It's just a puzzle box!

The Lament configuration - art by artist064 "The original Lament Configuration was created by Simon Sayce for the original Hellraiser movie". I have attempted to remain faithful to the original design, however it is apparent that multiple on-screen props were used with differing configurations, even some faces mirrored. To this end I have settled on the widely used HR1 configuration by Xane.

Another variation of the theme, for lost PLA investment casting. I have my heart set on having this in bronze, maybe with a two tone pewter braze polished flat and shiny. For some reason I've got it in my head this object should be unnaturally pristine with no signs of aging (but that would work, too).

Added a sprue and swapped out the press-fit retaining rings with a 26mm (nominal) external circlips. Drill the sprue out from the bottom with a large drill bit and cut the remaining top connections.

I think the embossed version would be more suitable for bronze as the raised surfaces will be polished. Previously, there was some loss of detail in the top and bottom faces with thin walls so they have been thickened with uniform emboss height of 0.6mm all over.

All the fiddly bits on the top few layers seem to be quite susceptible to over-extrusion. I found best results from dialing back flow rate to the point of causing under-extrusion elsewhere (75%). There are a lot of retractions so settings like ‘Retraction extra prime amount’ will also add unwanted material. A trick to getting this right in Cura is to add a support blocker cube positioned over the top few layers and change it to ‘Modify settings for overlaps’. This allows the flow rate to be tweaked only for the detailed layers as described here.

Shopping list:
27.5mm reduced shank drill bit
2 x 26mm nominal external circlips
Melt furnace on the way
Some experimentation required for ceramic shell using local ingredients...
Pewter in silicone is also an option

Work in progress (as always). Tinkercad has been glitching out on me lately with random defects that are not immediately apparent, so I have switched to Blender for the final boolean - just need to allow a little overlap.