I have been running tabletop RPG campaigns for about twelve years now. Mostly D&D, occasionally Pathfinder, more recently a long-running Blades in the Dark campaign that has somehow lasted three years. In that time I have spent more money on character art than I am willing to admit to my partner, and I have still ended up most weeks staring at a half-prepared session with no visual for the character my players are about to meet.
Anyone who has ever run a long campaign knows this problem. Your players will encounter dozens of NPCs over the course of a year. They will retain about four of them in their long-term memory unless you give them something visual to anchor to. The retired sea captain in port becomes a real person the moment your players see his face. Without an image, he is just another voice you do for forty minutes and then forget.
The visuals matter. The visuals have always been the bottleneck.
What “Theatre of the Mind” Quietly Costs You
There is a long-standing argument in the tabletop community that visual aids are unnecessary. “Theatre of the mind” purists argue that the imagination is more powerful than any drawing, and that giving players a specific image of an NPC actually narrows what they can picture. There is some truth to that, but I have come to think it is mostly a rationalization for the fact that visuals were prohibitively expensive for most people for most of the hobby’s history.
In practice, sessions that use even rough character portraits land harder. Players remember NPCs by face. They form relationships with the recurring tavern keeper, the disgraced noble, the suspiciously friendly customs officer, because they have looked at those faces during scenes. They make different choices when an NPC has a face than when an NPC is just a voice you remember to do consistently.
So the question for me as a DM has not been “do visuals help,” because they obviously do. The question has been “can I produce enough visuals, fast enough, cheap enough, to keep up with a campaign that generates new characters every session.”
The Old Ways of Getting Character Art
Before AI tools entered my workflow, my options were the same as everyone else’s, and none of them were really working.
Commissioned art from human artists is the gold standard. A good fantasy portrait artist on Twitter or Reddit will charge anywhere from forty to three hundred dollars per character, with a wait time measured in weeks. For a player character that the group will look at for six months, that is reasonable. For the third innkeeper your players are about to meet this campaign, it is absurd.
Stock art from sites like ArtStation or token packs is the middle option. It is fast and cheap, but the art never quite fits. The character you imagine is a weathered dwarven artificer with one mechanical eye and a leather apron full of pockets. The stock art is a generic dwarf with an axe. Close enough to use, never close enough to feel right.
Hero Forge, Picrew, and similar character creators are the third option. They are great for player characters but they all share a visual style that becomes obvious if you use them for everyone in the campaign. After a while every NPC looks like they came from the same 3D model library, because they did.
The fourth option is drawing the portraits yourself, which works if you can draw. Most DMs cannot. I have shelves of half-finished sketchbooks proving I am one of those most DMs.
My First Sessions Using Nano Banana
The first time I used Nano Banana for a session, I had three new NPCs to introduce that night and forty-five minutes before my players were going to walk through the door. The old me would have either pulled three random stock images and made them work or done the session with no visuals at all.
Instead I described the first NPC — an aging dockmaster in a port city, gray beard, weathered face, military bearing he had not lost since leaving the navy, dressed in a faded blue coat with brass buttons — and got back a portrait that looked like the actual person I had been describing in my head. I did the same for the customs officer and the harbor master. Three portraits in twelve minutes. I had time to actually eat dinner before the session.
What happened at the table that night was the thing that made me a believer. My players asked questions about the dockmaster’s missing brass button, the scar on the customs officer’s jaw, the harbor master’s tired eyes — details that came from the portraits, not from my prep notes. The visuals had given them hooks I had not even planned. The session ran an hour longer than usual because the world felt more populated.
Why Visual Consistency Matters Across a Whole Campaign
A single good portrait is one thing. A campaign full of portraits that all look like they belong in the same world is a different problem.
This is where the workflow with Nano Banana actually pays off, because you can write a campaign-wide style description and apply it to every character you generate. My current campaign uses a “weathered illustrated fantasy” style — slightly painterly, muted color palette, soft brushwork, the kind of art you might find on a 1980s fantasy paperback cover. Every NPC I have generated this year follows that style description. Sit them all next to each other and they look like they belong on the cover of the same book.
That cohesion does something for the players that I did not expect. When every visual matches, the campaign starts to feel like a real setting instead of a series of disconnected scenes. The sea captain and the desert merchant and the mountain hermit all live in the same world because they are all rendered in the same hand. The players stop noticing the visual style and start treating the world as solid.
The NPC Portrait Problem at Scale
The thing that nobody warns new DMs about is the sheer number of NPCs a long campaign produces. My current campaign log lists one hundred and seventeen named NPCs after two years. Some are minor — a guard mentioned once, a barmaid with two lines — but a few dozen are recurring characters the players know and care about.
Before Nano Banana, I would have portrait art for maybe six of those NPCs. The most important ones, the ones I had commissioned at the start of the campaign. The rest would be either stock images that did not really match or no image at all.
Now I have portraits for about ninety of them, generated with Nano Banana over the course of an evening every few weeks. Major NPCs got dedicated time and multiple iterations. Minor NPCs got one quick generation each, often as the session started. The remaining unportrayed NPCs are mostly the ones who only appeared once, where it was not worth the effort.
That coverage changes how the campaign feels. When the players ask “who else is in the tavern tonight,” I can pull up portraits of the regulars rather than wave my hands and describe them. The world has faces.
Player Character Portraits and Iteration
Player character portraits are a slightly different beast, because they need to be exactly right. The player has a clear vision of what their character looks like, and the art has to match that vision or it will sit unused.
The workflow I use here is collaborative. The player describes their character to me. I generate a first draft. We adjust together — make the eyes more piercing, change the cloak from green to deep teal, add the small scar on the left cheek, make the posture more guarded. Most player characters land in five to eight rounds of iteration, which would be expensive and slow with a human artist and is genuinely fast with Nano Banana.
What players have told me they appreciate is the ability to see the same character in multiple contexts. Their cleric in armor at the battlefield. Their cleric in casual clothes at the inn. Their cleric kneeling in prayer at a small altar. Three portraits of the same person, doing different things. That kind of variation used to be unthinkable for player characters because each scene would have been another commission. Now Nano Banana turns it into another half hour of iteration.
Battle Scenes and Set-Piece Moments
The use case I did not expect to lean on as heavily is the dramatic set-piece illustration. When the party finally confronts the campaign villain after a year of buildup, I want to be able to show them the scene. Not a token on a battlemap. An actual illustration of the villain, in the place where the confrontation happens, in the mood the moment deserves.
These set-piece images take longer to get right than NPC portraits. I usually iterate eight to fifteen rounds with Nano Banana, adjusting composition, lighting, the relative positioning of figures, the background detail. But the impact at the table when I drop a finished battle illustration on screen is worth every minute of prep. Players audibly react. They remember the moment for sessions afterward. The campaign log gets a piece of art next to that session’s notes that the group will revisit when they look back on the campaign.
What Nano Banana Cannot Do for a Campaign
Maps are still mostly outside this workflow. Generating a coherent tactical battlemap with grid alignment, clear terrain features, and the right scale for combat is something dedicated map tools handle much better. I use Nano Banana for scene illustrations and atmospheric backdrops, and dedicated map software for the actual playable map.
Established intellectual property is another limit. If your campaign borrows heavily from existing fantasy worlds — official D&D settings, published modules, branded characters — generative tools will not consistently produce content from those worlds, and trying to force them often produces something that is technically off-brand and visually confused. Homebrew settings work much better because there is no “right answer” the tool is trying to match.
And honestly, the iconic NPCs of a campaign still benefit from human-commissioned art. The villain who has been with the players for two years deserves a piece of art that an actual person spent real time on. I still commission those. The difference is that I can now commission them with a clear visual reference instead of a vague description, which makes the commission process faster and the result more accurate.
Why This Shift Quietly Improves Campaigns
The hobby has been gatekept by visual capacity for a long time, in the same way solo game development has been. DMs who could draw, or who could afford to commission, ran more visually rich campaigns. DMs who could not did their best with stock art and imagination, and their campaigns suffered from a thinner world.
That gate is mostly down now. A DM with no art skill and a small budget can run a campaign that looks as visually rich as a published Wizards of the Coast supplement. The world has faces. The villains have presence. The set-piece moments have illustrations. The players remember more, ask more questions, and form deeper relationships with the world.
For most of the hobby’s history this kind of visual richness was a luxury reserved for the small minority of DMs who had either talent or money. Tools like Nano Banana have not made everyone a great DM, but they have removed one of the longest-standing reasons a great DM might have run a less visually rich campaign than they wanted to. That is a quietly important shift, and I think the hobby is better for it.
This content is brought to you by the FingerLakes1.com Team. Support our mission by visiting www.patreon.com/fl1 or learn how you send us your local content here.