How to Create Transparent WebM VTT Tokens
A practical workflow for turning character art into transparent WebM tokens that loop cleanly and remain readable in Foundry VTT or Roll20.
Practical answer
A transparent WebM token is not simply a character video with the background removed. It is a small, looping game asset whose silhouette, alpha channel, motion, framing, file weight, and fallback image all need to survive real tabletop play.
Start with a source that the motion model can preserve
Use one clearly separated subject. A portrait, full-body character, top-down token, creature, vehicle, or object can work, but the important anatomy and equipment must already be visible. A model cannot reliably animate a hidden hand, complete a weapon that ends at the crop, or separate two overlapping characters without inventing detail.
For Upload mode, PNG, JPG, and WebP are suitable starting formats. Prefer a transparent image or a simple background with strong contrast. Remove browser chrome, captions, watermarks, and decorative frames before upload. A larger source helps preserve texture, but composition matters more than raw pixel count: the intended subject should be easy to identify without zooming in.
- Use one subject with a readable outline.
- Keep important hands, wings, tails, weapons, and effects visible.
- Leave practical movement room around the parts you want animated.
- Upload only art you are permitted to process.
Choose one creation mode and write one motion idea
Folosa offers three input paths. Build assembles a brief from guided character fields. Prompt is for a direct written design. Upload follows existing art and accepts an optional motion request. Only the active mode is submitted, so text left in another mode does not secretly combine with the task.
For an uploaded character, leave motion blank when identity and pose preservation matter most; the system can use a subtle pose-preserving idle. When you do write a request, describe one action that uses visible parts: “raises the staff, cloth and hair react, then returns to the opening pose” is clearer than a list of unrelated attacks, camera moves, costume changes, and effects.
Motion should be visible at token scale without forcing the character outside the frame. A readable weight shift, weapon-ready movement, controlled spell charge, wing beat, or creature breath is usually more useful than a full-screen cinematic sequence.
Generate for the tabletop you will actually use
Choose either Foundry or Roll20 as the delivery target. Standard VTT produces a compact 512p result intended for live scenes. HD Pro and 4K Master preserve more detail for larger creatures, close review, creator delivery, or archives, but a higher-resolution master is not automatically the best file for a crowded encounter.
A completed Foundry package keeps the transparent WebM, static fallback, module metadata, README, notices, and import guidance together. Roll20 delivery emphasizes the WebM and static asset that can be uploaded directly. Creator users can optionally generate a ProRes 4444 MOV from an eligible completed HD or 4K task for editing; that MOV is a post-production master, not a third tabletop platform.
Check the loop, alpha edge, and frame before delivery
Quality control should catch major delivery failures, not reject every intentional motion. Review the first, middle, and final parts of the loop. The character should remain recognizable, important body parts should not be cut off, and the final movement should return naturally enough that the repeat is not distracting.
Inspect transparency over both light and dark backgrounds. Green residue, opaque rectangles, doubled silhouettes, edge halos, or a frozen first image layered over the animation are delivery defects. Also check that the still image represents the same character and that the downloaded video has no audio track.
| Check | Pass condition |
|---|---|
| Identity | Face, costume, equipment, and silhouette still match the intended character. |
| Framing | Important content remains visible throughout the loop. |
| Transparency | No solid background, green spill, white box, or doubled subject. |
| Loop | The transition back to the opening frame is usable in repeated play. |
| Fallback | The WebP or PNG still is transparent and matches the animated token. |
Import and test at map scale
Do not judge the asset only in a full-screen video player. In Foundry, assign the WebM to the Actor's Prototype Token texture or let Folosa Bridge create the Actor after it verifies the files. In Roll20, upload the WebM, place it on the tabletop, set its grid size, and test it in the browser used for the game.
Run one representative encounter with lighting, maps, and several tokens. If performance drops, use the 512p WebM for live play and retain the larger output as a master. Keep the static fallback beside the animation so the character can still be used in previews, actor portraits, low-motion scenes, or tools that do not play transparent WebM.
FAQ
Does a transparent PNG automatically become a good animated token?
No. Transparency helps extraction, but the source still needs a clear subject, complete visible details, and enough room for the requested movement.
Should I use 4K for every VTT token?
No. Standard 512p is usually the practical live-play choice. Use HD or 4K when the token will be shown larger, reviewed closely, archived, or used in creator editing.
What should I keep after generation?
Keep the transparent WebM, the matching static WebP or PNG, the Foundry ZIP or Roll20 delivery, and the task ID used for support and reuse.