For this week I decided to design a ball jointed doll version of my desk robot.
Unlike my planned robot, I was going to need to figure out legs for the doll version.
I asked chatGPT to give me some suggestions based on my desk robot design.
Using the image as a reference, and the model I had from week 1, I started to design the body arms and legs of my doll version.
Creating the main body went quickly and easily.
I used the body form from my desk robot to act as a sort of skirt bottom and extended the shape upwards to create the torso.
I made the torso part as two separate pieces and created a center piece so that there can be a pivot point on the doll at the stomach.
Next I worked on the basic shape of the legs and arms which were just modified, elongated cylinders.
At each joint connection point, shoulder/thighs/elbows/knees, I added a sphere as a place holder.
For a ball joint doll to work properly, a spherical connector point is needed at these locations to allow for smooth movement of the joints.
While trying to design the shoulder joint, I realized I couldn’t remember what the standard joint looks like on this type of doll.
I grabbed one of the few I keep on my desk and examined the shape.
I created about 5 different shoulder connector points, tweaking each one a bit at a time, until I found a design I was happy with.
With each design I would test the rotation in the shoulder socket to see how the movement looks and where the stop points were.
It took a few iterations before I had a design that I liked the look of, and also worked.
For the hands and feet, I wanted to make something simple that looks clean and neat against the arm and leg ends.
My plan was a simple cylinder that I could embed a magnet into with a small loop attachment.
Since the hands and feet are crucial for the stringing process, I had to make sure the loop fit cleanly inside the arm and leg channels!
This allows the hand base to look neat and tidy on the printed doll.
With the hidden magnets, the hand also has the hidden ability to have robot hand and feet attachments via magnets, the ability to hold metal objects in the hands, and additional foot support on metal.
I wasn’t sure if the channels in the body would be enough to count for this week's individual task.
I showed my completed print to two of my labs Fab instructors, David Taylor and Garrett Nelson, and they both said maybe but weren’t sure.
One of my instructors, Garrett, suggested I create a more advanced part for my robot body.
We sat down and brainstormed some ideas and came up with a movable heart inside the chest cavity.
That evening, I went back to my blender file and began working on creating the new chest piece.
Since the heart would need more space, I had to extend the chest piece down and shorten the middle piece.
I had a chance to work on updating my hands with the holes for the magnets to be inserted.
Since I was not at my home computer, I did not have access to my full blender file but I did have the basic hand piece in my TinkerCAD account.
I had to make some minor modifications to the file since I had made updates in blender, but it was helpful since I had the measurements I needed.
I created a hole cavity in the hand piece to fit the magnet and slightly increased the cylinder part of the hand to adjust for the height of the magnet hole cavity.
I then created the first robot hand attachment which was a simple robot claw measured with a cutout to be able to fit around a standard pencil.
For embedding the magnets into the print, I had to add a pause at layer 16 in Bambu Studio.
Something important to note is that Bambu Studio performs the pause at the START of the layer, not the end.
Technically layer 16 is where the printer will cover the hole, but since the pause is at the start I can drop in the magnets and when the print continues it will cover the magnet.
One annoyance I had while doing the magnets, is that I kept getting my polarity mixed once I dropped it into the paused print.
I kept having to take them out and recheck that I have them in the print the correct way.
After a few annoying attempts, I had the bright idea to grab a sharpie and put a dot to represent facing up for each magnet.
I am a bit disappointed in the strength of the magnets after they were embedded in the print.
There are only 3 layers on each print covering the magnets, so a total of 6 layers in between the magnets.
On their own the magnets are very strong and even a little difficult to separate from the stack, but once they were in the prints the magnetic field felt really weak and while it does the hand attachment it doesn't take any force to knock it off.
For 3D scanning, I tried to use the PolyCAM scanner app on my Samsung Galaxy S24 phone.
I was able to download the app and create a free account which should have given me a few free scans, but the app would crash every time I tried to exit the sign up for a plan pop up page.
I tried a few times with no luck. One of my group mates offered to let me use their phone to take my scan.
I accepted, but in the end did not end up using their phone.
While waiting for my turn to use their phone I downloaded a couple more apps from the app store to test out.
The first one I downloaded ended up being a bit misleading in the name and description, because it ended up being a 360 degree video recorder and not a 3D scanner like the description said.
It is possible that the app did have 3D scanning functionality, but it wanted me to set up a paid account for most of the features and the only free feature gave me 3 scans to create a “360 video and model.” The video part did sort of work, but I was never able to get a model from that app.
I looked through this week's page for app suggestions and researched the ones listed and settled on trying Kiri next.
Kiri was an open source 3D scanning app and according to the site was free for unlimited 3D scans.
I went into the app expecting a pop up to buy a subscription like all the rest, and there was one but it also did give a free unlimited 3D scan option.
The free scan option had some limitations on quality settings and features, but my first scan with the app worked great!
I found a little paper crane on the desk that a student left behind and I scanned the crane.
Within a few minutes the scan was turned into a decently accurate 3D model.
It sent me the file download link via email within about 2 minutes and I was able to bring the file directly into Bambu studio and send it to the printer.
It did have some parts that were thin and ended up missing in the prints, but overall I was happy with the app and print considering I did not do additional work to fix the mesh.
Review:
The design came out cute, not quite what I had in mind but close.
However, I couldn't figure out how to download the model to try printing it.
Every option and file type I tried to use brought me to a payment page to upgrade my plan from free which I wasn't planning to do while just trying the services.
Cost Breakdown:
The cheapest plan, with a new subscriber offer, starts at $11.94/month and you receive 3000 tokens.
Normally this plan will cost $19.99/month.
This makes each token approximately .004 cents each with the discount or .007 cents without the discount.
The higher plans offer more credits, and once you are a subscriber you can also buy additional credits.
Review:
The overall look was alot different than the Tripo model.
I found it really interesting how detailed and realistic this generated cat model was compared to the other two.
The other two softwares generated more cartoony files in comparison.
When I went to download, I tried all of the export options and they were all blocked behind a higher paid tier.
After a little bit of research and finally hovering over all the buttons, I realized that the model was generated with Meshy 6.
Meshy 6, is not a part of the free plan export options.
It cost me 10 tokens to regenerate the model using Meshy 4 to allow for exporting.
In my opinion, there was a very large difference between the model generated with Meshy 6 and Meshy 4.
Cost Breakdown:
Meshy AI offers a free plan which gives you 100, non accumulating, credits per month.
The cheapest plan, with a new subscriber offer for the first month, starts at $15.17/month and you receive 1000 non accumulating tokens.
Normally this plan will cost $20.00/month.
This makes each token approximately .015 cents each with the discount or .02 cents without the discount.
They offer a higher plan with more credits meant for teams and billed by seat/person per month.
Review:
The last service I tested was the MakerWorld PrintMon tool.
This tool creates a good, cute, image and it was closer to what I had imagined and prompted compared to Tripo.
It created a cute smoke/magic effect coming from the cat hand, which would not actually work as a printable object because it is floating unattached from the model.
I liked that I was also able to make basic edits right in the software, such as adding a base and moving and scaling the model.
It was also really easy to download the model for printing.
When downloading it gives you the OBJ, the 3MF, and a painting reference image which I found as a nice touch.
Cost Breakdown:
Makerlab offers a free plan which gives you 20 credits per month, and you can earn additional credits through interacting with MakerWorld and Bambu.
According to the MakerReward program information, the approximate value of 100 MakerReward Points is equal to an $8.20 Bambu Lab Store gift card.
You can convert MakerReward Points to MakerLab credits at a rate of 1 coin = 10 credits.
This makes each Reward Point approximately $0.082 cents and each credit approximately .008 cents.
In my repo is a zip folder containing files for my week 5.