GifproGifpro

AI Image Prompt: Witch Reading Spellbook in Castle Library by Candlelight

AI Image Prompt: Witch Reading Spellbook in Castle Library by Candlelight
P

Prompt

candid documentary photograph of a mature witch in her 40s with silver-streaked dark hair, wearing a heavy embroidered velvet robe over a linen tunic, sitting hunched at an antique oak table in an old castle library, reading a thick leather-bound spellbook, holding a brass candlestick with a flickering beeswax candle, wax drippings on wax-sealed grimoires and stacked parchment scrolls, tall stone bookshelves vanishing into shadow, dust motes floating in the narrow beam of candlelight, slight motion blur on the flame, natural ambient light from the single candle, shot on Canon EOS R5, 35mm film grain, natural depth of field, asymmetric composition, real life moment, natural skin texture with pores and subtle imperfections, no face fully visible, profile view, detailed five-fingered hands
画像モデル: FLUX dev (Black Forest Labs)

Stable Diffusion 再現情報 (Replicate)

完全プロンプト (SDに実際に送信したテキスト)
adult, mature features, safe for work, wearing mythical or fantasy attire suited to the scene, anatomically coherent single-creature design, detailed hands with five fingers, natural authentic photograph, candid moment, shot on 35mm film, slight grain, natural depth of field, candid documentary photograph of a mature witch in her 40s with silver-streaked dark hair, wearing a heavy embroidered velvet robe over a linen tunic, sitting hunched at an antique oak table in an old castle library, reading a thick leather-bound spellbook, holding a brass candlestick with a flickering beeswax candle, wax drippings on wax-sealed grimoires and stacked parchment scrolls, tall stone bookshelves vanishing into shadow, dust motes floating in the narrow beam of candlelight, slight motion blur on the flame, natural ambient light from the single candle, shot on Canon EOS R5, 35mm film grain, natural depth of field, asymmetric composition, real life moment, natural skin texture with pores and subtle imperfections, no face fully visible, profile view, detailed five-fingered hands
ネガティブプロンプト
child, children, kid, minor, underage, teen, preteen, toddler, baby, loli, shota, young girl, young boy, little girl, little boy, school uniform on young face, nude, naked, topless, bottomless, nipples, areola, genitalia, penis, vagina, vulva, porn, hardcore, sex act, cum, orgasm, hentai, lewd, nsfw, real person, celebrity, actor, actress, politician, real photograph of celebrity, gore, gory, bloody, decapitation, beheading, torture, self-harm, bestiality, zoophilia, necrophilia, modern bikini, modern swimsuit, modern lingerie, panties, thong, sports bra, shirtless modern man, office worker shirtless, human legs on mermaid, legs and fish tail together, two tails, dragon tail on mermaid, lizard tail on mermaid, reptile tail on mermaid, scales on human legs, torso to tail seam mid-thigh, mermaid with knees, mermaid with feet, lowres, bad anatomy, bad hands, extra fingers, six fingers, seven fingers, fewer fingers, four fingers, fused fingers, twisted fingers, deformed hand, mutated hand, malformed hand, extra limbs, missing limbs, watermark, signature, text

上記の完全プロンプトとネガティブプロンプトを Replicate / ComfyUI / Forge / Automatic1111 等に貼り付けると、近い結果を再現できます (シードや解像度の違いで完全一致はしません)。

このプロンプトは、古城の図書室で蝋燭の灯りを頼りに革装の魔法書を読む魔女の情景を再現します。fantasy pool では現代的すぎる服装が混入しやすいため、ベルベットローブ・麻のチュニック・刺繍などの衣装を明示して「現代の部屋着」に化けるのを防いでいます。光源が蝋燭1つだけのローキー構図は、FLUX の低照度・暖色温度・舞う埃の表現力を確かめるのにも向いています。

AI画像は放っておくと過剰に美しく整いがちなので、”candid documentary photograph” “35mm film grain” “slight motion blur” “natural skin texture with pores” などのリアル系キーワードを混ぜてCG的な完璧さを崩しています。魔女は40代設定にして若年女性偏重を避け、横顔+伏し目で顔を完全に見せない構図にすることで、物語性と陰影の深さを優先しました。

より雰囲気を強めたい場合は、ファンタジー光源系LoRAとフィルムグレイン系LoRAの併用が効果的です。解像度832×1216または1024×1024、steps 28〜35、CFG 4〜6 程度で、蝋燭の暖色を飛ばさずに生成するのがコツです。

Prompt Overview

This prompt recreates the atmosphere of an ancient castle library where a witch studies a leather-bound spellbook by candlelight. The fantasy pool restricts modern clothing, so the robe, tunic, and embroidered velvet are explicitly described to prevent unintended modern outfit artifacts. Candle-heavy scenes with a single light source are excellent for testing how FLUX handles low-key lighting, warm color temperature, and floating dust particles.

Since AI tends to over-polish such scenes, the prompt mixes documentary keywords like "candid documentary photograph", "35mm film grain", "slight motion blur", and "natural skin texture with pores" to break the digital perfection. The witch is described as a mature woman in her 40s to avoid the typical young-female bias, and the profile view keeps the face partially hidden for a more story-driven, atmospheric feel.

For stronger results, a fantasy-atmosphere LoRA and a film grain LoRA can be combined. Generating at 832x1216 or 1024x1024 with 28-35 steps and a low CFG (4-6) is recommended to preserve the candlelight realism without over-saturation.

Was this prompt helpful?

Comments

Comments appear after moderation

Homeヘルプ