Prompt para Transformar Foto em Desenho: 14 Estilos para Copiar e Colar

19 min read

Você já viu a foto de perfil dos amigos virando pintura do Studio Ghibli ou desenho de giz de cera de caderno de infância — e quer copiar o prompt certo antes que a trend esfrie. O problema é que boa parte das listas em português usa um prompt em inglês genérico, que troca seu cabelo cacheado por um liso de anime e apaga a sua cor de pele real.

Entre no WhatsApp Novo fluxo de imagens e vídeos todos os dias

Aqui embaixo tem 14 prompts prontos para colar no ChatGPT, no Gemini ou para rodar direto no LightX Image-to-Image. Cada um foi montado para manter o seu rosto reconhecível enquanto muda só o estilo: Ghibli, Pixar, anime, giz de cera, caricatura da profissão, Lego, Minecraft e mais. Todos partem de uma foto sua — nada de gerador que troca sua cara por um rosto genérico de banco de imagem.

Escolha o estilo que bombou no seu feed, cole o prompt, suba a sua foto e manda o resultado no grupo da família ou no Stories.

A fórmula por trás de cada prompt (não pule esta parte)

Todo prompt bom de foto-para-desenho segue a mesma ordem, e é por isso que os prompts genéricos falham: eles pulam etapas. Primeiro vem o travamento de identidade (o rosto, o tom de pele e a textura do cabelo da sua foto não podem mudar). Depois vem a roupa (estilizada, não trocada por outra), a pose, o cabelo no estilo escolhido, o cenário, um lugar real do Brasil como âncora, o enquadramento, a lente, a luz e o acabamento (textura de tinta, giz, grafite). Por fim, toda linha termina com um “Avoid” — a lista do que a IA não pode fazer com seu rosto.

Copie os prompts abaixo exatamente como estão. Eles já vêm nessa ordem.

1. Prompt Ghibli no ChatGPT: selfie em pintura do Studio Ghibli 


A girl dwaing image in ghibli look

Para quem tirou uma selfie sorrindo em casa e quer o efeito que já passou de 1 milhão de usuários testando no mesmo dia — cabelo cacheado vira mecha pintada, sem alisar.

Edit the uploaded photo, keeping the exact same facial identity and likeness of the Brazilian woman (preserve her mixed-race/parda skin tone, face shape, and curly hair texture); transform her into a hand-painted Studio Ghibli-style anime character wearing an earth-toned cotton dress reminiscent of a Ghibli heroine; pose her in three-quarter view with her curly hair lifted by wind and a soft smile looking past the camera; style her curly hair in loose painted strands that keep their natural volume and coil pattern; place her in a lush green hillside setting with red-roofed cottages, fluffy clouds, and dappled sunlight filtering through leaves; anchor the background with a painted version of Corcovado and Tijuca National Forest in the distance; frame as a vertical 4:5 portrait; render with soft painterly brushstrokes, visible watercolor-like texture and no plastic shine; warm golden-hour lighting; finish with hand-painted Ghibli film grain and matte color grade. Avoid: photorealistic skin, glossy 3D render, symmetrical AI face, straightened hair, generic Japanese village without the Brazilian landmark, text or watermark.

2. Prompt para foto estilo Pixar 3D cinematográfico

A boy in the pixar look in black clothes

Para quem tem uma foto de cintura para cima na rua e quer o render 3D com luz de cinema — sem virar boneco laranja.

Edit the uploaded photo, preserving the exact facial identity of the Black Brazilian man (keep his skin tone, fade haircut, and facial structure); dress him in a Pixar-style 3D polo shirt and jeans rendered with physically accurate cloth folds; pose him standing with one foot forward, open genuine smile, looking directly at camera; render his fade haircut with individually sculpted 3D hair strands and controlled specular shine; set him on a blurred commercial street with warm bokeh lights; anchor the background with a stylized 3D version of Avenida Paulista and the MASP building silhouette; frame as a medium shot 4:5; simulate an 85mm cinematic lens with shallow depth of field; use cinematic three-point studio lighting at 45 degrees; finish with Pixar-grade subsurface scattering skin and saturated but natural warm color grade. Avoid: orange/plastic skin tone, porcelain-pale default face, flat cartoon shading, generic mall background, watermark or text.

3. Prompt para transformar foto em anime japonês

A girl drawing in white shirt anime look

Para quem quer virar personagem de cel-shading com olhos grandes, mas mantendo o cacho real do cabelo — não o liso genérico de anime.

Edit the uploaded photo, keeping the same facial likeness of the Brazilian woman with parda skin and curly hair pulled back; convert her into a clean cel-shaded anime character wearing a lightly stylized white shirt over a tank top and ripped jeans redrawn as flat-color line art; pose her seated cross-legged with a relaxed smile and large expressive anime eyes retaining her real eye color; style her curly hair as defined anime-style coiled strands, not generic straight anime hair; place her against a flat stylized pink-sky background with illustrated flowers; anchor the scene with a cel-shaded rendition of Escadaria Selarón’s colorful mosaic steps behind her; frame as a square 1:1 portrait; use crisp vector-line linework with soft cel shading; bright even anime studio lighting; finish with saturated flat colors and clean outline ink. Avoid: straightened or generic ‘anime spike’ hair, pale skin default, blurry linework, western cartoon proportions, text or watermark.

4. Prompt giz de cera: print do Instagram vira caderno de infância

Um desenho de uma garota com visual em tons de ciano

Para quem quer a trend mais recente do feed — o print do perfil virando desenho de giz de cera com cor errada de propósito, tipo caderno de criança.

Edit the uploaded Instagram profile photo, keeping the same facial identity of the Brazilian woman while converting the entire image into a child-like crayon drawing on plain white paper; keep her same black t-shirt but redraw it in wobbly crayon linework with intentionally ‘wrong’ colors like pink skin and blue-tinted hair strands; pose her in a simplified front-facing portrait crop like a school photo; render her long hair as thick, uneven crayon scribbles with childlike volume; surround her with hand-drawn stars, flowers, and doodled clouds on the white paper background; add a small crooked doodle of Copacabana’s beach kiosks in the corner as a local anchor; frame as a square 1:1 crop mimicking a phone screenshot; simulate visible wax texture, uneven pressure strokes, and paper tooth grain; flat, bright nursery-book lighting with no shadows; finish with a warm scanned-paper color cast. Avoid: smooth digital vector lines, photorealistic skin, adult proportions, glossy render, watermark or text overlay.

5. Prompt Gemini: foto vira doodle no caderno

Um esboço de um casal em papel

Para casal que quer aquele rabisco de caneta vermelha e amarela em folha pautada, com a mão do “artista” segurando a caneta na borda.

Edit the uploaded couple selfie, keeping the exact facial identity of both the pardo Brazilian man and the Black Brazilian woman; redraw their same casual t-shirts as a quick red and yellow ballpoint-pen doodle sketch; pose them close together as in a selfie, with a realistic hand holding a red pen visible at the edge of the frame mid-stroke; keep their hairstyles as fast but recognizable pen scribbles that preserve his curls and her natural coily texture; place the doodle on ruled notebook paper with small handwritten notes, comic-style arrows, and an eraser nearby; add a tiny scribbled sketch of Parque Ibirapuera’s lake in the corner of the page; frame as a flat-lay square 1:1 shot of the notebook page; sharp close-up macro lens simulating a phone camera photographing the page; soft desk-lamp lighting; finish with visible blue-lined paper texture and slight pen-ink bleed. Avoid: digital vector cleanliness, missing likeness, generic stock notebook, watermark or text beyond the doodle itself.

6. Prompt Disney clássico 2D (princesa/herói)

Uma garota de vestido azul com visual de princesa da Disney, em estilo de desenho.

Para quem quer o traço pintado à mão da Disney antiga, preservando o cabelo crespo em vez de alisar tipo Rapunzel.

Edit the uploaded photo, preserving the exact facial identity of the Black Brazilian woman including her natural afro-textured hair pattern and skin tone; dress her in a vibrant classic Disney-style princess gown with a fitted waist and flowing hand-painted fabric; pose her in three-quarter view with one hand on her chest and an enchanted upward gaze; style her natural crespo hair in a voluminous Disney-inspired updo that keeps its coily texture, never straightened; place her under soft fairy-tale window light inside a colorful colonial building interior; anchor the background with the colorful colonial facades of Pelourinho, Salvador, visible through an open window; frame as an upper-body 4:5 portrait; use a classic 2D hand-painted cel animation lens equivalent with soft focus edges; warm storybook lighting with a golden rim light; finish with painted film-cel texture and rich saturated color grade. Avoid: straightened Rapunzel-style hair, pale default Disney skin, generic castle background, plastic 3D render, watermark or text.

7. Prompt Os Simpsons: pele amarela e traço chapado

Um menino desenhado no estilo de *Os Simpsons, de pé perto da praia.

Para quem quer o clássico “vira Simpson” — mas com o corte de cabelo real, não um tufo genérico.

Edit the uploaded photo, keeping the recognizable facial identity of the Brazilian man with glasses and his real haircut faithfully translated; give him flat yellow Simpsons-style skin and a solid-orange polo shirt with shorts in flat cartoon color; pose him standing three-quarter with a wide overbite smile and oval cartoon eyes; keep his hair silhouette true to his real short haircut in flat Simpsons style, not a generic tuft; place him on a flat-colored boardwalk with a solid blue sky and simple cartoon houses; anchor the setting with Ipanema’s beachfront calçadão pattern redrawn Springfield-style with the sea visible; frame as a flat 2D full-body shot in 4:5; simulate classic TV cel-animation flat lighting with no gradients; finish with thick black outlines and flat matte color fill. Avoid: 3D shading, gray or brown skin instead of yellow, realistic textures, generic American suburb without the Ipanema boardwalk cue, watermark.

8. Prompt South Park: cabeça redonda, corpo pequeno

Desenho de um garoto de South Park de vestido preto, em pé perto da estrada.

Para quem tem uma foto de corpo inteiro na rua e quer o recorte de papel clássico, cabeça enorme e corpo minúsculo.

Edit the uploaded full-body street photo, keeping the recognizable facial identity of the pardo Brazilian young man; redraw his hoodie and sneakers as flat colored construction-paper cutout shapes typical of South Park style; pose him standing front-facing with arms at his sides and an oversized round head on a tiny cutout body; render his hairstyle as a simple flat silhouette cutout matching his real haircut; place him against a simple cutout-style street background near a stadium; anchor the scene with a paper-cutout rendition of the Maracanã stadium exterior behind him; frame as a full-body flat shot 4:5; use flat construction-paper stop-motion lighting with hard-edged shadows; finish with visible paper-cutout edges and matte flat color. Avoid: 3D rendering, realistic proportions, smooth gradients, generic snowy mountain town, watermark or text.

9. Prompt desenho a lápis realista (grafite)

Uma pessoa fazendo grafite e usando óculos

Para quem quer um retrato sério de ateliê, com hachura visível e grão de papel — sem parecer filtro digital.

Edit the uploaded close-up photo, keeping the exact facial identity of the Brazilian man with his glasses and mustache faithfully rendered; suggest a simple turtleneck collar in loose graphite hatching lines; pose him in a three-quarter close-up with a serious studio-artist expression; render each strand of his hair and mustache with fine cross-hatched pencil lines; place him against a plain off-white paper background with no distracting elements; anchor the setting with a small graphite sketch detail of Parque Lage’s arched courtyard visible faintly in the corner; frame as a tight portrait crop 4:5; simulate a fine 0.5mm graphite pencil on cold-press paper with visible tooth grain; soft single-source studio lighting casting gentle hatched shadows; finish with monochrome graphite tones and visible paper grain. Avoid: color, digital smoothness, photorealistic skin, blank generic background, watermark or text.

10. Prompt para transformar foto em aquarela

A girl in white top aquarela style drwing look

Para quem quer o efeito de tinta molhada escorrendo no papel, com borda solta em vez de contorno duro.

Edit the uploaded photo, keeping the exact facial identity of the parda Brazilian woman; redraw her white tank top with translucent watercolor washes that let the paper show through; pose her standing near a window holding a glass of water, looking gently to the side; render her long hair in soft wet-on-wet watercolor blooms with loose flowing edges; place her beside a window with blurred potted plants and soft indoor light; anchor the background view through the window with the Baía de Todos-os-Santos and the silhouette of Elevador Lacerda in Salvador; frame as a vertical 4:5 portrait; simulate a wet-on-wet watercolor technique on cold-press paper with visible pigment blooms and paper tooth; soft diffused daylight; finish with translucent color layering and light paper texture. Avoid: opaque flat color, digital vector lines, hard edges, generic tropical background without the Salvador landmark, watermark or text.

11. Prompt caricatura da profissão no ChatGPT

a caricatura só fica boa quando trava o seu rosto e puxa os objetos do seu trabalho de verdade.

Para quem viu a trend do G1 de fevereiro — a caricatura só fica boa quando trava o seu rosto e puxa os objetos do seu trabalho de verdade.

Edit the uploaded office or home-office photo, keeping the exact facial identity and curly hair texture of the parda Brazilian woman in her 30s; dress her in an exaggerated version of her profession’s uniform (lab coat, blazer, or apron) with oversized job props like a stethoscope, laptop, or whistle; pose her in three-quarter view with her head enlarged to about one and a half times normal proportion and a confident LinkedIn-style smile; keep her curly hair recognizable but lightly exaggerated in volume for caricature effect; place her at a colorful cartoon desk scattered with objects related to her real job; anchor the background with a window view of Avenida Faria Lima’s skyline in São Paulo; frame as a 4:5 caricature portrait; use a bold colored-pencil-and-marker caricature illustration style; bright even editorial lighting; finish with thick outline ink and flat-to-shaded color fill. Avoid: realistic proportions, generic office stock background, straightened hair, watermark or text.

12. Prompt foto Minecraft (mundo em blocos)

Para quem quer aparecer em voxel com o Pão de Açúcar ao fundo, sem perder a cara nos blocos.

Para quem quer aparecer em voxel com o Pão de Açúcar ao fundo, sem perder a cara nos blocos.

Edit the uploaded beach or overlook photo, keeping the recognizable facial identity of the young Brazilian person; redraw their team jersey as a blocky voxel-textured wool/leather Minecraft skin; pose them standing Steve-like with slightly elongated blocky arms while keeping their real face readable; render their short hair as a voxel block shape that still matches their real haircut silhouette; place them in a blocky beach or jungle biome with a cubic sky; anchor the background with a low-poly voxel version of Pão de Açúcar and its cable car; frame as a full-body 1:1 game-screenshot style shot; simulate a Minecraft in-game camera angle with a subtle game-UI hotbar hint at the bottom edge; flat game-engine lighting; finish with pixel-block texture and saturated game color palette. Avoid: smooth realistic skin, photorealistic textures, missing game blockiness, generic desert biome without the Pão de Açúcar cue, watermark.

13. Prompt foto Lego / boneco na caixa

 

Para quem quer virar minifigura dentro do blister, sem logo registrado, com o Cristo Redentor mini de acessório.

Edit the uploaded full-body photo, keeping the recognizable facial identity of the Brazilian person as a smiling printed minifigure face; redraw their outfit as a glossy plastic Lego minifigure torso reproducing the same shirt print and colors; pose them standing like a minifigure with pin-jointed arms and a printed smile; render their hair as a matching plastic Lego hairpiece molded to imitate their real haircut, including a textured piece for coily or curly hair; place them inside a clear plastic blister display box with a plain yellow cardboard backing, free of any brand logos; anchor the box art with a small toy-scale Cristo Redentor accessory piece beside the figure; frame as a square 1:1 product-style shot; simulate a macro toy-photography lens with shallow depth of field; soft studio product lighting with gentle reflections on the plastic; finish with glossy plastic texture and bright toy-box color grade. Avoid: real skin texture, brand logos or trademarks, cloth fabric folds, watermark or text.

14. Prompt transformar foto em desenho para colorir (estilo Bobbie Goods)

família inteira em line art de caderno de colorir, com o crespo e o cacheado desenhados de verdade

Para quem quer a família inteira em line art de caderno de colorir, com o crespo e o cacheado desenhados de verdade — não um cabelo de boneca.

Edit the uploaded living-room family photo, keeping the recognizable facial identity of the pardo Brazilian father, the child, and the baby as a clean black-and-white line art illustration with no fill color; redraw their same casual outfits as open uncolored outline shapes ready for coloring; pose them grouped together exactly as in the original photo, hugging or sitting close, with no shading; render the father’s and child’s hair textures as accurate crespo/cacheado line art loops, not doll-like curls; place them inside a simple apartment living room outline with an outlined Christmas tree; anchor the scene with a simple line-art silhouette of Cristo Redentor visible through the window outline; frame as a landscape-safe square 1:1 coloring page composition; use single-weight black ink linework with no gradients; flat even lighting implied only by line weight; finish as a pure white page with clean printable outlines. Avoid: any color fill, shading, doll-like generic curls, photorealistic rendering, watermark or text.

Qual foto usar de base (e qual foto não usar)

O resultado depende mais da sua foto de entrada do que do prompt em si. Uma boa base tem seu rosto de frente ou em 3/4, luz natural batendo direto (sem contraluz), e sem óculos escuros cobrindo os olhos.

Faça: use foto recente, nítida, com o cabelo do jeito que ele realmente é (cacheado solto, crespo natural, liso — não importa, desde que apareça). Prefira fundo simples para os estilos de caricatura e desenho a lápis.

Não faça: não use foto com filtro pesado de beleza já aplicado — a IA vai copiar a pele de plástico. Não use foto de grupo apertada demais para os prompts de família; dê espaço para cada rosto aparecer.

Como escrever seu próprio prompt de foto para desenho

Se nenhum dos 16 acima bater exatamente com o que você quer, monte o seu usando a mesma fórmula:

  1. Comece travando a identidade: pele, formato do rosto, tipo de cabelo.
  2. Descreva a roupa que já está na foto, só dizendo em que material ela vira (aquarela, giz, plástico Lego).
  3. Diga o que a pessoa está fazendo (pose), não só “posando”.
  4. Nomeie um lugar real do Brasil — isso muda o resultado mais do que qualquer adjetivo.
  5. Termine sempre com uma linha “Avoid” dizendo o que não pode acontecer com o rosto.
  6. Como rodar esse prompt no LightX Image-to-Image

Abra o LightX e escolha a ferramenta de Image to Image (foto para imagem). Suba a sua foto de base — a mesma que você escolheu na seção acima. Cole o prompt em inglês no campo de texto, mantendo a ordem completa (identidade, roupa, pose, cabelo, cenário, lugar, luz, avoid). Ajuste a força da imagem original para um nível que preserve o rosto sem travar o estilo, gere e compare lado a lado com a foto de entrada antes de postar.

Ferramentas relacionadas no LightX

LightX AI Image Generator — para gerar o cenário do zero antes de editar.

LightX Image to Image — a ferramenta usada em todos os 16 prompts acima.

LightX AI Photo Editor — para ajustar cor e nitidez depois do desenho pronto.

Conclusão

Os 16 prompts acima cobrem a trend que está no seu feed agora — Ghibli, giz de cera, caricatura da profissão — e os clássicos que continuam sendo buscados todo mês — Pixar, anime, Simpsons, lápis. Escolha um, cole no ChatGPT, no Gemini ou no LightX, e mande o antes/depois antes que a próxima trend chegue.

Entre no WhatsApp Novo fluxo de imagens e vídeos todos os dias
Receba prompts diários gratuitos de IA para imagens e vídeos! Participe agora