TL;DR
Get business pricing on tools and workshop supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
A maker says Google Gemini helped generate OpenSCAD code for parts of a Fallout-inspired Halloween prop he is building with his son. The AI outputs needed repeated corrections, but provided a starting point for visualizing and printing components; the project is still in progress.
A maker is using Google Gemini to help design and 3D-print parts for a Fallout 4-inspired Halloween prop he is building with his son. In a report published by ToolGuyd, he said Gemini generated OpenSCAD code for several components, but its designs required repeated prompting and manual adjustments; the prop is still unfinished.
The project began after the maker’s son asked for a Fallout-themed costume. The original idea of building DIY power armor was too ambitious for the available time, so they shifted toward a more manageable costume with a decorated chest piece and a distinctive handmade prop. A planned motorcycle chest protector did not arrive as expected, and the maker said they would instead use foam and craft materials for the costume.
For the prop, the maker used Gemini to develop a truncated-cone component with a mounting ring. After several iterations, the AI supplied OpenSCAD code describing two connected truncated cones. The maker said the code allowed him to adjust parameters and test dimensions. A print made with filament he described as very wet was good enough to help him plan a larger revision, including mounting holes for attaching the part to a wooden box.
Other AI-generated designs were less reliable. The maker reported that Gemini missed instructions about keeping side arms flush and placing pill-shaped holes only on the main face, and added chamfers he did not want. He also asked the service to break a fantasy tool into printable subassemblies; he found that result unhelpful, although it supported his existing idea to use mounting pins for glued assembly.
AI as a Workshop Design Aid
The report offers a practical example of how a generative AI tool can fit into a hands-on making process: not by producing a finished prop, but by helping turn an incomplete idea into editable code and testable parts. That can be useful when a maker is still working out geometry or has difficulty translating a mental picture into a model.
The account also shows the limits of that assistance. Gemini did not consistently follow design constraints, and the maker still had to inspect the code, adjust dimensions, print prototypes and decide how parts would connect. The tool helped with one stage of the work; it did not replace the judgement and fabrication needed to make a functional assembly.
For readers experimenting with AI-assisted design, the distinction matters: a generated model or script is a starting point to review, not proof that a part will fit or work. The maker’s iterative approach—changing parameters and printing a low-density version before committing to a final part—illustrates how digital suggestions can be tested against physical results.
Popular right now
Dreame L50 Ultra Robot Vacuum and Mop
Shark Navigator Lift-Away Upright Vacuum (NV360)
Dreame L40 Ultra Gen 2 Robot Vacuum and Mop
As an affiliate, we earn on qualifying purchases.
From Costume Plan to Printed Parts
The maker said he has experience modeling objects he has built before, but finds it harder to create a model when the design is not fully imagined. He described planning as especially difficult when a project involves multiple parts that must fit together. For this prop, his current approach is to attach individually designed components to a small wooden box, using screws, glue or other hardware where needed.
He is also considering a hard hat and a flashlight to create a scavenged look, along with printed fittings and air hoses. The green fittings shown in the report came from McMaster-Carr 3D files and were used to visualize how parts might fit. The maker said he had not decided whether to buy inexpensive metal fittings, and used the printed examples to avoid ordering more varieties than necessary.
He may later remodel components manually in Fusion or similar software, which he expects would give him more control over related dimensions and the overall assembly. For the current project, however, he said time constraints favor getting a rough first version made rather than redesigning every part.
“The AI made a lot of mistakes, and I’m also learning how to better constrain dimensions in my prompts.”
— The maker, in the ToolGuyd report
The Prop Is Still a Prototype
The maker has not said whether the current prints will fit the final assembly or whether the prop will be completed by Halloween. He described the existing parts as preliminary and said a low-density version of one component was printing while he considered its next revision. The final color, dimensions and construction method have not been settled.
It is also unclear whether he will use Gemini-generated code for the full design or move more of the work into manual CAD modeling. He said the AI’s suggested subassembly breakdown was not very helpful and that he may remodel the handle in Fusion or similar software. The article provides the maker’s account of this project; it does not establish how Gemini would perform on other designs or under different instructions.
A Rough Build Before Halloween
The maker’s stated next step is to assemble a rough first version of the prop, using a small wooden box as its main structure and attaching separately modeled components. He plans to print and test revised parts, then decide whether to keep adjusting OpenSCAD code or remodel pieces in CAD software. He hopes to refine the prop before Halloween, but has not set out a firm completion date.
The costume itself is also taking shape through foam and craft materials after the delayed chest protector fell out of the plan. The maker said he and his son intend to build the prop together; its final identity—such as a detector or another fictional device—remains open. Further details will depend on how the parts fit and how much time remains.
Key Questions
What is the maker building?
He and his son are making a Fallout 4-inspired Halloween costume and prop. The original idea of DIY power armor was scaled back to a more achievable costume and a distinctive handmade device.
How did Gemini help with the design?
Gemini generated OpenSCAD code for components, including a truncated-cone part with a mounting ring. The maker could change parameters in the code and print versions to test the geometry.
Did the AI produce accurate designs?
Not consistently, according to the maker. He said Gemini ignored some requirements about arm alignment and hole placement and added unwanted chamfers, so he had to revise its outputs.
Is the prop finished?
No. The maker described the project as ongoing, with prototype parts and further revisions planned. He hopes to refine the build before Halloween but has not confirmed that it will be complete by then.
Will he keep using AI for the rest of the project?
That is not settled. He may continue editing OpenSCAD code, but also said he might remodel some components manually in Fusion or similar software for more control over dimensions and assembly.
Source: rss
