From Impossible Prototype to Hanayama Cast Puzzle: How Metal Puzzles Are Made

Pick up a Hanayama Cast Puzzle, and the finished object gives almost no indication of how much work went into it before it reached your hand.

The surfaces are smooth. The pieces fit with controlled tolerances. Curves that appear almost impossible to manufacture pass through one another exactly as the designer intended. Even the plating can look purely decorative.

It isn't. A rare behind-the-scenes account by Dutch puzzle designer Oskar van Deventer shows just how much design, engineering, tooling, casting, finishing, testing, and manual work can go into producing a single Hanayama puzzle.

His case study follows the development of Cast Twist, a puzzle that began as an experimental 3D-printed design considered impractical to mass-produce and eventually became a commercial Hanayama Cast Puzzle.

The process reveals something important about mechanical puzzles:

Designing the puzzle is only the first puzzle. Figuring out how to manufacture it can be another one entirely.

Oskar documented the complete process after visiting Hanayama's production facilities in China with George Miller, Roxanne Wong, and José in 2013. His original six-page report, How It Is Made: Hanayama Cast Twist, provides an unusually detailed look at what happens between a designer's idea and the finished metal puzzle. 

How Are Hanayama Cast Puzzles Made?

For Cast Twist, the documented production process included the following stages:

Each stage can affect how the finished puzzle behaves. With a mechanical puzzle, a small change in tolerance, polishing, or plating can alter how the pieces move and fit together.

Before Cast Twist, There Was a Puzzle That Couldn't Be Manufactured

The story of Cast Twist began long before the puzzle appeared in stores.

In 1986, Oskar van Deventer created a puzzle called Oskar's Disks. Puzzle designer and collector Wil Strijbos suggested that the two disks should be able to pass completely through one another.

That idea contributed to several later designs, including Hanayama's Cast Disk, released in 2000. 

Then 3D printing changed what was possible.

After George Miller introduced Oskar to the technology, Oskar designed a much more complex version in 2004 called Snake Ball. Miller produced the physical prototype on his 3D printer.

The puzzle could now exist.

The problem was that Oskar and the others involved believed the geometry could not realistically be mass-produced using conventional manufacturing methods.

So the project was abandoned. This is an important distinction in product design.

A shape that can be created by a 3D printer is not automatically a shape that can be economically die-cast thousands of times.

A prototype only has to exist once. A production design has to be made repeatedly, consistently, at an acceptable cost, while still behaving exactly as the puzzle designer intended.

Eight Years Later, the Puzzle Came Back

In early 2012, puzzle collector Brett Rothstein asked Oskar for a Snake Ball sample. Oskar produced one through Shapeways.

Later that year, at the International Puzzle Party, he also gave samples to Teddy Sakamoto and Takeshi Onishi of Hanayama.

They wanted to know whether Hanayama's Kyoo Wong could find a practical manufacturing solution.

He did. Kyoo Wong, working with a Chinese CAD specialist, found a way to divide the difficult Snake Ball ring geometry into three separate castable components.

That changed everything. 

The idea itself had not suddenly become simpler.

The manufacturing strategy had changed.

Hanayama released Cast Twist in October 2013, crediting Oskar van Deventer as its designer. Hanayama rates it Level 4 and describes its challenge around sliding protrusions through grooves in the puzzle's asymmetric form.

See Hanayama's official Cast Twist page

The Solver Sees Two Pieces. The Factory Sees Something Else.

This may be the most interesting part of the entire story.

A puzzle solver experiences the final object as a small number of finished puzzle pieces.

But those visible pieces do not necessarily correspond directly to the parts that come out of the casting machine.

For Cast Twist, some of the geometry had to be divided into multiple castable components and later assembled.

Why?

Because a metal part has to be removable from its mold.

Imagine casting a complicated shape with hooks, undercuts, curves, or surfaces that wrap around one another. The molten metal can fill the shape perfectly, but once it hardens, there may be no physical way to pull the finished part out of the mold.

The solution is not necessarily to simplify the puzzle.

Instead, manufacturing engineers can determine how to divide the geometry into parts that can be cast and then assemble those parts afterward.

This creates an unusual situation:

The puzzle designer solves the mechanism. The production engineer solves how to make the mechanism.

The First Production Step Is Not the Mold

Before Hanayama commits to production tooling, the company creates hand samples machined from brass.

These samples allow the production team to test the puzzle physically and check the dimensions, movements, tolerances, and other manufacturing details.

If the original designer supplies a physical sculpted prototype rather than a digital model, Hanayama can first 3D-scan the object so that its geometry can be brought into CAD.

In Cast Twist's case, the design already existed digitally, so that scanning step was unnecessary. 

This stage matters because something that works in a designer's prototype may behave differently when translated into another material or manufacturing process.

The hand sample creates a bridge between the design prototype and the production puzzle.

How Is the Mold for a Hanayama Cast Puzzle Made?

Once the hand samples are approved, work begins on the production mold.

The mold itself is first designed in CAD. The engineers have to consider not only the final shape of the puzzle but also how the newly cast parts will be removed from the tooling.

For Cast Twist, negative versions of the required mold cavities were machined in copper.

Those copper forms were then used to create the actual cavities in steel through a process called spark erosion, also known as electrical discharge machining or EDM. 

What Is Spark Erosion?

Spark erosion is a manufacturing process that removes metal using controlled electrical discharges.

Instead of cutting the steel mold with a conventional blade, the copper electrode and steel are brought into conditions where repeated electrical sparks erode tiny amounts of material from the steel.

Gradually, the shape of the copper electrode is reproduced as a cavity in the steel die.

Oskar described the process as extremely slow, proceeding spark by spark until the required geometry was formed.

This allows manufacturers to create complex mold surfaces with the precision required for small mechanical components.

And precision matters enormously here.

A decorative casting can tolerate small dimensional variations without affecting its function.

A mechanical puzzle may not.

What Metal Are Hanayama Cast Puzzles Made From?

In the documented Cast Twist production process, Hanayama used a special molten zinc alloy.

Once the steel mold was completed, Hanayama first performed a small trial or "zero-series" production run to make sure the tooling worked correctly.

During production, molten zinc alloy was pressed into the mold cavity. The metal solidified inside the water-cooled die, after which a robotic arm removed the still-hot casting and placed it on a slowly moving conveyor to cool. 

The result at this stage does not look like the finished puzzle you see in a store.

The pieces are still attached to extra casting material.

What Is a Sprue?

During die casting, molten metal needs channels through which it can flow into the mold.

The solidified metal left in these channels is known as the sprue.

Once the casting cools, the sprue is removed from the puzzle parts. In the Cast Twist process Oskar observed, that excess zinc alloy was returned for recycling. 

Only then does the individual puzzle component begin moving toward its recognizable finished form.

Casting Is Only the Beginning

Freshly cast pieces still require considerable work.

According to Oskar's factory account, Hanayama used specialized machinery and manual operations to remove excess material and perform processes including cutting, threading, bending, and other adjustments.

The pieces were then buffed to remove sharp edges. After manual inspection, they went through a vibratory tumbler, which further smoothed and polished the metal surfaces. This part of manufacturing is easy to underestimate. A mechanical puzzle is something the solver repeatedly handles, turns, slides, rotates, and rubs against other pieces.

A rough casting edge would not simply look unfinished. It could interfere with movement or make the puzzle unpleasant to use. That is why finish is part of function. We explored this idea from the solver's side in What Makes a Great Metal Puzzle? Hanayama and Tetso Offer Two Very Different Answers.

Why Does Plating Matter to a Mechanical Puzzle?

After polishing comes one of the most surprising stages: metal plating.

Hanayama can test different plating finishes to achieve different appearances and surface characteristics. Some may provide higher scratch resistance, while others create a more antique-looking surface.

But choosing a finish is not purely cosmetic.

Different plating systems can add different amounts of material to the surface.

Oskar reports that the dimensional differences can matter at the micrometer level, and that in some cases Hanayama must correct the mold to compensate for the thickness created by the chosen plating. 

Think about what that means. The puzzle has already been:

designed, engineered for production, prototyped in brass, converted into production CAD, tooled in steel, die-cast, trimmed, machined, buffed, and polished.

Then the final surface finish can still alter the fit enough to require changes further back in the production process.

For mechanical puzzles, appearance and tolerance cannot always be separated.

Why Do Tolerances Matter So Much?

Mechanical puzzles live in the space between pieces.

A gap that is too tight can make a legitimate movement feel impossible.

A gap that is too loose can expose a mechanism, create unwanted movement, or allow the puzzle to be solved in a way the designer never intended.

The challenge is not simply producing two accurately shaped pieces.

The manufacturer has to preserve the designer's intended relationship between those pieces.

That is why the manufacturing story behind Cast Twist is so revealing.

The mechanism does not exist only in the shapes.

It exists in the clearances between the shapes.

Are Hanayama Puzzles Assembled by Machine?

Not entirely.

Oskar's account describes significant manual work during final assembly and packaging.

One particularly interesting detail is that workers were trained to assemble the puzzles efficiently without necessarily being taught how to solve them in reverse.

The individual components were protected during handling so that finished surfaces would not scratch against one another before assembly. The completed puzzles were then secured in retail packaging, boxed in larger quantities, placed on pallets, and prepared for international shipping. 

By the time a puzzle reaches the solver, most evidence of these production stages has disappeared.

You see the puzzle. Not the factory problem that had to be solved to create it.

Are All Hanayama Cast Puzzles Made This Way?

No, and this distinction is important.

Cast Twist gives us an unusually detailed case study, but it should not be treated as a universal manufacturing specification for every Hanayama puzzle.

Different geometry can require different manufacturing methods.

Oskar gives a good example on his own Hanayama page: his Cast Möbius design had also once been considered difficult to mass-produce, but Hanayama eventually manufactured it using zinc spin casting with silicone molds, rather than the steel-die process documented for Cast Twist. 

So the better conclusion is not:

"Every Hanayama puzzle is manufactured exactly this way."

It is:

The manufacturing process has to be engineered around the particular puzzle.

That is a much more interesting idea anyway.

Where Did Hanayama Cast Puzzles Begin?

Hanayama's own history places the beginning of its Cast Puzzle line in 1983, with early releases including Cast Key, Cast A.B.C., and Cast Star.

Over the following decades, the collection expanded to include designs from numerous independent puzzle creators and eventually hundreds of variations, redesigns, and new mechanisms. Hanayama's timeline records Cast Disk in 2001, NEWS in 2003, Vortex in 2008, Cylinder and Twist in 2013, Infinity and Padlock in 2016, and many later additions. 

Explore Hanayama's official Huzzle history

Today, the name on the package might say Hanayama, but the underlying mechanism may have come from an individual designer working somewhere else in the world.

That is why we have increasingly separated Designer and Maker in Kubiya's product information.

A mechanical puzzle is often the result of both.

Designer and Maker Are Not the Same Thing

Cast Twist makes the distinction unusually clear.

Designer: Oskar van Deventer
Maker: Hanayama

Oskar created the puzzle idea and its mechanical behavior.

Hanayama and its engineering partners worked out how that idea could become a repeatable commercial product.

Neither role replaces the other.

This relationship appears throughout the world of mechanical puzzles. Hanayama has produced designs by Nobuyuki Yoshigahara, Vesa Timonen, Akio Yamamoto, Jin-Hoo Ahn, Kyoo Wong, Oskar van Deventer, and many others. 

At Kubiya, you can explore our broader Hanayama puzzle collection, including several other Oskar van Deventer designs such as Nutcase, Hanayama Chain, and Cast Key 3. 

3D Printing Changed Puzzle Design, but It Didn't Eliminate Manufacturing Problems

The Cast Twist story also captures an important moment in modern puzzle design.

3D printing allowed Oskar to create a physical Snake Ball prototype with geometry that would have been difficult to prototype using traditional machining.

That expanded what a puzzle designer could test.

But the prototype did not eliminate the next question:

Can this be produced thousands of times?

The answer initially appeared to be no. The breakthrough came only when the manufacturing team reconsidered how the geometry could be divided and assembled.

That relationship still matters today. Modern designers can produce increasingly complicated prototypes using 3D printing, CNC machining, resin printing, and other tools. But when a design moves toward commercial production, questions about mold release, material behavior, tolerances, surface finishing, assembly, durability, and cost return immediately.

The computer can tell you that a shape exists. It cannot automatically tell you that the shape makes sense on a production line.

What Makes Cast Twist Such a Useful Case Study?

Many manufactured objects hide their complexity.

Mechanical puzzles hide it especially well.

When Cast Twist reaches the solver, the production process has disappeared into the object. There are no visible copper electrodes, steel molds, sprues, polishing machines, CAD revisions, plating tests, or unsuccessful prototypes.

There is just a small metal puzzle.

That simplicity is deceptive.

The finished object represents decisions made by the original designer, CAD engineers, toolmakers, casting specialists, finishing teams, assembly workers, and Hanayama itself.

And all of those decisions have to preserve one thing:

the puzzle.

Not merely its appearance.

The movements.

The restrictions.

The clearances.

The false assumptions.

The solution.

Why This Matters When You Hold a $20 Metal Puzzle

One of the remarkable things about Hanayama is that this level of production engineering ends up inside a relatively inexpensive object.

That does not mean every Hanayama puzzle requires the exact development journey of Cast Twist. Some designs are easier to manufacture. Others may require entirely different techniques.

But Cast Twist demonstrates why judging a mechanical puzzle simply by its number of pieces can be misleading.

A puzzle might contain only two visible pieces and still require sophisticated CAD work, custom tooling, die casting, several finishing processes, plating experiments, manual assembly, and years of design history before those two pieces reach your hands.

The physical object may be small.

The process behind it isn't.

From Idea to Puzzle

The history of Cast Twist can be reduced to a simple timeline:

1986: Oskar's Disks introduces the original idea.
2000: The concept contributes to Hanayama Cast Disk.
2004: Oskar designs Snake Ball after being introduced to 3D printing.
2004: The puzzle is considered impractical to mass-produce, and the project is abandoned.
2012: New prototypes bring the design back to Hanayama's attention.
2012-2013: Kyoo Wong and a CAD specialist redesign the geometry into castable components.
2013: Tooling, casting, finishing, plating, assembly, and production are developed.
October 2013: Hanayama releases Cast Twist. 

More than nine years separated the first Snake Ball prototype from the commercial puzzle.

The solution was not simply a better puzzle design.

It was a better way to manufacture the design.

The Puzzle Behind the Puzzle

Mechanical puzzle solvers spend a lot of time asking:

How do these pieces fit together?

Why can't this part move?

What is preventing these two shapes from separating?

Where is the hidden path?

Manufacturers have to ask surprisingly similar questions.

Can this shape come out of the mold?

Can this geometry be divided into manufacturable parts?

How much clearance will remain after plating?

Can the puzzle be assembled efficiently?

Will a tolerance change accidentally reveal the solution?

Can thousands of units behave like the designer's prototype?

In Cast Twist, solving those manufacturing questions is what transformed an abandoned 3D-printed experiment into a Hanayama puzzle sold around the world.

Perhaps that is the most fitting part of the story.

Before anyone could solve Cast Twist, someone first had to solve how to make it.

If you're new to the terminology behind take-apart, disentanglement, interlocking, packing, and other physical puzzles, our Mechanical Puzzles: A Beginner's Guide to a Much Bigger World explains the major puzzle types and how they differ. And if you want to explore the finished objects themselves, browse Kubiya's metal mechanical puzzle collection.