In this week I modelled and 3D printed a russian doll type of a creation. As for the weekly assignment, I 3D printed a braid which would be quite difficult to create with subtractive methods.
The printer I used was Prusa M4KS and as software for slicing the model I used PrusaSlicer 2.9.4 So I began by downloading the software, opened it and went through the initial set up.
These are the printers that I chose for now to use on my computer. Additionally, a SLA printer was tested on another computer.
Clicked None on the right and then selected which printers I expect to use. Following instructions I selected Creality, Prusa, and Ultimaker, and selected a few printers from those manufacturers, specifically Prusa M4KS which I was going to use now. For filament I selected generic PLA and Generic PTEG. These settings can be edited in the drop down menus on the right -> Add/Remove filaments/printers
I wanted to print a phone stand to help in filming my documentation with my phone (freeing my both hands for demonstrating), so I searched for a one in Thingiverse and downloaded this one which was small enough to be printed quickly enough, 22 minutes to be exact (preference for this task was 10min).
I opened the file with Prusa Slicer, and clicked Slice. Slicing means that the 3D model is sliced into many “2D” -layers that are stacked on top of each other and have X and Y coordinates, which the printer understand. Then I examined how the print is going to be printed with the sliders, one layer on horizontal slider and multiple layers on the vertical slider (in video below I grabbed a wrong handle at first and moved up only one layer).
Then I clicked Export G—code at the bottom right and saved it on a flash drive that I had taken from the 3D printer, and then plugged it back in the printer.
Before starting to print, I realised that the filament was running out, and received some guidance from a class mate on how to change it. It was quite straightforward: in the main menu press Filament and then Change filament (alternatively first Unload and after that Load). Then wait for the nozzle to heat up so the filament can be removed by pulling it out. Then we added new roll of PLA filament above the printer, cut the twisted parts away and pushed the head of the filament in the nozzle. It didn’t go too far in it to be enough. Chose PLA and affirmed the load. The nozzle purged a bit of the new filament, and it is good idea to purge a bit more to ensure the readiness for the print.
Then I was ready to go on with the printing... Pressed Print in the main menu and selected my file from the list, and then pressed Print. Printed started to print and 25 minutes later I came to see the outcome which was a success and works as it’s supposed to.
First we did overhang test, which I downloaded from this week's lecture material and imported on PrusaSlicer.
In the settings on the right I chose the speed, material, the printer and selected None for supports (as this is a test for not using supports). This also raised a warning about print stability (second image).
Then I exported the G-code in to the Flash stick, sticked it into the printer and pressed print. I did the same procdeure for the bridging test (downloaded from the lecture materials)
Examining the construction of the print with the sliders, there can be seen on different color the specific parts of the print. On the second image in blue is the actual subject of these tests: the overhang perimeter.
And here are the printing and results for these tests. The bridging has surprisingly clean outcome.
Then we did a The jolly 3D printing torture-test by CreativeTools.se, which was done with Prusa MK4S MMU to allow multiple colors (many rolls of filament on the image below).
So I downloaded the files and opened with PrusaSlicer those files that had multi-part and single in their name. I changed the unit of measure from inches to millimiters by clicking yes to the prompt and accepted that the files are considered as one object consisting of multiple parts. I removed the extra part that opened along with the files. I chose the right printer from the drop down menu and selected the colors in the software to correspond the color in the printer.
Sliced and imported the G-code to a flash drive which I then sticked to the printer. As in the image below, I was able to see the colors in their place and a model of the upcoming print. Then I hit print and left it to materialize for about 7 hours, and came to collect it after the weekend.
In the last image there can be seen the block that the printer created as every time it changed a color/filament, it sprouted the previous color out of the tubes/nozzle. It might have changed colors multiple times per layer so there's a lot of extra PLA being used in the printing process.
Stereolithography allows a fascinating way for additive manufacturing.
We did a wall thickness test with the SLA printer and here's the process described. The Printer was Formlabs From 3 SLA 3D Printer. Gloves need to be used when in contact with the machine and its parts. The machine works with Wifi connection and here we have their IP-addresses printed in the front of each machine which need to be selected from the computer. Check that there
The first thing was to open the ventilation hole at the back of the printer.
We selected the material as white in Formlabs menu and opened the file. The model was rotated so that there's a minimal small part of it touching the platform and proper support were selectd on the Supports tab. Similar ooutcome was also achieved with One-Click Setup (last image)
As the supports were in place we were ready to start printing.
As printing was finished (after the weekend), became the time for post-processing which is more demanding than with FFF (if done any). Begin by taking gloves (this is more important than caring about creating waste). You don’t want the liquid resin to touch anything non-disposable. Then close the ventilation above the cartridge on the top back of the printer.
Then open the orange lid, lift the handle above the platform to release the lock, hold the handle around the lock handle and pull the platform out while as soon as possible turn around thew platform to avoid splashing any of the liquid resin on the platform or the print.
Take the terminal to a place for cleaning it up properly. Here we had a specific holder for the platform.
Use a tool to detach the print on a paper and put it in the grill of the cleaning machine, where the print soakes in isopropyl alcohol, and then choose the time for soak (10min ok) and press start. After ten minutes the print pops out and is still left to dry up for about 15 minutes. Let the print dry up so the surface finishes nicely. There is still some liquid resin (although diluted) on the print so do not touch it by hand.
Then clean up the printer platform. Spill generously isopropyl alcohol on the platform, swipe it up, and repeat until the whole platform is clean from resin all around. Again do not worry for creating too much waste but use as much paper as needed. Then you can put the platform back in the printer, press down the lock handle, and close the lid. The machine asks if the print came out clean: press yes and proceed to the main menu.
Then it was time for curing the print. The isopropyl has removed most of the liquid resin from the surface, but there is still some left, probably diluted in the isopropyl, so take new gloves (or use tweezers) and move the print to the curing station. Curing solidifies all the resin with light and heat. We let the support system of the print in its place during the curing, but it could also be removed at this point to enhance the final surface, but that would be more hazardous regarding the resin and removal of the supports. We set the temperature to 60 degrees and time for 30 minutes. These values vary between printing materials and can be checked in the formlabs website. Then pressed start (and began documenting all this).
And here's the final print after the curing. Three of the smallest holes are stuffed with the resin so smallest width you can use with this printer is 0.4mm. Two of the smallest "walls" are quite loose and flex easily if touched but do not break down.
The tests showed the rather obvious fact that I cannot print in the air without support systems. However, it was surprisingly possible: PLA material is quite sticky and it was possible to build relatively long bridges without very visible malformation, as seen in the video above. This was not the case with overhangs, the material quite soon starts to sag down. Also the support system are not always very optimal as they are somewhat tedious to remove and leaves small marks on the print that need post-processing. There is this video on ways to go around the need for supports. The surface with FFF/PLA is not very clean in any case. Also the tests with SLA printer showed that making thin holes is much more difficult than making "walls" of the same thickness. The material, even with SLA was sticky enough to fill up small slots and holes.
In addition to these test learnings, I learned the processes and workflows of working with three different types of printers (as documented above) well enough to use them independently in the future
We scanned someone's brain left in the lab. This was done using Artec Leo 3D scanner with a price tag of modest 35k! (compared to for example Creality Ferret that costs few hundreds). Artec Leo has a flash, two cameras and runs a computer vision algorithms. It can scan from 0.1mm up to 30m object sizes. In addition to the scanner, the accompanying software is also relevant when choosing the scanner.
We selected a new project and set resolution to normal. Then just hit the trigger once. The scanner was kept on unchanging distance as the it was moved around the brain, so that the image on the screen turned to green, meaning that that particular spot was properly scanned. Then hitting the trigger again finishes the scanning.
As the whole brain on the screen was turned to green, we plugged the scanner into a computer with ethernet cable and named the new scan on the device.
The device was made into a DHCP server by going into Settings -> Connect to Leo -> Wired (DHCP Server): Enable, and ultimately ethernet was being enabled (there was some problems that I wasn't able to fully follow).
Then we opened Artec Studio 16, imported a new project and there was the model of the brain on the computer screen.
Then we began editing the model. First we erased onnecessray parts with the eraser tool. We chose base selection and painted the base (with Ctrl + left click) so that the software was able to recognize it and remove it
Applied tools such as outlier removal (run global), fast fusion, small objects filter, hole fitting , isometric refresh, fixing the holes (press Tab and select holes to fill them), and mesh simplification (more polygons, more detail, 500 good enough). Also the texture was finished here.
As the model was complete, we imported it as a mesh which was then opened with Blender. .Stl-files are good for 3D printing but .obj-files also include the texture for processing that can handle it.
We played with my child in the weekend with traditional russian doll (matryoshka), where I got the idea for printing an object that couldn't be laser cut. After thinking about creative angles for approaching a creation of a russian doll, I asked Leo Ai what creatures have layers but didn't lead to far. Then I asked more straight question of what funny ways there are to create russian dolls and there was a couple of interesting ones. First idea was to create a Russian doll of Russian presidents/leaders since Lenin to represent the Russian/Soviet political history that has been built on the previous one. Those are for sale on the internet and likely in Russia as well but the faces of the presidents are printed on the dolls.
Another idea was creating russian dolls out of babushka cat, a meme of a cat with a scarf on it's head looking cute/funny.
The point is to try to print them already in their nested positions so that it wouldn't be possible to make the subractively.
Now later as I look at the Fabacademy assessment page, this does not qualify as a design that can't be made subtractively as it can be disassembled. :/
I need to begin to check the assessment page before starting anything. This far I've remembered it only until I've finalized documentation and I need to check if I've documented everything expected. (The fabacademy website still confuses me a lot and I find it difficult to navigate without getting overwhelmed by all the links and paths that are just listed there and there's is so much and totally new things to cover in any case.)
To try a shortcut with designing the model for print, and inspired by the AI recitation, I decided to try out connecting Claude to Blender using MCP and create my rather complex idea for a print with that. I followed Pablo's great documentation and a little bit of this medium article. Creating the connection between Blender and Claude was surprisingly simple.
For the first idea, Claude refused to print Vladimir Putin, or any real public figure whatsoever as it could be used in misleading or disrespectful ways. I had also thought about the political correctedness of the idea so I didn't mind leaving the idea. After seeing the latter idea coming into (digital) reality, this would have been way too difficult task for Claude. I'm not sure if it is feasible at all.
I did multiple iterations in Claude-Blender-MCP style and took a few screenshots. It took Claude a few minutes to make the models, of which none of them wouldn't have really worked as a matryoshka, usually because of the thin connection between head-torso. It wasn't able to fix it although prompted for doing so. I wish I had saved each iteration in separate files as there was on somewhat potential design.
This one was probably the first design after a prompt "Can you create me a design of a Russian doll that resembles a babushka cat (as the meme goes) with five nested layers?" On the left Claude lists what has been included in the design.
This was the second one as I asked for larger ears and more cat like face. Also a fit-snap mechanism for the lower and upper parts of the body. I also asked for an alternative design where the scarf covers the whole head. This one was most close to fit as nested inside each other.
Here I asked to make it resemble more of a actual matryoshka and put the parts inside each other and the outcome was this.
After spending some time with this, mostly waiting for Claude to finish, I decided that I wont probably get a design to 3D print. The process might have also failed because of my prompting: this idea was rather difficult to describe and put into prompts.
In the next day I realized I hadn't had the "Use assets from Poly Haven" checked in Blended's MCP-menu which would probably make a change in the quality. I didn't know what that is but now I know that it is a public 3D asset library. I tried checking that box and try again. Claude was checking and searching Poly Haven assets for 10-15 minutes and finally there was a message that "Claude's response could not be fully generated" and I wanted to to continue creating the 3D model to print.
I searched youtube for a Blender tutorial for how to make a russian doll and found this one by PIXXO 3D and began working on that. Maybe I can add cat ears later to have at least a slight twist to the original one.
So I created a cylinder by pressing Shift + A -> Mesh -> Cylinder and downloaded the model jpg for the doll from here as instructed. I wasn't able to add the image just by dragging, as it created only a single-plane image, so I added it from Shift + A -> Image -> Reference and selected the right image. Now I was able to move around in "space" and the image would not disappear. I pressed G to move it to the right spot and scaled it up to approximately match the cylinder by pressing S. I didn't add plywood texture because I wasn't making an animation as in the tutorial.
I toggled the X-ray view so that the cylinder became transparent. Then I selected the top points in the cylinder and moved (G) them slightly down and selected the bottom points and scaled (S) them to match the image.
Still the bottom points selected, pressed E to extrude and Z to allow horizontal movement. Made it again match the image. Then had the mouse there on the recent extrusion and pressed Cmd + R and scrolled the mouse to create two more lines, double right-clicked them and pressed S and Z to scale them further away from each other, and S to scale the middle of the pedestal out.
Then moved mouse over the body and with Cmd + R added more lines which I slightly scaled (S). Then continued up from the middle with extrusion (E) and scaling (S) until the whole image was "covered".
Then toggled the x-ray mode off, switched to object mode, and by right-clicking the object selected Shade smooth to smoothen up the (horixzontal) edges.
Back in edit mode, selected the middle edge and with Cmd + B I duplicated it and with X and selecting faces I deleted the space between them. Then from Modifiers (the wrench icon) on the left -> Add Modifiers -> Generate -> Solidify. Adjusted thickness aptly.
At this point I realized that my matryoshka was 4.8 meter tall and I went to adjust the dimensions to the Object mode and pressed N. I set the height to 0.1m (10cm), which I then canceled as the doll became too small to operate with.
In the object mode I clicked Apply in the modifier tab. Then pressed Z and chose solid to see the object in two parts. In edit mode, selected one vertice and pressed Cmd + L to select all of them, then G to select all of them in the upper part. Then using Cmd + R and clicking on the thin bottom area I splitted it in two and extruded (E) the inner half. I did the same for the bottom part except extruded the outer area.
Then having one doll ready I duplicated it (Cmd + D) and scaled (S) to 75%. Then did the same for the smaller one and repeated two more times to get 4 layers of dolls. The first image is the first version and the second was modified after visiting prusaSlicer with the first one.
Then I exported the file as STL, opened PrusaSlicer and imported the file there. It appeared there really tiny so I scaled it up in PrusaSlicer by 1500% for the height to be about 7.2cm. I also set support settings to everywhere to allow Prusa to do what is necessary to print this nested thing.
I realized that there is not likely enough space between the upper and lower parts, so I went back to Blender, Cmd + Z back to the only one "doll", moved (G) the upper part slightly up, and did the duplications, exprting to STL and importing to PrusaSlicer. Now it imported larger so I scaled it down to 6,2cm and sliced it.
I asked Kris to check the print before beginning it, and it appeared that there should be support structures visible in PrusaSlicer so we went to Print Settings -> Support material tab -> in Style selected Organic, and now the support structure was visible. We also made it back to 100% size so height around 12cm. I hadn't really thought about that 2cm size of the smallest doll was really small, especially regarding the gap between upper and bottom parts.
Here's the difference between no support and support. The supports ended up being over 30% of the used material.
Here are the final settings and proportions of the used material.
The print time was estimated to be 3h50min. Looking at the early print, I began to doubt if I can break the supports without breaking the whole thing. I probably should have made the dolls thicker, I had just adjusted the thickness roughly without thinking about it.
And as I thought, the print didn't as supposed to. What I didin't expect that the printer won't finish the print, because the largest top part detached from the whole and began wobbling around the nozzle,ruining the print. Also the assembly didn't work as the dolls were too thin and the snap-fit mechanism even thinner. I feel a bit stupid not thinking it through properly before starting the printing (I was busy getting this done before the regional review and lecture). These videos show what went wrong and how.
Now, with more time I've come back to fix the issues in this assignment. First, I want to make the doll work as it's supposed to work. Then, I also want to print a piece that is not possible to be created subtractively. For the matryoshka doll that's easy if all the pieces are done separately. I began by fixing the doll and then do the latter task.
I made a tolerance test for the printer I'm using to know how much the PLA widens before it hardens. I modeled the test with FreeCAD. I added (too thin) numbers with Shapestring tool in Draft workbench to indicate the difference, and connected everything with boolean union. To make Pads out of the numbers, the compoundoing had to allowed in their body data tab.
I found out the best fit is gained with 0.15mm tolerance.
I made the doll from the start with the same method as described before. I had an issue that the whole cylinder was scaling when I was trying to scale only one line of faces. I asked Gemini what the issuea about and it was solved by clicking off the Proportional Editing on the top row.
Another problem I ran was the measures. I got confused about how can the whole thing be 7.7 cm tall but 20cm thick but still look like this.
Asked Gemini and this was solved by sleecting all parts in Object mode and pressing Cmd + A and selecting All Transforms. Now the scale was the same across the parts and the scales for X, Y and Z became 1.000.
Then I applied the tolerance test to make the top and bottom halves fit each other well. I figured this Edge Slide-value is 0 in the middle and -1 or 1 in the edges. Thickness is 5mm and half of it (2.5mm) represents 1 and -1 on Edge Slide. So I calculated 0.15mm tolerance to mean 0.15/2.5 = 0.06. An Edge Slide value 0.06 should be equal to 0.15mm.
Now I had the pieces ready to be printed. I exported the blender file as an stl-file and imported it to PrusaSlicer. In the setting I set the Autogenerated Supports.
The supports were now supporting each piece from bottom to top and the print was a success.
After the supports were stripped, I found out that the dolls did not have bottoms. They had been not fully attached to the main bodies and had been stripped with the supports. Also, the end result wasn’t a good fit, and the reason I suppose is that both the top and bottom parts have the extra 0.15mm after the print and I had taken into account only one extra 0.15mm. This feels like an obvious mistake. Also, I had scaled the inner parts down from the outermost after solidifying the piece, so the tolerance were also scaled down by 75%, making the inner parts fit slightly better. Thus, this experiment was a “failure” as well. However, the supports were strong enough to keep the print attached over the printing. Closing the bottom and fixing the width of the top and bottom pieces should fix these errors.
The thin bottom looked like this in Blender before the fix. I scaled to bottom edges and typed 0 make them connect in the middle. But then trying to fix the actual issue, the solidify-modifier created these type of odd solutions. It took almost half an hour to make a simple solid bottom and I'm not really sure what exactly solved the thing in the end.
Then I continued to fix the fit-issue. Now I added the tolerance of 0.15mm (Edge Slide: +-0.06) on both the top and bottom parts and extruded the opposing sides for the parts.
After copy-pasting the doll and scaling the copy down by 75%, and repeating it once again, I realized I had made a mistake with the solidifying midifier and fitting the pieces. Now the thickness and the tolerances became scaled down as well. Instead I should have solidified each doll separately. I didn't want tto start again from the start, so I just prompted Claude to make the thickness the same for all the dolls. The prompt: "Despite the scaled down size, I'd like you to make the solidifier modifier to be the same as the largest cylinders (5mm) and the extrusions that are vertical between the pairs to be the same width as in the largest cylinders. So edit only the four of the smaller ones to be the same thickness as the largest cylinders."
I also did one more addition with Claude to create more meaning into the nesting dolls. I prompted Claude: "Great. Can you now make simple relief sculpture type of faces to each of them? The largest should have a raging face, the middle should have very terrified face and the smallest should have very sad face and a broken heart (vertically) in the middle of the doll." The result looked like this.
But this time the fit was too loose and I don't think I should use anymore PLA for this thing. It was a mistake compensating both of the pieces by 0.15mm. I wonder why the fit was too tight in the first try because that should have done the job right. There has likely been another mistake there then.
Instead of printing more dolls, I decided to test the fit with a few simpler pieces. I began by setting the units to fit the scale that I’m using: on Scene Properties I set unit scale to 0.001 and set length to millimeters. I created on cylinder which I scaled right and Cmd + A > All transform to have the scales in line. I deleted the top face of the cylinder, added a Solidify modifier and set thickness to 5mm (as in the dolls before). I prepared additional bevels to the sides.
Then I copy-pasted 8 of them and started creating different sizes for testing tolerances. However, I came to think that maybe I had applied 0.06 edge slide to the doll with too tight fit but I should have applied -0.06. In that case I would have compensated the tolerance to wrong direction. So I decided to try with only two pieces with a correct tolerance correction: outer rim extrusion is in the middle and the inner one has -0.06 edge slide, translating into 0.15mm tolarence to correct direction.
The first print failed, although the pieces layd down on the build plate, which was weird. I cleaned the build plate with Isopropyl Alcohol and I added 5mm brims to the pieces. Then print came out fine but the fit was a little bit too loose, so I quess I had done it correctly previously.
I made a new piece with 0.14mm tolerance (Edge slide -0.056) and same printer preferences.
However, I forgot to "Shade smooth" and having not-exactly-round pieces affects a little to the fit as it depends on the angle the pieces are attached.
For some reason the printer printed a previous print although I had selected the correct one. I went to delete the previous prints from the USB-stick and then it was able to select the correct file.
Then the print failed again, so I canceled and reprinted. And it happened again at the same spot. In the image below there's the previous failure on the front. This makes me doubt there's something going on with the printer. It looks like the nozzle collides with the print at some point. There is nothing visible on the build plate, though.
I will leave this experiment here as there's not much to gain anymore.
I was working on this at the same time as trying to fix the nesting dolls. My first idea was to create 3D knot demonstrations as they usually are in 2D. For the purpose of creating these models I used Claude-Blender MCP. As the connection was established and Blender updated, I prompted Claude: "I would like you to make a rope with 1cm diameter and make it into a bowtie with at least 2cm space between the rope, i.e the rope not touching itself at any point." and then clarifying that I meant bow knot, but it didn't make it correctly.
Then I gave it this link of multiple knots and prompted: "Can you make me more similar knots, but only 1cm space in between the ropes (without touching itself): numbers 1,3,4,7, and 15 from this website" and now the results were more plausible.
However, I came to think of a braid instead and wanted to continue with that idea further. I prompted "Can you make me a braid from five ropes of different colors, 2cm wide each and 1cm space in between the ropes?" and the result was excellent.
Then there were multiple prompts for testing with textures and such
I insructed first to have maximum 45 degree angles but the ropes didn’t intertwine very much after that, so I reinstructed to maximum of 65deg angle (horizontally). My original plan was to try to print it with only a raft and no other supports. An angle test I found at the lab shows that even 20 degree angle (vertically, 70 degrees horizontally) can be possible without failure.
I continued testing composition and angles with Claude. Then I proceeded with this one.
Now I needed to export them as different objects to have them printed in different colors. I asked Claude to make them separate objects. But I didn’t have option to export 3mf-files so I needed to add an extension to do that in Blender: Edit > Preferences > Get Extensions and searched for 3mf.
Now 3mf export was available which I then opened on PrusaSlicer. It didn’t matter to me which rope was which color so I didn’t adjust any colors. But I adjusted the speed to be a bit slower on the perimeters, from 250mm/s to 150mm/s. I tested with different support options on PrusaSlicer. I adjusted the overhang treshold to 25 degrees and set 10 raft layers with 20mm expansion to have enough support on the build plate.
I went back to ask Claude if it could bring the ropes closer to reduce the overhangs and re-exported the result.
I added a new PLA to the multicolor printer by pushing PLA filament to the tube. Before printing I got a warning that the G-code is for five PLAs but two of the loaded filaments are PETG, which has higher melting point. I go back to PrusaSlicer to set two colors as PETG. On PrusaSlicer, I got a warning as in the image below, telling me that they require different temperatures. I asked Gemini what to set to those settings, and it said among other that PETG and PLA do not bond to each other, which I figure makes the support system not working for these materials. I decided to replace the PETG-rolls with PLA-rolls.
The print was estimated to take over 30 hours and there was a lot of switching between the colors. Each color was basically printed one level for under 1 cm^2 area and then the color needed to be changed. This created a lot of waste and process-wise wasn’t very well though out. The print was paused a couple of times because the filaments were not spotted by the sensors. Then at one time it couldn’t be fixed easily without aborting the print and there was only one fifth of the print done at this point. So I decided to use only two colors, use another printer with wider, 6mm, nozzle, and increase the line height for the print (which I learned is possible after I had started the previous print).
I added a printer in PrusaSlice: Printers-tab and in the drop down menu select Add/Remove printers. I clicked next until I was able to select the printer I was going to use: Prusa XL 5T Input Shaper and selected the nozzle sizes of 0.6mm and 0.8mm. Later I added also 0.4mm nozzle.
Right-clicking the print pieces under Extruders-column I was able to select which color to use.
I forgot to set the layer heights but later when I remembered, I searched for "height" in prusaslicer and find layer height in the results. It is the first line in Print Settings > Layer and perimeters. The print didn't come as planned. Although there was a heavy raft under the print, the surface area of the ropes was so small the they fell down from the touch of the nozzle ripping the top layer of the raft with them. I adjusted a few things in PrusaSlicer settings, including as 3 layers of enforced support for the first which I found in the settings. I also tested how to print would look laying down, but it became basically mostly supports, which I wanted to minimize. I ended up asking Claude to create wider legs to keep the ropes from falling down. Now I also used the 0.4mm nozzles because they had nicer colors installed to them.
I heard a thunk from the printer and the beeping and it had an error message. >I moved the carriage by hand to the dock 5 so that the mounting pins were in the dock and slided the carriage right to detache the tool from it. Then I clicked continue on the screen and the process restarted. It was only now that I realized that the middle rope had again fell. There’s a clearly an issue in attaching the first layer. This has gone wrong so many times that I’m not sure should I still try this again with reducing the speed of the first layer (which I should have done already but had forgotten). I tried it anyone so I can learn. I decreased the first layer speeds by 50% each. I also set layer height again from 0.2mm to 0.4mm to speed up the print and set infill density from 20% to 30%. However, it seems that I can’t have 0.4mm layer height with 0.4mm nozzle so I set layer height to 0.3mm. Checking this assumption from Gemini, it suggested a 75%–80% rule for the layer height/nozzle ratio.
The printer crashed again in similar way and I fixed the issue as before and the printing continued. Then it happened again with more serious sounding beeping from the machine and 5th nozzle was dropped slightly from the carriage. I cleaned up the nozzle to have it back up on its place and then I was able to move the carriage to the dock. Now I restarted the print with 0.6mm nozzles as it seems the 5th nozzle has problems that needs someone else than me to fix it.
I also asked Claude to create 1mm horizontal pillars between the ropes so that they support each other during the print. And finally the print finished without problems. As soon as I had detached all the supports, the pieces began separating from the braid, and it was quite difficult to attach them back because you would need five hands to keep each in place. I should have kept the raft attached so the ropes would have been standing by default. It should had to be taller to have enough intertwining that would’ve keep them together. I used tape to have them in their place and rebuilt it, based on the photo of the finished print.
I continued trying to make this myself on Blender. I watched two youtube tutorials (this and this) that used curves to make meshes to follow those curves. This was done with adding an Array (legacy) modifier to the mesh, selecting Fit to curve, and selecting the curve that was made. Then adding a Curve modifier and selecting the same curve, makes (or is supposed to make) the mesh to follow the curve. I found out that it is crucial to have the mesh and the curve to begin from the same spot. This was adjusted with Shift + S > Cursor to Selected to both objects.
After an hour of thinking about different possible processes to make it, I decided that it is too difficult to comprehend in digital space (at least right now) or make with one rope at a time. I decided to make it with three curves instead. With a lot of adjusting I was able to make the curves resemble a braid well enough.
After trying a little while, I was able to make one rope. But for the second one the problems were on the next level. The second image below was a situation that I wasn't able to solve before I realized you can thicken the curves themselves in the data tab by adding Bevel depth.
With this new way, I was encouraged again to try out with 5 ropes and kept the previous print as a model which made it much easier. I adjusted the vertices and their handles for an hour or so and was able to create this type of a thing.
Then I set the Bevel depth to 0.01m on the data tab and solidified them (with the modifier). I also scaled one cube mesh as a pedestal for keeping the pieces together and as a support structure.
Then I exported it as stl and imported to PrusaSlicer. I decreased the print speed to try to ensure that the nozzle won't tip over the ropes and to help with the overhangs. I used automatic supports and set the overhang treshold to 15 degrees for not to fill up the whole thing with supports. I also added 20mm brim to keep it from falling over.
The print came out nice on the first try and is surprisingly robust. Also the overhangs did not cause too serious glitches on the surfaces.