ファンタジー・SF

Fantasy RPG character card

Turn yourself into a collectible RPG character card, complete with a rendered name plate and class stats.

A painterly fantasy trading card of an elven ranger in a misty forest at dusk, framed by an ornate border with a rendered name and class plate.

プロンプト

A vertical fantasy trading-card layout: dark parchment-textured background glowing faintly at the corners, framed by an ornate bronze filigree border running along all four edges. In the lower two-thirds, a battle-worn elven ranger with braided auburn hair and a leather-and-fur cloak stands ready in a misty pine forest clearing at dusk, a recurve bow slung across her back. Lit by soft blue-hour ambient light with a warm rim light from a distant campfire, rendered as rich painterly illustration with visible brushwork texture. In the card's top banner render the text "SYLVANA THORNWOOD" in a bold engraved fantasy serif font, and in a small plate near the bottom render "RANGER - LV. 24" in the same font; this is a collectible character card for a personal social post, portrait orientation, no watermark, no extra text beyond those two labels.

Original recipe · oimg.ai team · Jun 19, 2026

Original example prompt authored by the oimg.ai team — free to copy and adapt.

モデル: gpt-image-2
アスペクト比: 3:4
カテゴリ: ファンタジー・SF

パーソナルゲームアートSNSイラスト油絵キャラクターファンタジーテキスト

変数を入れ替える

画像ではなく構造を拝借しましょう。ハイライトされた各{{TOKEN}}を自分の被写体に置き換えて、生成します。

A vertical fantasy trading-card layout: dark parchment-textured background glowing faintly at the corners, framed by an ornate {{BORDER_STYLE}} border running along all four edges. In the lower two-thirds, {{SUBJECT}}, {{ACTION_OR_POSE}}, in {{SETTING}}. Lit by {{LIGHTING}}, rendered as rich painterly illustration with visible brushwork texture. In the card's top banner render the text "{{NAME}}" in a bold engraved fantasy serif font, and in a small plate near the bottom render "{{CLASS}} - LV. {{LEVEL}}" in the same font; this is a collectible character card for a personal social post, portrait orientation, no watermark, no extra text beyond those two labels.

効果的な理由

Quoting the exact name and class/level string gives gpt-image-2's strong text rendering a literal target instead of letting it invent illegible fantasy lettering.
Naming "the lower two-thirds" plus a named border style is what makes this read as an actual trading card and not just a portrait with a frame slapped on after.
A named light source (the campfire) doubles as both the in-world explanation for the rim light and the dramatic sub-brightening real card art uses for a heroic silhouette.
Keeping the border, background, and lighting fixed while swapping only subject, name, and class is what makes a batch of these look like one deck instead of unrelated images.

自分だけのバージョンを作る

1プロンプトをコピー
2自分の被写体やディテールに入れ替える
3生成する

注意点

Long names or full sentences in the quoted text tend to misspell; keep {{NAME}} to one or two short words.
Dropping the "lower two-thirds" and border instructions usually collapses the whole thing back into a plain character portrait with no card framing at all.