VTT Tokens vs Full Portraits: which one to commission (and when)

A token is a name tag with a face on it. A portrait is the character. Most clients who write me asking which one to commission end up wanting both, but rarely for the reasons they expected.

I get a version of this question almost every week. Someone is two sessions into a new Foundry campaign, the DM keeps pulling up a Google image of a generic half-elf with the wrong sword, and the player wants to fix it. The brief usually arrives in two halves: "I need a token for the VTT" and "also can it be a portrait I can frame." Those are two different paintings. Sometimes one piece can serve both, but the math is rarely as clean as people assume when they write the email. This guide walks through what each piece is actually for, when each one is the right answer alone, when you genuinely need both, and how I scope a token-plus-portrait combo when the character is going to live in two places at once.

Contents

• What a token is actually for
• What a portrait is actually for
• The "I want both" path
• Roll20, Foundry, Owlbear: the platform shape problem
• Why a portrait crop fails as a token
• Token packs and the bystander problem
• Animated tokens, and when they earn the spend
• Common mistakes I see in briefs
• How to decide before you write the brief

What a token is actually for

A token's job is identification at speed. It sits on a battle map at sixty-four to a hundred and twenty-eight pixels, gets dragged across squares, gets rotated by tired DMs at one in the morning, and has to read clearly when six other tokens are stacked around it inside an ooze's grapple. It is not a portrait. It is the visual equivalent of a sticky note with a face drawn on it, and that's not a downgrade, that's the entire design problem.

A token painted as a token has the head filling roughly seventy percent of the circle. The features are sized for legibility at small scale, not for anatomical balance. The palette pushes toward two or three high-contrast colors so the character reads in a quick glance. Hair color, helmet shape, a single visible weapon, a distinctive cloak collar. If your wizard wears red and your bard wears red, one of those tokens is getting a different hat.

Tomasz, a Foundry GM I worked with last October, ran a party of five where three players had picked variants of "human in a brown cloak." He sent me a screenshot from session two with the message, "I literally cannot tell my own NPCs apart from the party at this point." We solved it with token-side decisions, not portrait-side ones. The fighter got a deep ox-blood collar visible above the cloak. The rogue got a hood up with the face mostly shadowed, only a glint at the eye. The ranger got a leather strap across the brow and a green that read as forest at thumbnail. The portraits used totally different palettes when we did them later. But the tokens were solving a battle-map problem first.

What a portrait is actually for

A portrait is the character at full resolution. It is the piece your player opens the character sheet PDF to and pauses on. It is the file they set as their Discord avatar. It is what gets printed and framed when the campaign ends after three years, or what gets shared in the post-session group chat with the message "okay but Sera crushed it tonight." A portrait works at twelve hundred pixels minimum, often higher, and the painting is built for that viewing distance and that attention span.

The compositional choices are different in every direction. A portrait can use a three-quarter view because the eye has time to read it. The hands sit in frame and do storytelling work, gripping a hilt, half-closed in a casting gesture, resting on a holy symbol. The light is allowed to be subtle, the rim barely catching the temple, the shadow side carrying most of the mood. None of that survives a downscale to sixty-four pixels. It isn't meant to. A portrait is a piece you look at slowly. A token is a piece you look at hundreds of times per session but never for more than half a second. Different paintings.

The "I want both" path

Two out of every three character commissions I deliver end up shipping with a matching VTT token attached, even when the original brief only asked for the portrait. Once a player sees their full portrait they want it on the battle map, and the portrait does not crop down well. So we add a token pass to the back end.

The combo path looks like this. I paint the portrait first, full resolution, three-quarter view, all the storytelling choices the brief specified. Then, with the palette and likeness already locked, I paint a separate token from the same character: head and shoulders, designed for the circle, scaled for the platform. The two pieces share the character but not the canvas. The token is faster than a standalone token commission because the design work is already done. The bill goes up by less than people expect, because I'm charging for the second piece of painting work, not for a second design pass.

If you already know you want both, tell me at the start. I'll design the character with both end uses in mind: a head silhouette that reads at thumbnail, a portrait composition that holds a story, a palette that survives both. That's cheaper and faster than commissioning the portrait, sitting with it for a month, and then coming back to ask for a token. Same final result, less back and forth.

Most clients who say "just the token" come back for the portrait within six months. Most clients who say "just the portrait" come back for the token within one. The order doesn't really matter. What matters is whether the painter knows from day one that both pieces exist.

Roll20, Foundry, Owlbear: the platform shape problem

The VTT platforms each handle token shape slightly differently, and that affects how the piece is painted. Skipping the conversation produces tokens that look slightly off in ways your players cannot articulate.

Roll20 applies a circular mask to tokens by default, with an optional square. The mask is hard, no feather, which means anything you paint near the corner of a square canvas disappears entirely once the mask is on. I paint Roll20 tokens on a square canvas with a circular safe zone marked, and I keep the composition inside the circle. Anything in the corners is decorative bleed, not load-bearing.

Foundry VTT supports both circular and square tokens depending on the system. Foundry also handles transparency in PNG tokens well, which opens a door Roll20 doesn't: a token can have a soft edge, a feathered shadow, even a small breakout where a wisp of hair or a weapon tip pokes outside the canonical circle. I use this sparingly. The whole point of a token is fast read, and a fancy silhouette only earns its place if it doesn't slow that read down.

Owlbear Rodeo is the loose, theater-of-the-mind cousin. Tokens are usually square, the platform is forgiving, and the painted edge can be more painterly. If you tell me Owlbear in the brief, I paint with softer rendering at the edge.

If your GM has not told you which platform yet, default to a Roll20-safe token: circular composition, full bleed inside the circle, no critical detail near the corners. That file works everywhere. The companion piece, why a VTT token deserves more than a portrait crop, goes deeper on the composition tricks I use to keep token heads readable at every zoom.

Why a portrait crop fails as a token

People assume you can take a finished portrait, crop in tight on the face, downsize to two hundred and fifty-six pixels, and call it a token. You can. It will look fine to the person who painted it. It will look slightly wrong to everyone else, and they will not know why.

The reason is pixel budget. A portrait painted at twelve hundred pixels has the head filling maybe thirty to forty percent of the canvas. The rest is body, weapon, environment, background. When you crop to the head and downsize, the head occupies the full token canvas, but the detail was painted at the resolution of a head that was a third of the original. The brushwork looks soft. The eyes look slightly blurred. The features are correctly placed but not legibly carved.

A token painted as a token works the other way. The head is sized for the token from the first thumbnail. Every feature is sharpened for the small-scale read. The brushwork is tighter, the contrast pushed. At sixty-four pixels the features still pop. At a hundred and twenty-eight, they look intentional. That's why my token-specific service tier is priced as its own product rather than a free crop of a portrait. Different brief, different file, different painting.

Token packs and the bystander problem

If you are a GM running a long campaign, the question gets more interesting. A single token is one painting. A pack is a design system.

The party pack is usually the first ask. Five characters, matched in style, lighting, palette logic. The work I did on Wren's Curse of Strahd NPC pack is the deeper case study, and the principles transfer cleanly to token-only packs: one light direction across the set, a palette agreed up front, a consistent crop ratio so the head sizes match, and a brand-style ring that signals "this token belongs to the party set."

NPC token packs are where I see GMs underspend in a way they regret. A campaign module gives you twenty to forty named NPCs. Painting all of them is overkill. Painting none and using stock fantasy avatars is the path most GMs take, and it is the reason their players spend session three asking who Ismark is again. The middle path I recommend: a token pack of the eight to twelve named NPCs the players will interact with most, painted in a unified style, plus a small set of generic "city guard" tokens for bystanders.

The bystander question is worth pausing on. A named NPC needs a face. The third town guard from the left does not. Painting bystander tokens at the same fidelity as named NPCs is a budget mistake I see GMs make often. The fix is to commission a small pack of three or four generic tokens in the same palette and lighting as the named set, with the faces deliberately less defined, the helmets covering more of the head, the body language interchangeable. The players never ask "wait, who is that guard," because the visual language says "generic city guard, do not bother."

• Party pack: 4-6 tokens of player characters, matched lighting, palette agreed up front, identical head crop ratio
• Named NPC pack: 8-12 tokens of the campaign's recurring antagonists, allies, faction leaders
• Bystander pack: 3-4 deliberately generic tokens that fill the "anonymous human" role, reusable across sessions
• Token-plus-portrait combo: full portrait for the player character, matched token for the battle map, single design pass

That stack of four covers about ninety percent of the campaigns I work on.

Animated tokens, and when they earn the spend

Animated tokens are the once-in-a-while justification. A subtle loop, three to five seconds, of a character's cloak shifting or a flame flickering on a torch. Foundry supports them natively. Roll20 supports them with the Plus tier and a webm file.

My honest take: most animated tokens are wasted budget. A token spends most of its life sitting still on a battle map, ignored, while the GM describes what is happening in narration. The animation rarely catches the eye unless the player is looking at exactly that token at exactly that moment, and the cost of the animation pass is significant compared to the static piece.

Two cases where it earns its keep. The first is a recurring antagonist the players will see across an entire campaign arc. Strahd's eyes shifting, slowly, every few seconds. A lich's pupils dilating when the players cast detection. The animation does not need to be flashy. It needs to do one small thing, repeatedly, so when the players see the token, the character feels alive in the room.

The second is a player character token for a player who really, really wants one. Some players spend more time looking at their own token than anyone else's, because that is the avatar of their participation in the story. It is a luxury, but luxury is a legitimate reason to commission art. If you want it, ask for it. I just want you to know what you are paying for before you sign off. For most parties: static tokens for everyone, one animated token for the primary antagonist, reassess after a few sessions.

Common mistakes I see in briefs

Three patterns I see often enough to flag.

The first is assuming the portrait will crop into a token. It will not, not well. If you want both, brief both from the start. The combo costs less when planned together than when ordered as two separate commissions six weeks apart.

The second is over-specifying the token. A token does not need a backstory. It needs three visual hooks: a distinct silhouette, a dominant color, and a face that reads at thumbnail. If your token brief is three paragraphs long, you are overthinking it. The work on writing a commission brief covers this in more detail.

The third is forgetting to mention the VTT platform. I have painted Roll20-safe tokens that the client then loaded into Foundry and complained the edges looked too tight. I have painted Foundry-soft tokens that got loaded into Roll20 and the soft edge got eaten by the hard mask. One line in the brief, "this is going on Roll20" or "this is for Foundry, Pathfinder 2e," saves a revision pass.

A fourth I will mention briefly: confusing style choice with piece type. Painterly versus anime versus lineart applies to both the token and the portrait. Token versus portrait versus both is a separate decision. The sketch-to-color-to-final process walkthrough covers the style side; the pricing guide covers the piece-type side.

How to decide before you write the brief

The short version, the one I would give you on a video call before your session started.

Portrait only if the character lives primarily on a wall, a character sheet, a Discord avatar, or in your head. No VTT. The piece exists to make the character feel real to you and to the people you share it with.

Token only if the character lives primarily on a battle map, the GM has asked for a unified pack across the party, or you want the at-table identification piece without the full portrait spend. Tokens are also the right starting point if you are not sure yet whether the character will survive past level five. Commission the portrait once they have earned it.

Both if the character is a long-term commitment, the party is doing weekly sessions, and you want the same face on the battle map and on the wall. Most long-running player characters land here. So do most GM-side primary villains, and the named NPCs the players will see again and again.

If your campaign is a published module with a known cast, a GM-side NPC pack is usually the single most useful commission you can make. Eight to twelve named characters, matched in style, with tokens for each plus full portraits for the ones the players will encounter dramatically. For party portraits where the whole table wants the same painted treatment together, the party portrait service is the matching tier on the player side.

If you want the genre-specific version of this conversation, the D&D 5e commission guide covers how it shapes up for a standard fantasy table, the cyberpunk character guide walks through how neon palettes and cybernetics change the token-read problem, and the sci-fi commission guide handles helmets, hardsuits, and the "helmeted hero" issue where a token has to read without a visible face.

If you already know which piece you need, the order form has a field for piece type at the top. If you are still deciding, the portfolio is the cleanest way to see how tokens and portraits sit next to each other in the studio's house style. Either way, the sooner you write the one-line brief, the sooner the character has a face the table can actually point to.