I was at the nearby Bluffs a few weeks ago with my mom, sitting on a rock and staring out to the ocean. I was out on a photography trip with my newly unearthed Nikon D700 DSLR camera. It was a very picturesque day, with a bright blue sky and clear weather. We were in a small eave surrounded by light foliage, boulders, and trees in the park. A few yards ahead of us, were the beautiful pebble beaches formed through thousands of years of erosion and sedimentary deposits. At my feet, there were bits of grey gravel- slate, stone, granite, and even bits of quartz. I picked up a handful or two, feeling it in my palm. It was just the right size for a craft… and that was how the idea was born.

Consumables

1. Sakrete Sand Mix
2. Water
3. Cooking oil
4. Green slate
5. Grey gravel
6. Black polished pebbles
7. Beach stones
8. Miscellaneous decorative rocks
9. White decorative sand
10. Glass craft beads
11. Acrylic craft paint (I used green and white)
12. Fiberglass mesh
13. PLA filament
14. Plastic wrap
15. Paper towel/tissue
16. Newspaper/work-surface protection
17. Q-tips
18. Plastic bags
19. Disposable containers
20. Paper plate

Tools

1. 3D printer
2. Computer
3. Autodesk Fusion
4. Python
5. Cura
6. Digital kitchen scale
7. Hammer
8. Mixing stick
9. Paintbrush
10. Plastic knife
11. Metal ruler
12. Small measuring container
13. Scissors
14. Massage gun
15. Small screwdriver/chisel
16. Old toothbrush
17. Stopwatch/phone
18. Safety glasses
19. Gloves
20. Respirator/dust mask

Previously, I had decide that it was about time to make a concrete project and learn how to use this incredibly versatile medium. I also loved pretty mosaics and the look of stained glass artworks. Upon seeing those rocks, it all clicked! I was going to make a mosaic coaster/trivet with all sorts of colorful glass and stones, bonded together with concrete for a strong base!

A few days later, I went to my local dollar store to see what craft rocks they had in stock, and I was not displeased by their arsenal of stone, sand, gravel, slate, and everything in between. I ended up getting a box of green colored slate, polished black pebbles, clear glass beads (that I intended to shatter and dye, but ended up waging a war against them- more on that later) and some white sand to put in the concrete (which screwed me over later…) I also used the few small handfuls of Bluffs gravel as grey filler aggregate. On another note, please never take river rocks from the side of a river- those are necessary to keep the river at a stable level and not free rocks for crafts. Consider going to a rock quarry instead!

Since I had no silicone at home, I decide to try using a 3D printed mold instead. I would design the mold in Autodesk Fusion, a highly powerful free software for CAD design beyond Tinkercad. The trivets would be hexagon shaped to make them modular and able to freely connect at the sides. An individual trivet would be 6cm per side, 12cm wide, corner-to-corner, a comfortable size for a single cup or bowl. If you have a larger pot or bowl, assemble 3 or 4 together to easily accommodate it! The final trivet collection will be arranged in a classic hexagonal flower, 1 in the middle and 6 on the sides.

To add some extra flair, I also decided to add Voronoi patterns to the mold. Unfortunately, that means much more complex modelling that would be tedious and honestly too difficult for my skill level, especially since I wanted them to connect seamlessly to create a beautiful visual pattern. Given my history in procedural generation, I can instead create a script to draw those Voronoi patterns for me, paste it onto each tile, and give me a separate .dxf file to import as a sketch into Fusion, and work from there. This means I have a fully customizable, fast, and fluid workflow, from choosing exactly which pattern and seed I want, to doing simple extrudes to efficiently create all 7 molds in a matter of minutes.

I was inspired by other concrete projects with embedded fireglass or tiles to create a terrazzo-style finishing, which they grinded down with an orbital sander to reveal the pretty surface underneath the concrete. I did not have an orbital sander at home, so I decided to use another casting method to imitate this effect: face-down casting. I would print a hexagonal mold with the Voronoi ridges at the bottom, carefully arrange the stones before pouring concrete over them. When it finishes curing, I can demold and flip it over to reveal a flat surface with all the rock faces preserved and safe from a concrete layer.

NOTE: please expand all picture sections, because there are A LOT of important pictures!

I used the following Python libraries:

The script is attached below!

In a nutshell, the script gathers data (the variables you provide), does a bunch of math, calculating borders, divisions, edge length, hexagon's center, etc. It then creates a digital blueprint saved in data, and renders that into a compiled image of the 7 assembled hexagons- flower_preview.png and cast_preview.png (the flipped version that you will get after demolding.) Most importantly, it uses this digital footprint, divides it into individual hexes, and systematically transforms it into .dxf sketch files with a continuous Voronoi pattern.

Here are the variables I used to get my mold:

Changing the seed will change the result you get, so play around and see what design you like! I recommend finding a pattern that looks visually appealing regularly and flipped, and has bigger areas. Small areas are annoying to cast properly... lowering the CELLS_PER_HEX value can be helpful!

Attached below are the .dxf files I used in Fusion, if you would rather replicate my specific pattern :)

Now that you've gotten your .dxf files, you have the flat 2D blueprint for your molds. However, a blueprint or plan is nothing without ACTION! Let's actually build these molds in 3D and get to printing them as soon as possible.

Workflow:

1. Import the .DXF file (Insert > Insert DXF), select the tile you want to work on first, select the XY plane, and set it to One Sketch Per Layer
2. Toggle visibility off for all sketch layers except TILE_OUTLINE and offset the hexagon by 2mm outward (keybind O) creating 2mm thick walls.
3. Select both the hexagon and the newly created frame and extrude (E) 17mm upwards. We are going to create a 2mm thick base.
4. Select just the hexagon and extrude-cut 15mm upwards, starting with an offset of 2mm to keep the base.
5. Switch to the RIBS_BASE sketch layer and extrude-join the voronoi ridges up by 4mm (with an offset of 2mm). This means the deepest ridge will be 4mm into the rest of the concrete. Additionally, set the taper angle to -5 degrees for a better draft and easier mold release. Without this draft, the concrete will grip onto the mold, making for a difficult demolding process.
6. (I understood this principle for the Voronoi ridges... I did not apply it nearly well enough everywhere else... this will become important later.)
7. Switch to the BORDER sketch later and extrude-join the border by 4mm (offset 2mm) with no taper to avoid geometry issues.
8. Check finished mold with the RIBS_TOP sketch layer visibility on- the sketch lines should match up with your geometry exactly. Make sure to view from the top, with X horizontal and Y vertical!
9. Optionally, you can drag a Fusion plastic appearance onto your mold to make it look better.
10. File> Export to STL!

Repeat for all 7 molds.

I sliced all seven molds in Ultimaker Cura. Since these are casting molds rather than finished display pieces, I used fairly quick print settings:

Print Settings

1. Layer Height: 0.3 mm
2. Wall Thickness: 0.8 mm
3. Wall Line Count: 2
4. Top/Bottom Thickness: 1.1 mm
5. Top Layers: 3
6. Bottom Layers: 2
7. Infill: 20%
8. Infill Pattern: Triangles
9. Printing Temperature: 210°C
10. Build Plate Temperature: 60°C
11. Print Speed: 100 mm/s
12. Outer Wall Speed: 50 mm/s
13. Inner Wall Speed: 50 mm/s
14. Travel Speed: 150 mm/s
15. Supports: Off

The walls of the molds don’t need to be super super smooth, so the 0.3 mm layer height saves a lot of printing time. The layer lines do show up on the concrete during casting, but I actually liked how it looked. It gave the concrete a bit more life instead of just being "boring flat grey concrete side". I was also a bit stingy on the top and bottom layers, because when printing something large and wide, those extra layers tend to eat the most time. For something like a mold that doesn't require significant structural support or padding on the top/bottom, this is absolutely fine.

Each mold took roughly 2h 30m to print, using about 35 g of filament, although this will vary depending on the exact Voronoi pattern and slicer settings. In total, I printed 8 molds (I ruined one AAAH more on that later) using ~290g of filament.

I gathered all the aggregate I bought or collected and sorted them into small containers. Honestly, the moral of the whole Instructable, in my opinion, is to keep as many little containers as possible. You will always find a use for them in the end. I've used cracker packaging, the lid on some cream rolls, yogurt containers, paper plates, and so many more little containers in this project.

1. Black Polished Pebbles- Smooth, dark polished rocks provide strong contrast against the lighter concrete. Their small-to-medium size also makes them useful as attractive filler between larger pieces.
2. Grey Gravel- Mostly used as practical filler. The small, irregular pieces are great for squeezing into narrow gaps and awkward corners where larger stones simply won’t fit. They also add a rougher, more natural texture. This really carried the whole natural, rocky feel of the piece.
3. Green Slate- One of my favorite materials in the project! The slate is thin, flat, and smooth, with a bold green color and beautifully angular edges. The flat pieces are especially good for tiling larger sections of a Voronoi cell while adding large areas of color.
4. Miscellaneous Rocks/Slate- These are the weird, unique pieces I collected for as accent pieces. These were larger and had some unique characteristics, like their color, flatness, layers, shape, etc.
5. White Decorative “Sand”- I've been lying to you!! This isn’t normal sand, it's actually used at the bottom of candles for purely decorative purposes. My original plan was to spread it across the bottom of the mold to create a clean finish without needing white paint (unfortunately I was all out at home), while also keeping wet concrete away from the visible faces of the stones. This caused some... rather catastrophic effects during my first prototype… more on that later.
6. Craft Glass Beads — These were intended to become the main visual highlight. My plan was to shatter the clear beads into crystalline chunks, tint them a beautiful green, and use them to pave large areas for maximum glam. In sunlight, the glass would SPARKLE and create a really pretty contrast against the comparatively outshined stone and slate. There was only one minor problem: I severely underestimated how difficult these things would be to shatter. Curse you, glass beads.....

My goal sounded simple enough: smash the beads into small, satisfying chunks that I could use to fill the mold. Not glass powder, tiny slivers, or giant pieces. Just nice crystalline shards. Turns out, the glass had other plans.

Safety comes first. I put the beads inside two plastic bags and wrapped it in a thick cloth, forming a thick cocoon to keep the glass shards from shattering and flying everywhere.

I also wore work gloves and safety glasses and covered exposed skin. Don't start smashing on your living room floor- go outside to your balcony or backyard instead and keep people away from your working area. Glass shards can get real messy, real fast, so it's best to keep it contained in a space.

SO I HAMMERED!

And hammered.

And hammered more.

And… nothing happened.

I was sweating my ass off outside, and cloth was getting holes in it, and the plastic bags underneath were certainly breaking apart too. And after all that effort I had successfully broken... two glass beads. One of those had been pulverized into tiny, practically unusable smithereens.

This was going extremely well!!

At this point I got really pissed and I was determined to destroy this dumb sack of glass beads. I asked ChatGPT multiple times, each more desperate than the last, "how the flip do I smash these beads???"

I went on a side quest. I went downstairs to the garbage room in a desperate attempt to find a glass bottle, hoping, praying that it would be easier to break. I came back with an empty bottle of cooking wine. Surely this would be easier....

Nope. Even that wasn’t breaking the way I expected. My gentle impacts weren’t doing much, and this finally made me stop and think about what was actually happening instead of throwing around a hammer in increasingly dangerous ways.

After calling a friend for advice, several problems suddenly became obvious.

The thick cloth and multiple plastic bags were absorbing a lot of the force. I had essentially created a fluffy blanket to take all the blows. The soft wooden floor wasn't doing me any favours either. The rounded shape of the glass beads also meant that they could slip around and move from each other when struck. They were basically dodging my hammer- and when I did manage to strike them, their curves spread the force to a negligible degree.

I eventually managed to break the bottle with more advice from my muse, but then I realized that bottle glass SUCKS.

It was dirty, the wrong color (it was a dark brown), and, most importantly, the shards are curved pieces. I needed flat, irregular chunks that could sit nicely against the bottom of my molds, not bent slivers of sad brown bottle.

So after all that, I ended up going back to smash the glass beads anyway...

IT WORKED!!!

Instead of trying to smash an entire moving pile at once, I worked on individual beads, with less padding (unlike the giant safety blanket I made early). This was a slow, grueling process, but I finally started getting the kind of irregular pieces I wanted.

It was still an absolutely miserable process. I was bent over, my back hurt, I was sweating, and each bead took individual attention—but slowly the pile of usable glass started growing.

Eventually, I had a nice amount of clear glass shards in a tray, ready for dyeing.

Victory!!!

A NOTE ON SAFETY:

Breaking glass can produce not only obvious sharp pieces, but also tiny fragments and fine glass particles!

PLEASE wear eye protection and gloves and avoid breathing dust. Don't swing your hammer around like a madman, you can seriously injure yourself or give yourself glass splinters. Keep the fragments contained during both breaking and cleanup, and don’t use tiny glass flakes or powder in the mold.

This is the fun part!

I didn't have nail polish, or Mod Podge for the "standard" tinted glass baking method using food colouring, but I did have some acrylic paint. If you can't tell, I'm all for cheap ass household alternatives- this is just base dollar store craft paint. I don’t know how it will hold up in highly alkaline concrete long-term, but it works for now!

Get a small container for paint. I used a lid of some cream sticks or whatever from years ago. Put in a small dollop of paint, dilute it slightly, and put in color of your choice- I did green because it’s my favorite colour and it goes well with the slate! Remember, you only need a small amount of paint and water. The consistency should be that of basically water, but it should look opaque in the disk. Dab lightly on the glass using a paintbrush. I used a paper plate to hold the glass, which makes them look more colorful then they actually are, because the paint seeps into the plate.

After going over the glass with the green, I added a bit more blue to my mix to get a different hue, again dabbing lightly around different areas. Make sure to leave some chunks mostly untouched to get a nice variation! Multiple layers can make the color deeper, but don't color so much that the paint layer becomes opaque. You can also try different colors and shades together to make a more interesting look, akin to natural crystal.

Let it dry for at least 30 minutes and carefully scrape it off the plate- it may stick to the paper fibres. Do not use your bare fingers. They will get cut and shredded on the plate like a cheese grater.

It looks beautiful under the sunlight! I see this as a great cheap alternative to craft glass aggregate or fire glass, for purely decorative purposes. There's not much difference, actually!

Oil it up, baby!

For mold release, I didn't have any fancy products at home to use on the plastic, but I heard cooking oil (vegetable/canola) does just as well, if not better. I dipped a Q-tip in a small amount of oil and slowly went over the sides of the mold, the top of the edges, the voronoi ridges, as well the the sides of those ridges. I left the section faces intact.

Please note that you do not want dribbles and blobs of oil! Only spread enough oil on the molds so it is visibly wet. If you end up with large amounts of oil on your mold, it can affect the concrete when it casts.

This next part is arguably the most important, beauty-wise. The art of arranging aggregate is actually more complicated than you think! I tried to use different kinds of aggregate in each section, and made sure to use the same type of aggregate in sections that connect through the hexes to really make it look connected. I always put glass chunks in every single hex, because they're just so pretty! However, each and every piece of aggregate must be placed carefully, with their largest flat face contacting the bottom of the mold. This ensures the hex comes out looking clean and uncovered by concrete, with their best face revealed.

Here is the final design I created. I'm really happy with it, and it's arguably the most picturesque thing in the whole project!

For casting, you should have a:

1. large disposable container, around 1L, (I used a yogurt container)
2. a mixing stick / scraping tool (I used the handle of a paintbrush)
3. a screeding tool (I used a plastic knife and a metal ruler)
4. Newspaper or another surface covering for your workspace.
5. a bowl of water
6. a small scale (I used a digital scale that we use for cooking and baking at home)
7. a roll of paper towel, tissue paper, or toilet paper for wiping away concrete
8. and your oiled, arranged, and finished mold within reach

Make sure to wear PPE: vinyl gloves, safety goggles, a respirator/approved filtration mask and work in a ventilated environment. Replace your gloves if they wear out or develop holes- cement is VERY alkaline and caustic. A slight tear can introduce wet cement into your glove and trap it there, causing severe chemical burns.

I calculated that the base volume of the the inside of each mold was about 127.6 cm^3 by merging a body in Fusion. The aggregate takes up some of this volume, so the exact amount of concrete needed will depend on how densely you arrange your rocks and glass.

MIXING THE CONCRETE

I start every hex by weighing the dry cement powder, scooping it into my container with a glove. I measure 40 ml of water on the scale, making sure to reset the measurement with the empty container on. I found it helpful to START A STOPWATCH as soon as you start mixing in the water. I mix it together slowly, starting with 20 ml, mixing, and then add another 10 ml and mix for 2-3 minutes total. For the last 10ml, I add 2 ml bit by bit until it reaches a consistency of creamy peanut butter. It should visibly look smooth and hold its shape, without running or flowing. Don’t feel like you absolutely have to use the entire 40 ml, or only 40ml. The final consistency is more important than using an exact amount.

PACKING THE LAYERS

Carefully dab in small blobs of cement in between rocks with large spaces. I like to go section to section, working the cement into every little nook and cranny visible. Don’t take too long on this: aim to be done packing the base layer by 10 minutes. Pay special attention to the corners of the hexagon- if they look visibly more dry and crumbly than other parts, remove the cement, mix around your existing batch, and apply fresh cement to the area. Lightly tap the bottom of the mold for around 20 seconds, then rap your mixing stick against each side for 5 seconds each. This first layer is especially important because it has to hold onto the decorative aggregate.

Continue layering your mix. I tend to do 3 layers of cement- the base, packed layer, a second layer to bring it to a flat surface, nearing ½ or ⅔ of the mold’s capacity, and a final layer to completely fill the mold to the rim. Make sure to repeat your tapping after each layer. Insert fiberglass if used after the 2nd layer- do not directly put it on the bare aggregate.

SCREEDING & VIBRATING

After completing the final layer, screed the top of the hex with a metal ruler and/or plastic knife, or any other tool you have on hand. Scrape and fill in linear motions right-to-left until you create a smooth, flat surface. Keep filling any obvious low spots and screeding again until you’re happy with the surface. Remember that this side will eventually become the bottom of the finished trivet, so it has to be flat in order to sit balanced on a table. Don't spend like 20 minutes on it though... it will dehydrate and lose strength.

I then used a massage gun to vibrate the mold, once each on opposite sides (making sure not to contact the mold directly), and once on the bottom of the mold. This method was far more effective than hand tapping and I saw lots of air bubbles rise to the surface! I held each for around 45 seconds each, but you can go for even longer if you want to be extra thorough.

FINISHING

Wipe away any concrete that has gotten onto the rim or outside of the mold. I used a paper towel/tissue to gently clean the edges so it looks more presentable. This is important, because you don't want any hardened globs of concrete accidentally fusing your trivet with the mold, and it also looks a lot neater!

Around 6 hours after each hex, when the cement has hardened, I cover it with plastic wrap to trap humidity inside and make sure it doesn’t dry out and become brittle and fragile later on. Cement needs to stay moist in order to sustain the chemical reaction, called hydration, that makes it solid. Trapping moisture helps it cure better and reach its optimal strength. This means that you shouldn't try to dry concrete "to make it set faster." If too much moisture escapes, there may not be enough water left for the chemical reaction to continue properly, making the concrete weaker and more prone to cracking.

Not every hex turned out perfect first try. I did a total of two prototype hexes and the 7 final hexes.

My first prototype was… not exactly successful.

I used 300 grams of dry Sakrete and 40g of water, with a consistency like that of wet sand. Had around ⅓ of the cement left over at the end.

1. DENSE sacrificial sand layer
2. Did NOT pack stones (BIG MISTAKE!!)
3. Did NOT use fiberglass.
4. 48 hours before demolding
5. Did NOT tap mid way through putting cement in

I used a dense layer of decorative white sand across the face of the mold, hoping it would create a white surface and protect the aggregate from the concrete. I had a bunch of leftover cement too.

The first red flag was when I picked it up and it sounded like a damn maraca. After demolding, sand poured everywhere and many of the rocks simply fell out with it.

The problem was that the thick sand layer had prevented the concrete from properly reaching and surrounding the aggregate. Combined with my very dry mix and lack of packing/vibration, there was barely anything holding the rocks in!

Lesson learned: decorative aggregate needs direct contact with well-packed concrete.

Second prototype was better!

I added sparse layers of white sand in between the cement layers, and significantly lessened the amount on the face of the tiles. I also added white sand on the bottom face of the tile (top surface of mold).

1. LIGHT sacrificial sand layer
2. DID pack stones
3. DID NOT use fiberglass.
4. 18 hours before demolding
5. DID tap the mold by hand

This time the aggregate actually stuck and the Voronoi pattern appeared! However, the sand still brushed away and some areas remained dry and crumbly.

At this point I was nervous and ready to see the result, because I didn't have much time left for producing the final 7 pieces, so I demolded after only 18 hours. My fatal mistake...

The concrete felt hard, but hard does not mean strong. The piece survived my careful demolding, but instead decided to break apart when I bumped it by accident!! AAAAAagh!! All that effort wasted just for me to break it on accident.

Prototype 2 taught me two things: ditch the backstabbing sand, and be patient with the concrete!

For the seven production hexes, I combined everything I had learned:

1. NO sacrificial sand layer
2. DID pack stones carefully into every section
3. DID use fiberglass reinforcement
4. CREAMY, hyrdrated concrete
5. Fiberglass reinforcement
6. 48-52 hours before demolding

Adding Fiberglass

My friend, my muse, told me of the wonderful strengths of fiberglass mesh. In case the horrid thing from Prototype 2 happens again, hopefully it would help hold it together more and avoid disaster on these 7 final pieces. I happened to have a roll at home, so I cut seven hexagonal pieces roughly 5 mm smaller than the mold cavity.

I placed the fiberglass after the second layer of concrete, rather than directly against the decorative aggregate. This keeps it embedded inside the concrete instead of becoming exposed on the finished face or edges.

Casting all seven production pieces took me approximately four hours... my back hurts :(

The molds had one major design flaw I hadn’t appreciated enough in Fusion: the outer walls weren't drafted at all, sticking up at an incredibly harsh 90 degree angle.

The Voronoi ribs had a roughly 5° taper, but the outer mold geometry held onto the concrete incredibly tightly. I had forgotten to taper the most crucial part!

I slowly worked around all six corners, alternating between them instead of pulling hard from one side. Tiny separations appeared, but progress was literally millimeter by millimeter.

Eventually I had enough of it, and I kind of just demolished the mold just to free the casting.

After an absurd amount of fighting…

IT CAME OUT WHOLE.

You don't know how stressful it was to pry it out of the mold, feeling ready to break any second. Multiple chunks on the sides did come off, so it was even thinner than planned.

And, most importantly, the aggregate actually stayed embedded.

If I redesigned these molds, adding proper draft to every release surface would be my #1 change.

I used a toothbrush and some water to polish the mold, giving it a shiny finish. I also scraped away small bits that ended up covered by concrete with a screwdriver. A chisel would definitely be ideal, though! This part was SOOO satisfying- I felt like an archeologist digging up fossils! It was almost like Minesweeper, guessing where the rocks were.

I also painted a white border along the edges and the Voronoi ridges to create some more contrast!

And here it is!

Compared with both prototypes, the production process was a massive improvement. The stones and glass remained embedded, the Voronoi patterns actually appeared, and the face-down casting method exposed the decorative aggregate without requiring an orbital sander or grinder.

At the time of writing, the remaining production hexes are still curing in their molds. Rather than repeat the mistake of Prototype 2 just to finish them before my deadline, I’m leaving them alone.

We shall wait!

When I get back, I plan to:

1. Demold and clean the remaining production hexes
2. Assemble and photograph the complete seven-piece Voronoi flower
3. Continue curing the pieces before putting them into regular use
4. Test different arrangements with cups, bowls and pots
5. Add final hero photographs to this Instructable
6. Redesign the molds with significantly better draft
7. Improve the Python generator to eliminate tiny/awkward Voronoi regions
8. Potentially experiment with a concrete sealer/finish
9. Use purpose-made glass aggregate next time instead of spending an afternoon fighting craft beads with a hammer

This project started with me picking up some cool rocks from the beach and somehow turned into Python-generated Voronoi geometry, Fusion modelling, seven 3D-printed molds, shattered and dyed glass, two failed prototypes, fiberglass mesh, and a lot of learning about how concrete actually works.

Prototype 1 taught me the importance of packing and aggregate contact; Prototype 2 taught me patience; and the production pieces taught me just how important mold geometry and draft are.

For now, I have a successful production piece, several more curing, two very educational dead prototypes, and significantly more respect for concrete than I had when I started.

Fun stuff!