Now that I’m printing larger, longer prints (36 hours in some cases) with the H2D I’ve discovered the drastic impact that infill can have on print length.
One print is 1 day 18 hours using gyros (my typical) and only 25 hours with rectilinear. My mind was blown.
I have been printing gyroid to minimize the print lines crossing and the print getting knocked loose. I am not sure what the best compromise is between time, strength, and safely finishing the print.
I made this thread specifically in H2D to see if anyone has new opinions given the machine, or like myself is now learning about the large print world.
Note: For now I’m finishing prints using gyroid; strength is important for these prints since they are for cosplay and may get knocked around a bit.
This has been on my mind as I get into larger prints. I tend to use crosshatch, which can take a bit of time over rectilinear too.
Uhh, wouldn’t rectilinear be dicey? doesn’t it cross it’s own path? So on tall thin prints I’d be prone to knocking stuff over or out of whack. I’ve had crosshatch as my default for awhile now, so forget how all the infills compare exactly.
I’ve had issues with tall thin prints. The quality degrades some as it gets towards the top, and the quick infill printing is a portion of that problem. I’ve considered making custom supports to help stabilize the pieces.
I guess that makes sense, with longer lines in grid/rectilinear the printer can utilize its high acceleration, whereas with something like Gyroid it doesn’t have those long straight stretches to speed up on
I’ll try it out. Seems to take slightly less time than gyroid so I’ll give it a go.
I’ve decided to go grid for prints less than 500 layers and gyroid for over that for now. It’s too much of a pita to reprint these 1kg models to save 12 hours.
The path it prints crosses with itself, not just travel. It needs to put the filament down on top of itself where the lines cross. Z-hop would work for travel collisions as a band-aid but not for printing
Odd… in every test I’ve done for myself, I don’t see these attributes. I always find Gyroid relatively weak for the z compression. Maybe I’m not testing the right geometries.
Gyroid infill can feel weak under Z-compression for a few practical reasons:
Specimen geometry: Thin walls or non-standard test pieces let its slender struts buckle more easily.
Low infill density: Under ~20 %, the curves form thin beams unsupported by horizontal layers—bump density to 30–40 % or widen extrusion for better Z strength.
Layer bonding: Gyroid’s continuous path needs rock-solid interlayer adhesion; any under-extrusion or uneven cooling shows up under compression.
Relative comparison: A grid or cubic pattern carries pure Z loads with straight horizontal beams, so they often outperform gyroid in simple compression tests.
To really see gyroid’s isotropy, use standardized cubes/cylinders, test at multiple densities, and print samples at 0°, 45° and 90° orientations.
Still, it’s squirting out molten plastic and colliding with a very thin layer of recently molten, presumably soft plastic, so I’m surprised that that generates enough force to knock over a print.
First ask yourself, “do I really need all the strength for this print?”
If yes, by all means go for gyroid or your favorite strength infill. But you should also be increasing the wall count, the top & bottom surface count, and increasing the infill line width 0.2 over the size of the nozzle size.
If no, aligned rectilinear is king (IMO). It’s quick, quiet, keeps layer times even, & doesn’t intersect.
I always run my infill line width 0.2mm over nozzle size anyway. So with a 0.4 nozzle it’s set to 0.6. With a 0.6 nozzle it’s set to 0.8. I don’t exactly do this for the strength, but more so for reliability along side slowing infill print speed down to 80mm/s and disabling reduce infill retraction.
I’m not exactly worried about speed. I’ll gladly allow a print to take twice as long if it guarantees better quality prints and print process reliability.
Honestly, I would assume the problems people have with the “crossing” paths is more related to their situations. My guess, wet filament. Poor extrusion coupled to crossing paths, equaling bad and failed prints. Since PLA is so tolerant to moisture, people tend to not worry about it. But there are points where it could come back and haunt them, like the guess I just made. Obviously, I can’t back it up or confirm, but I tend to think if I NEVER run into that problem and exclusively use crossing patterns, the problem is not as simple as a crossing pattern.
I also have never run into any issues with rectilinear infill. It’s been my go-to for both aesthetic and functional prototypes for years. I predominantly print with PLA and PETG though, so I can’t speak to all the high temp, engineering grade materials.
What @L0rdS474n said. Couldnt have said it better myself.
Side note: Grid was the go to infill until the printers got fast. They were so slow before, that they had time to melt the overlapping sections as it passed over them. Now the printers just smash through the high points.
What I do for consistent and reliable infill is lower the infill speed to 80mm/s, increase line width 0.2mm over nozzle size, and never pick an infill that intersects on the same layer.
I’ve grown fond of rectilinear, aligned rectilinear, and gyroid. Both of the rectilinear infills keep layer times consistent for basic models, which keeps the outer wall quality consistent and free of visual defects.
This has never failed me in terms of reliability and quality.
I know it’s been awhile but I wanted to chime in since I didn’t see this in the comments. Grid and rectilinear create the same pattern but rectilinear doesn’t crossover any lines. Rectilinear prints one direction on a layer and then prints the opposite direction on the next layer so the nozzle never bumps any printed lines.
Grid can have a very slight increase in strength over rectilinear but when printing that pattern infill, I always use rectilinear because it’s faster, doesn’t hit any printed lines and doesn’t add to nozzle wear over time like grid does
