominant failure mode.
In the animation at 5:25 you can see the problem with Gyroid as it can happen that none of the surface connects with the infill if you are unlucky with your design. that leads to weak spots in terms of infill strength. I think that the crosshatch pattern gives the best compromise but it's not widely available in all slicers.
most surprising result is the quality of aligned rectilinear, very cool. There should be some setting where you can just make the last few layers of a print before the top aligned rectilinear and then the rest cubic. Overall, excellent video!
I'm ride or die for adaptive cubic! Use it for everything, and I mean everything, it's just my standard infill, mostly cosplay/display props but also helmet/armour which are like 2-3mm thick infill space, and even large miniatures.
You took a different approach to this than other youtubers, which I liked, in adjusting the infill to get the same amount of material in the part. Some of the same mistakes, in my opinion, focusing on the top line modulus of rupture. For most of my prints, I care more about the rigidity of the object under normal loads rather than the ultimate strength. More testing of different elastic moduli is needed, as well as testing of different orientations, since additive manufacturing is fundamentally non-isotropic (outside of maybe laser sintering).
In my Bachelor thesis I did a similar test except I tested for tensile strength. In my tests I also got that more shells are better than more Infill and almost no increase in strength for more than 30% infill density (normalized by weight). I tested only four patterns against each other: Grid, 3D-Honeycomb, Gyroid, Triangles. Gyroid and Grid tied for first place than triangles lastly 3D-Honeycomb (also normalized by weight).
Shell thickness vs number of layer is something I have overlook for a long time. Its is important to understand that by default, top bottom layers need to be twice as mutch as side layers.
In some sense this video may have an argument for lightning for strength! Use lightning and the lowest infill you can get away with and use more perimeters.
There is one issue I've been having with Cubic as opposed to Gyroid. And that issue is that my prints with Cubik had a lot more issues with warping and coming off the bed where my Gyroid prints had zero warping.
The re-passing of infill lines when using certain patterns has knocked down tall or top heavy prints. Rockets for example. Thank you for the information.
All that to say that you validated my intuition that shell thickness is critical to strength. I can’t think of any place where I’d use more than 10% infill. A small lever arm, for example, would be solid with enough perimeters. Infill is mostly there to support the top layers
I've been maining cubic since the first CNC Kitchen analysis video on this, adaptive cubic on anything large, gyroid for soft parts or where infill is visible.
I occasionally use patterns from the "dubious" category, typically Octogram Spiral for either the Top or Bottom (or both) surface pattern. When I do this it is purely for cosmetic reasons, they can add a nice subtle surface appearance. Hilbert and concentric also work well for this. For infill itself, on typical prints I almost always use Gyroid or Cubic at typically 15-20% along with 3 walls, 4 if I am concerned about strength.
I can see, from your data, there would be good use made of taylored infill such that it rapidly went to zero, away from the walls (shell). Just like nature designed bones' trabeculae. Worth looking up for some ideas. It supports what you said about large area, thin parts too. Great video, thanks!
Depending on the application of our print, appropriate strength should be considered. For example, if it is going to be compressed a lot or carry some load, will it experience shear or bending forces. Even if this research covers a few cases, it still was very insightful.
Before I watch: - Gyroid if I need strength from multiple sides - Honeycomb if I expect a lot of pressure in vertical direction - Line if I just need stomething to support the top layers - No Infill and if there's not much top surface area and the walls are strong enough already
Will watch this later, literally just popped in to say that's an S-tier thumbnail.
Yep, concentric and gyroid brain.
Was about to start my most ambitious 10 hour print with gyroidal infill. Chose gyroidal on the basis that it looked cool. Decided to watch your video before I began. Changed to Cubic based on your results and saved 3 hours on my print time. Thank you sir for the tip. This was S-Tier TH-cam content.
Thanks for the tip, too!
What was the print?
Cubic is just slowing down everything on both my 4 2.4 Vorons
Awesome. I did 25% honeycomb on a print and immediately vowed to never do it again. Watching my M5C blasting along wobbling about 3mm side to side at 30mm a second was like the least efficient thing I’ve ever seen.
One of the patterns you thought was just in there, I actually have a specific but incredible use for. On the concentric pattern, I actually have two uses. One is purely aesthetic, on transparent filaments sometimes you don't want the infill to show and strength isn't paramount. The second use is more functional. On TPU prints, flexibility is always needed, but it can vary in how much flex is desired in one axis. I find the concentric infill makes for a very strong vertical strength but very weak lateral resistance. This is super useful if I don't want to spend hours in CAD to do the same.
very good points. I used to use lightning infill for transparent prints and it looked pretty good. but for "full" transparency I recon concentric is better indeed
I appreciate the hours/days spend doing this massive comparative evaluation
I like how the video doesn't waste time with lengthy testing recordings and just talks about the results. Also, I like cubic too though I go for grid when strength doesn't matter much as it's faster.
CNC Kitchen wants to know your location.
Yeah, but that's because we already know Thomas is good at testing from previous videos. If it were a newer TH-camr, I'd definitely want to see their methods before I trust their results. All the sciency 3D printing youtubers made a lot of testing methodology mistakes early on, and they keep learning and improving.
Cubic was my first love...
gyroid gang
you talked alot about the strength and efficiency of the print head movement, require to print the pattern. but i think those dubius pattern or any fractal based pattern existsted beause they are easier to callculate and able to fill iregular shapped print volume while also maintain great contact surface with the wall of the print.
what about adapt cubic ?
.. and it will be my last. 🎶🎵
i was cheating on gyroid with 3d honeycomb until orcaslicer 2.0 update made it official with the upgraded version
then crosshatch came out and i left 3d honeycomb
Concentric, Hilbert curve, archimedian chords and octagram spiral are great for top/bottom infill. They usually create a better looking surface finish than the standard linear infill.
Octagram spiral with a silk filament on a smooth sheet give amazing results.
Hilbert curve on top infill makes for a surface comparable to a bottom layer on a textured sheet - helps to make both sides look similar in PETG.
Yup, fuzzy skin and hilbert curve with 0.9 flowrate on top surface makes everything fuzzy :)
concentric my beloved
Depends if you use aligned rectilinear or not.
They're good for squishy tpu
Agreed. The only way I can get a perfect surface on production ASA parts that cannot have defects is to use hilbert curve on a powder coat bed. Saved my life. Absolutely nothing else worked.
So glad you did this in PETG. i make a line of products for the hair industry and primarily use PETG.This video has changed how I will be utilizing infill differently now. Thanks again.
A silly problem with cubic is that you can often see the infill pattern slightly through the outside in daylight and it makes the parts look cheap where with gyroid the inside looks fuzzed out.
Valid point!
aint the point of 3d printing to make things you dont need (low poly pikatsus or race boats) and have them on demant at cheap?
@@TommiHonkonenyeah, but it doesn’t mean it needs to look cheap just becuase it is
@@TommiHonkonen Not me. Almost everything I print is functional, including car parts in engine bays.
Just bump up your wall count and this issue disappears.
The amount of data and research you have done on this is amazing and very valuable for many of us who 3DP. Thank you for creating this video and helping to teach us all more about infill and strength of printed parts.
Concentric and archimedian chords have some use for flexible materials when you want a more flexible infill. It still provides support for top layers, but it doesn't make the part as rigid as something like cubic would.
Another vouch for the concentric, hilbert, archemides, and octogram is that as they're completely free of any isolated pockets and can be taken advantage of when you want to fill in the void fraction of a 3d print to further strengthen them. Such as as filling them in with resin.
You can't do that with a lot of the other infill options.
This is the case with gyroid as well. This also means hot air can be trapped in those pockets which creates pillowing occasionally. Good to know the different options for different use cases.
This aligns pretty well with my theories and experience. I generally used cubic wherever I wanted maximum strength and a faster to print infill where I just needed support. I also tend to rely on more shell thickness any time I want a part to be robust. Thanks for all the testing!
One advantage that Gyroid has is that it allows the print to be filled internally to add weight or possibly with foam for strength. I've printed some tall sculptures that I've added sand in the base to make them less "tippy". The closed patterns won't allow the sand to flow beyond their individual chambers. I also like Gyroid for it's multi-directional support.
Hilbert and other open variations are even better for pumping the model full of PU resin/foam, gesso, concrete or what have you.
Do you pause the print to fill it up or how do you do it? I printed a cool tall vase for my mom but it keeps falling 😢
@@tuliagz5329 Is the vase used for liquids or just used with dry plants or flowers (or nothing at all). Usually, you would drill a hole in the bottom of the object and use a funnel to fill the infill portion of the print - or at least partially at the bottom - then use something like a mixture of 5-minute epoxy and sand to plug the hole. You may have to roll the object around to distribute the sand or whatever you're using. If the vase is easily accessible from the top you could put the weighted media in from the top - sand or plaster or lead shot and seal it with a layer of epoxy. There are a few ways to do that, but without knowing the structure of the vase it's hard to make sure recommendations.
It's not often you can finish watching a video and immediately start using that information. Literally my very first print after watching this video was where I got to use this knowledge. With a complete enough understanding of the subject I can quickly make good choices.
How I look at that dropdown has completely changed. Before this video I would "play around" with various settings. I can still play around. But now I know what to look for. I know what to choose, what to watch out for, and what to expect.
This is really good stuff. Oh, and nice shirt!
I just started 3d printing and I'm working on my first project that involves modeling and slicing. This was incredibly comprehensive, thank you!
Also what about cross hatch, was it too late to include it? Looks very promising, aligned rectilinear meets gyroid 😅
yeah, i switched from only using gyroid to only using cross hatch, with PETG, would have been a perfect thing to include
He may have only been using Prusa slicer. I don't think Prusa slicer has cross hatch but I could be wrong. It would have been cool to see him try it though.
@@oddball119 yep, but everyone should be using orca imho :)
I've also switched from gyroid to cross hatch. Makes my desk shake less. Gyroid is still my favourite to watch though 😅
@@riba2233 True. Thought Cura was the best until I used orca then I realized how confused I was.
I use the cubic pattern as it's the strongest from my experience. I had printed an extender to connect 2 monitor posts 6 inch in length. While this worked initially... after about 3 months the print snapped dropping the monitor. I had used 100% infill. When I used the cubic at 50%... it did not snap it just eventually started to lean and then show failure at the same stress point as the solid peice... thanks for your tests, helps confirm what I already know
I appreciate the thoughtfulness and thoroughness of Tom's videos. I print a lot feet for air compressors, leveler feet for freezers, dish washers, workbenches, et al. My main concern is compressibility/axial loads and my default infill is gyroid.
honestly, i think gyroid is a great pic in your case since it performs quite evenly across all directions. I feel like it therefore does very well to not collapse if twisted/bent in a particular way.
I wish a slicer would implement a feature to be able to draw internal walls in parts. You can trick the software to do it, but it doesnt always work great. You could tie features together with thick walls, and use a low infill everywhere else.
Great tests and a perfect way to present the results.
Also, I have say that your ad integrations are very smooth without being annoying - love that!
Hilbert curve on a powder coat textured bed gives a flawless first layer with ABS, ASA and some others. Straight lines on parts that catch light will show through in many circumstances. Hilbert is my go-to first layer for anything that has to be perfect.
I'm a noob to 3d printing and I'm happy to learn that my intuitive choice of adaptive cubic with less infill and thicker shells is such methodically proven 😀
Damn, I was hoping for cross-hatch and the updated 3D honeycomb to see how it performs
Awesome to see that the pattern I've emprically has gravitated towards, aligned rectiliniar, gives the "best" top surface look. Even though I only ended up using that, since it was the combination of fastest and non-crossing that I've tried. I only print for "looks" as in terrain for wargames and such.
Rectilinear my very first pick, it is Zig-Zag in Cura. Have to switch to aligned rectilinear..
What’s with cross hatch? Orca Slicer.
Agree. Tests without Cross Hatch are "incomplete at best", since it combines the pros of gyroid with the higher speeds of non-gyroid infills. It has become my default infill type on all my machines because of that reason.
Orca Slicers new Cross Hitch infill pattern is GOATED. You def have to give that a try. It's a cross between Gyroid and Aligned rectilinear and its amazing.
For the multiple nozzle sizes in same print, try upping the extrusion multiplier by 1 or 2% on the nozzle that prints the infill. I find it helps with adhesion between perimeters/shells and the infill.
I think "infill anchor " is the right parameter to use instead, to keep other geometry intact. but I don't have multi heads to test... 😅
The thumbnail is just the best. Very informative.
I’ve been playing around with infills and shell thickness, trying to figure out what cases need which one. Thank you for saving many late nights! As always great video.
I was hoping to see cross hatch versus gyroid as orca has recently made cross hatch the default in many profiles.
Basically 90% of the video explored the "best" infill type to use. And the last 10% told us that the best infill to use is NOT to have an infill at all (and use the material for the perimeters instead), which makes pointless the first 90% of the video. Fantastic!!
Actually, the only reason to have some infill is to support the top layers and resist to compressive loads (and side walls buckling). But unfortunately the infills were NOT tested to resist compression, but bending (where clearly they do not contribute significantly, being by definition around the neutral axis of the section). And theoretically the best infills to resist compression are the 2D types (certainly not cubic!).
I have always used rectilinear, ~5%. Which also prints as fast as you can go and gives a clean top surface. This video confirms I was right!!
Hilbert curve is super useful in both infill and top/bottom layer patterns when a part is prone to warping. I use it to print PP or PETG-CF parts that would otherwise require an enclosure or a massive brim.
nice tip!
Thomas, our 3D printing explorer. He goes very far away to safe us from mistakes. Thank you Thomas for your dedication 😍😍😍
Such a shame you didn't test Bambus new "Cross hatch" infill pattern as well
from what ive tested in a very unscientific ooga booga bend until it breaks and crush it, its a tad less strong than gyroid in the bend and quite a bit less strong in the crush test, but it prints much faster and it doesnt shake the house, so adding one extra wall with it has worked fine for mee, but it does seem less consistent than gyroid, like sometimes it wont join well with the outerwall but that could just be me
Or the new 3D honeycomb from Orca.
@@woodcat71803D honeycomb is included at 4:40?
I do not own a 3d printer but I've watched a reasonable amount of 3d printing content just because one day I will own one... This info may be out dated by the time I own one but I must -MUST- watch this video because the thumbnail is just that cool.
I've seen a few others do these same tests. Different methodologies for each. I'm glad the tests continue, often showing the same results.
That Prusa XL looks so cool. A lot of asthetically pleasing touches to this video. Thanks for not printing the tests in yellow and red.
Another problem with the self intersection of grid, triangles, and stars is that the self intersections always happen at the same X/Y coordinates (3 lines thick for triangle!) This can lead to a noticeable bump forming that can catch on the nozzle tip much worse than cubic. It can also make a REALLY nasty rattling sound as it drags along.Found this out the hard way after building my Mk3S all those years ago and the included GCode for the Nefertiti bust failed several hours in from the buildup (and probably not perfect adhesion to the build plate).
Honestly it amazes me that grid is still the default pattern in PrusaSlicer (I don't know about the other forks) after all these years. As you've shown here, and as I've seen over thousands of print hours, cubic and it's direct derivatives, perform as good as or better than grid in most situations.
I think your thumbnail was one of the main reasons you got 50K views in one single day! -It's just perfect! 😃
-Anyway, I'd like to put a note up that I've used 1.2mm wall thickness for more than 2 years and I've gotten quite good results with both PLA and PETG. I'm using and old, old version of Cura and won't change anytime soon, so I'll not have the benefit of getting cubic infill (which I'd like to use).
[I never see any infill-artifacts on the surfaces, because I use so thick walls - except from of course when I use translucent filaments].
Excellent video; I could watch from start to finish without falling asleep (most YT-videos make me fall asleep nowadays). 😃
Wow this must have taken so much time to produce! Congrats and thank you, I know what I’ll be going for next print!
Personally I use rectilinear for its clean prints and very minimal internal wiping, it also prints nicely on both bedslingers and core xy alike. One thing I find it particularly useful for is checking that bridging and general extrusion is in point when I’m switching filaments for different prints as if it is off even slightly you see it won’t connect at walls or sags in the infill and is really easy to see. Top layers are nice due to the even square finish if you want top keep top layers to around 4 instead of going higher and using more filament :)
Never messed around much and stayed cubic - I‘m fine with that and now I know why 😁. But one thing I learned: I will not push the density over 30%. I sometimes cranked up to 50% for rigid parts.
I use rectilinear and aligned rectilinear when I need to use a really high infill percentage.
Also, hilbert curve has its application with about a 95-100% infill. That is when it shrinks due to cooling, the stresses from it go into the many curves in the hilbert pattern itself instead of into lone direction that cause the layers to peel up.
This was impressively comprehensive. Thanks for your work!
Awesome stuff, now if only someone would test which infill pattern is best for warping 😊
Yes, I prefer gyroid to cubic for ASA so I don't have straight lines passing through the entire part. But I'd like to see objective data on that choice.
Before the video I used cubic subdivision (I believed it was the more geometrically tough pattern). Now I'll still use it but with a great video to support my choice. Thanks Tom.
For thicker infill use infill line multiplier (as named in cura) so each layer is double thick. This is for strength as it add rigidity to an otherwise flexible thin wall. Half the infill and double the multiplier. Speed and material use-age are similar, but strength is better
Great video, I like how detailed all the tests are (especially with the data). Interesting on this topic would also be, how the infill change the damping/vibration/stiffness in parts. This would be great information for many mechanics (as an example when you build a printer like a Voron). Proper test would probaly be some measurements on the eigen modulus and frequency. Or are there already some tests with proper data on that (didn't find anything like that yet).
Gyroid! The strengh to weight ratio of infills was tested in 3D Printing Academys video around 5 months back. Its the reason i use Gyroid almost always. 3 walls all around and between 15%-20% infill for max strength to weight ration.
Exactly what I am preaching in discussions forever now: use rectiliniar/triangles as it is fast, efficient and provides decent strength. For more strength use more walls. And for non mechanical parts lightning can be a valid option to support top layers/internal structures.
One topic that wasn't covered: using infill for aesthetic reasons, e.g. I've used hexagon infill without top/bottom layers as radiator grilles, roof racks, etc. for my 3d printed RCs
Thanks Thomas, for doing this tedious work more of us would have wanted to got done to get the results you showed.
Good info, thanks! I wish slicers would offer infills designed for strength. For example, by using thick double lines to draw the familiar old patterns. Or triple or quadruple lines.
Fascinating stuff. I'm definitely going to try aligned rectilinear next time I have a purely aesthetic print with a broad, flat surface on top. It's a niche use-case, but far from an unrealistic one.
The dubious patterns just look great when they print, getting factual data about their strength is quite nice 👍I mostly use the gyroid pattern though, because most of my parts do not need much strength.
Thorough and "to the point" analysis, as always. Thank you for educating me!
Gyroid has been my go-to for the longest time now. Great strength to weight, reasonably quick printing, and minimal external surface artifacts. Another benefit is that if you do need to fill the print, all of the airspace is contiguous, so filler materials (resin, plaster, sand etc) flows fairly easily to fill the space.
Hilbert curve should be the least warping infill, I would love to see it tested
That is a good test to have. Infill patterns and warping.
lets vote for an adaptive hilbert :D
Good Video... in aerospace we use thin honeycomb and skins. The load paths are, as you suggest, only in the skin which takes all bending and direct tension/compression.
The infill is only for out of plane or transverse shear but also can prevent skin buckling in a compressive or bending load.
Honeycomb core has no bending strength and flexes just like you demonstrated in the video.
I believe cubic is still the pattern that produces the most fully sealed cells / air pockets, correct? It's a pretty rare use case, but this would be important for insulating properties.
The color grading in your videos is gorgeous.
Going into this video rooting for my beloved cubic subdivision, excited to see the results!
Edit: Hecc yeah, let's go cubic! Glad to see you mentioned cubic subdivision as well in the end, even if it wasn't tested.
If you ever return to the subject and make another vid comparing the physical attributes of the few best infills in more detail, I'd gladly watch!
Super interesting video! Recently dove into this and my takeaway was that grid was the day to day best for me with gyroid being great for looks and if you need the same strength in all 3 dimensions
But sounds like I’ll have to look into cubic again!
Here is a long description of my experience probably not many people have so I want to share in case someone finds it usefull.
Recently I have been printing very large molds for someone who then makes a thick fiberglass master model in them for further finishing and polishing by hand. The final serial production negative molds get then made on this polished master model. The printed molds therefore need to survive a single wet lamination process and usually get destroyed during the extraction of the very rigid master model out of them.
The designer of the molds maxed out the size of my 500x500x500 Rat Rig on some of them (he designed the sectioning of the mold arond the size of my printer). Finding settings that would produce the required parameters was interesting. Requirements are smooth surface, enought stiffness and no warping. I ended up using 4 perimeters with 0.4mm nozzle and 0.2mm layer height. This gives decent surface smoothness and some wall thickness in case imperfections need to be sanded down. For most of the parts I used 4% gyroid infill. This gives enough support to even very gently curved walls to hold their shape while having the resing squeezed out of the fabric by some rollers. In areas where M6 bolts are used to join multiple sections of the mold together I used 20% infill to better resist the compressive strength of the bolt. Gyroid got picked over cubic because in my experience the more long straight lines you have in such huge print the more it wants to lift its corners off the bed (or bend the bed off the magnet) and then some sections of the mold no longer fit together. Since PETG was requested this can be a serious issue on such big prints. Not as bad as ASA would be but noticeably warps more than PLA would. Gyroid with its waves does not pull on anything as the cooling extrusions contract. And as it turns out when they need to destroy the printed mold to get the master model out then gyroid seems easier to dig out with chisels in large chunks (not sure why).
One time I used lightning infill because it was by far the fastest and did not compromise the mold where it would matter. This mold was basically almost 500x500 base with something resembling a flat top mountain coming out of the base. All the functional walls wer curved in various directions relatively sharply and just the top enclosing the volume needed support to print cleanly. With 4 perimeters and lightning infill set to 20% it saved about 24 hours of printing and 1kg of filament compared to 4% gyroid. And all the curved surfaces where fiberglass was to go were still perfectly shaped and rigid enough. It is a rare case but here lightning was the best despite this beeing a technical functional print.
The Hilbert curve is actually very useful for some specific usecases. When boring into the part in 2 locations you can be sure both holes are connected and you can suck air through them, no matter where you bore or what shape the object is
Looks cool with translucent filament too
Another thorough and useful test Thomas. You are a gentlemen and a scholar my friend. I have been using the Voxel PETG + for a few months now and it has been performing great on my XMax3. I was hesitant to believe the claims it works on high speed printers but it certainly does. It sands much easier than PLA as well.
The "dubious" infills are often good for translucent materials. There are some aesthetic things I make with TPU, because you can print at almost any speed without affecting clarity. When I do this, I use concentric, because the clarity is relatively even in all x/y directions. So you can adjust your walls and infill to adjust the depth.
Fantastic. This is what I have been wondering for a while as I print mainly for structural rather than decorative use. I felt that more perimeters were the better option and defaulted to 7, but I also added more infill for parts I really wanted to be very strong and often hit 70% . I think a little experimentation similar to yours is called for on the infill density impact
Thank you for your effort
CNC kitchen did a pretty deep dive on this issue a while back, at least in respect to shell thickness and strength. Infill has relatively little impact on part strength even when it's set pretty high. I usually run 4-6 perimeters with as low density possible with lightning infill to save on material since infill contributes relatively little even in different use cases. The only con of lightning infill is slow print times though.
The problem with infill in regards to bending strength is that the infill contribution to the cross section's moment of inertia around the neutral axis is extremely low. Extra top and bottom layers put material where the majority of normal bending stresses occur.
Nice review. My infill is "default" because I never give it much thought. If I need a stronger part, increase the shell and infil percentage. Otherwise, I don't mess with the pattern.
Nice to see the 3D Cubic Infill I favor being the best choice 😁
My favorite infill for 90% of my prints is the new Cross Hatch infill! Its faster than Gyroid but has similar strength.
I think these results are similar to those of CNC Kitchen years ago.
Always nice to see two different tests show similar results.
I have been using cubic since the beginning for the majority of my projects. Just made sense to me.
I never tried the cubic, but I love the gyroid. Its wavy shape saved many complex prints-it allowed the building of difficult bridges.
From an engineering perspective a 3D print is basically a torsion box. As such, the infill should only be strong enough to prevent the surfaces from buckling before they break in tension, and not a bit more. Your perimeter-vs-infill results reflected this accurately, with the additional note that you can go pretty low before buckling becomes the dominant failure mode.
In the animation at 5:25 you can see the problem with Gyroid as it can happen that none of the surface connects with the infill if you are unlucky with your design. that leads to weak spots in terms of infill strength.
I think that the crosshatch pattern gives the best compromise but it's not widely available in all slicers.
most surprising result is the quality of aligned rectilinear, very cool. There should be some setting where you can just make the last few layers of a print before the top aligned rectilinear and then the rest cubic. Overall, excellent video!
I was really hoping to see the two different nozzle sizes! Great video
I like using cubic subdivision in Cura, it becomes more dense of a pattern in narrower parts and less dense when filling large gaps
I have been happy with cubic and now I have some justification from your testing. Thank you.
I'm ride or die for adaptive cubic! Use it for everything, and I mean everything, it's just my standard infill, mostly cosplay/display props but also helmet/armour which are like 2-3mm thick infill space, and even large miniatures.
You took a different approach to this than other youtubers, which I liked, in adjusting the infill to get the same amount of material in the part. Some of the same mistakes, in my opinion, focusing on the top line modulus of rupture. For most of my prints, I care more about the rigidity of the object under normal loads rather than the ultimate strength. More testing of different elastic moduli is needed, as well as testing of different orientations, since additive manufacturing is fundamentally non-isotropic (outside of maybe laser sintering).
I didn't expect cubic to be the winner but now im gonna start using that, especially because it prints faster.
In my Bachelor thesis I did a similar test except I tested for tensile strength. In my tests I also got that more shells are better than more Infill and almost no increase in strength for more than 30% infill density (normalized by weight). I tested only four patterns against each other: Grid, 3D-Honeycomb, Gyroid, Triangles.
Gyroid and Grid tied for first place than triangles lastly 3D-Honeycomb (also normalized by weight).
Shell thickness vs number of layer is something I have overlook for a long time. Its is important to understand that by default, top bottom layers need to be twice as mutch as side layers.
I use a variation of infill patterns especially for flexible materials to get the right properties
best thumbnail ever in 3d printing history
In some sense this video may have an argument for lightning for strength! Use lightning and the lowest infill you can get away with and use more perimeters.
There is one issue I've been having with Cubic as opposed to Gyroid. And that issue is that my prints with Cubik had a lot more issues with warping and coming off the bed where my Gyroid prints had zero warping.
The re-passing of infill lines when using certain patterns has knocked down tall or top heavy prints. Rockets for example. Thank you for the information.
All that to say that you validated my intuition that shell thickness is critical to strength. I can’t think of any place where I’d use more than 10% infill. A small lever arm, for example, would be solid with enough perimeters. Infill is mostly there to support the top layers
I usually go for simple grid. I think it's the strongest for compression forces in the vertical direction.
I've been maining cubic since the first CNC Kitchen analysis video on this, adaptive cubic on anything large, gyroid for soft parts or where infill is visible.
I occasionally use patterns from the "dubious" category, typically Octogram Spiral for either the Top or Bottom (or both) surface pattern. When I do this it is purely for cosmetic reasons, they can add a nice subtle surface appearance. Hilbert and concentric also work well for this.
For infill itself, on typical prints I almost always use Gyroid or Cubic at typically 15-20% along with 3 walls, 4 if I am concerned about strength.
Fantastic sanity check. Thanks for your work!
I can see, from your data, there would be good use made of taylored infill such that it rapidly went to zero, away from the walls (shell). Just like nature designed bones' trabeculae. Worth looking up for some ideas. It supports what you said about large area, thin parts too.
Great video, thanks!
Depending on the application of our print, appropriate strength should be considered. For example, if it is going to be compressed a lot or carry some load, will it experience shear or bending forces.
Even if this research covers a few cases, it still was very insightful.
Before I watch:
- Gyroid if I need strength from multiple sides
- Honeycomb if I expect a lot of pressure in vertical direction
- Line if I just need stomething to support the top layers
- No Infill and if there's not much top surface area and the walls are strong enough already