Runebit Dice: Witchfire Plasma D20 (Free Sample)

Free, commercial-use, drop-in. Twenty photoreal D20 result sprites with a green-cyan plasma core caged inside the gem and witchfire-orange emissive numerals burning across each face. One sprite per face, every face from 1 to 20.

1024 px · 512 px · 256 px   |   transparent PNG

What's in the pack

• Twenty D20 result sprites: one per face, 1 through 20.
• Three sizes per sprite: 1024 px, 512 px, 256 px (square, transparent PNG).
• Photoreal renders. Path-traced in Blender Cycles. Volumetric green-cyan plasma tendrils captured inside a transmissive gem outer, with witchfire-orange inset-etch numerals burning against the cool body.

Use it for

• Indie tabletop / TTRPG game UIs
• Sci-fi and fantasy VTT result graphics (Foundry, Roll20)
• Twitch / YouTube stream overlays for actual-play shows
• Prototyping while you decide if the full polyhedral set is right for your project

Want the full polyhedral set?

The matching Witchfire Plasma polyhedral set drops tomorrow. In the meantime, other photoreal polyhedral packs from the same pipeline are already live:

Already grabbed the Frost Crystal D20, Ember Runes D20, Opal Moonstone D20, or Amethyst Crystal D20 freebies? All five themes share the same render pipeline, so they composite cleanly side by side in a VTT scene or stream overlay.

About Runebit Dice

One-person studio. Every pixel path-traced in Blender Cycles from procedurally generated 3D geometry. No generative-AI imagery, no traced photos. The underlying scenes are regenerated from code, which is why new themes and shapes can be added on demand and stay perfectly consistent across packs.

If you need custom dice visuals for your Kickstarter, BackerKit, itch page, or board game, the same pipeline can produce bespoke renders, transparent PNGs, face-layout sheets, and quote-prep notes.

Credits. Numeral font: Barlow Bold by Jeremy Tribby (SIL OFL 1.1). Font is not redistributed in this pack. It is used only to generate the etched numeral geometry. All artwork is procedurally generated in Blender Cycles from Python/bmesh; no generative-AI models were used.