Shark Chase

Gallop Joy

Royal Blue

India Gate

Ride in Style

Pirate Pet

Ballet

"System Role: You are an expert AI Image Editing Assistant. Your task is to view the input image and generate a single, precise text instruction to edit the image. Logic Requirements: Analyze Identity: Identify the main subject. You must explicitly mention their specific visual traits (species/breed for animals; age, hair color, facial features for humans) in the output to preserve their identity. Determine Posture: If Animal: The instruction must make the animal stand in an anthropomorphic posture, standing fully upright on its hind legs with a vertical torso and forelimbs hanging naturally at its sides. If Human: The instruction must make the human stand naturally, with arms hanging naturally at their sides. Apply Scene & Costume: Wearing a delicate ballet-style little dress and wearing an exquisite princess crown (made of diamonds and of high quality), the scene and background remain unchanged. Style: Ultra-realistic, high-definition details"

Funny Cat

The pet in the picture assumes an anthropomorphic standing posture (with the front two paws raised and the hind legs on the ground; no extra legs should be present). The scene and background remain unchanged.

Beauty

巴西测试-chef cat

Juice Time

Recreate the reference as a playful drink portrait. The uploaded dog wears a green bucket hat and black sunglasses, holding a large cup of orange juice close to the mouth with paws around the cup. Light blue background, bright sunny color palette, front-facing crop with the drink in the lower center, relaxed vacation expression. Preserve each uploaded pet's actual species, breed, coat color, markings, ear shape, eye color, muzzle and facial structure. Do not assume a fixed breed, color, or species from the reference image; adapt the reference gameplay to the uploaded pet. Actively change pose, facial expression, gaze direction, camera angle and styling to match the gameplay reference. No extra animals beyond the requested uploaded pets. No human face unless explicitly requested; anonymous hands are allowed only when the gameplay needs hands. No brand logos, no watermark, no random readable text. Match the gameplay reference very closely in composition, crop, texture, lighting, pet expression, visible face angle, props, background, and image quality. Vertical 3:4, photorealistic unless the reference is a poster or sticker layout.

Magic Class

"Full-bleed vertical 3:4 photorealistic fantasy wizard-school portrait. The uploaded reference image is the exact pet identity and MUST be preserved. Highest priority: keep the same gray-and-white tabby cat from the reference photo: same head shape, same tall ears, same eye color and gaze, same pink nose, same white muzzle, same gray tabby forehead and cheek markings, same facial proportions. Do not replace it with another cat, do not simplify or invent a different face. Only transform costume, pose, and environment. Scene: old magical classroom matching the supplied gameplay reference style: antique wooden desk, arched leaded-glass windows, candlelight, potion bottles, stacked old books, brass scale, crystal ball, warm amber cinematic light, retro film-photo texture. The cat wears a plain black wizard robe and a soft tall black pointed wizard hat. Scarf must match the classroom reference: a knitted school scarf with HORIZONTAL STRIPES ONLY. It is not plaid, not checkerboard, not tartan, not crisscross. Colors are golden yellow and black/dark charcoal, repeated horizontal bands around the neck, visible ribbed wool knit texture, soft knitted fabric, wrapped around the neck with short hanging ends. No readable text, no logos, no watermark"

Hairstyle

Chef Cat

Caught in 4K

Recreate the reference as a humorous pet mugshot. The uploaded cat stands upright in an orange prison-style hoodie, front paws held together with small handcuffs, in front of black-and-white height measurement lines. The cat smirks mischievously with one eye half closed. White background, graphic booking-photo look, centered half-body crop, small decorative numbers allowed but no logos. Preserve each uploaded pet's actual species, breed, coat color, markings, ear shape, eye color, muzzle and facial structure. Do not assume a fixed breed, color, or species from the reference image; adapt the reference gameplay to the uploaded pet. Actively change pose, facial expression, gaze direction, camera angle and styling to match the gameplay reference. No extra animals beyond the requested uploaded pets. No human face unless explicitly requested; anonymous hands are allowed only when the gameplay needs hands. No brand logos, no watermark, no random readable text. Match the gameplay reference very closely in composition, crop, texture, lighting, pet expression, visible face angle, props, background, and image quality. Vertical 3:4, photorealistic unless the reference is a poster or sticker layout.

Floral Gem

Recreate the reference as a pastel-pink high-fashion studio portrait. The uploaded pet is shown from chest up in three-quarter profile, wearing a large glossy pale-pink satin bow tied around the neck. Delicate colorful flower-shaped gems and tiny crystal appliques are arranged symmetrically around both eyes and across the forehead. Smooth pink seamless background, refined softbox lighting, calm elegant expression, exact close crop and styling. The subject remains an animal throughout. No people or human hands. No real brands, logos, watermarks or readable text. Accurately preserve the uploaded pet's species, breed, coat color and markings, ear shape, muzzle and face structure, eye color and body proportions. Keep it instantly recognisable as the same animal; change only pose, scene, clothing and accessories. Vertical 3:4, photorealistic, crisp detail, with the pet as the unmistakable main subject.

Super Sticker

Use the uploaded pet image as the exact identity reference. Preserve the pet's recognizable face, fur color, markings, eye color, muzzle shape and ears while transforming it into the requested gameplay style. Create a vertical 3:4 image with this exact effect: orange background sticker sheet, multiple hand-drawn poses of the same pet, playful everyday gestures, thick white sticker borders, expressive cartoon eyes, bold simple linework, warm retro palette, sticker-pack composition. The generated image must strongly match the gameplay composition and texture described above. Keep the pet angle and amount of face/body visible faithful to the described style. If the style is hand-drawn, watercolor, felt, plush, sticker, toy, comic, or package design, reproduce that material and texture clearly. Keep the pet identifiable as the uploaded pet even when stylized. No logos, no watermark, no Doubao AI text.

Play Time

Say Hi

Hyper-realistic portrait photography, using the texture style of Fujifilm, is very popular on the Instagram social platform. In the picture, one hand holds an iPhone with a transparent frame as the foreground. The phone screen retains the original Apple Instagram social media interface, including the profile (with the uploaded person as the main character), the introduction, and the dynamic posts. The facial features of the image in the uploaded picture remain unchanged. The face shows a sweet and warm smile, waving towards the camera, with half of the body protruding from the phone screen, presenting a realistic 3D visual effect. The outdoor background is the Tuileries Garden in Paris, dotted with retro European-style buildings, green trees and roadside flowers, and green park benches. The soft overcast natural light creates a shallow depth-of-field effect like that of a movie, with exquisite and realistic skin details, 8K ultra-high-definition resolution, and a clear and sharp picture. A simple and clean white hand-drawn graffiti arrow is added to the picture, and some simple white handwritten English annotations are dotted on the clothing details. The graffiti style is soft and delicate, and it never obscures the face and body of the person. The picture composition is fresh and full of modernity. A pure white hand-drawn wood-grain style graffiti, with fine dot-shaped borders surrounding the phone screen, is dotted with some small decorative elements: clouds, star graffiti, heart-shaped graffiti, mini camera icon, decorative lines, lowercase "Hi!" words and soft lines. These graffiti are located at the edges and background positions, never blocking the main content, creating a simple and warm social media atmosphere. The overall weight is between 0.3 and 0.4 grams.

Storm Center

A dramatic cinematic photograph. A high-angle, close-up first-person selfie shot of the subject exactly as depicted in image_17.png (the young adult woman with the black hijab and yellow jacket, preserving her specific face, smile, age, and attire details, including textures and patterns). The subject is dynamically holding the smartphone in a tight, close-up frame, looking directly into the camera with her precise wide smile while being lifted by a powerful wind. The entire surrounding background environment is transformed into a massive, dark, churning tornado twister that spirals from the dark stormy clouds. Swirling in the chaotic wind vortex closer around the subject are the distinct floating cows, small houses, and many birds from image_17.png. The depth of field is shallow, rendering the foreground subject extremely sharp and detailed, with the background tornado swirling close behind her, all under action movie lighting with dynamic composition and a surreal, hyper-realistic digital art style. absolutely NO ALTERATION to the subject's identity, clothing, face, or age, only a specific happy-selfie-pose replacement with a tighter framing.

Drunk Cat

"System Role: You are an expert AI Image Editing Assistant. Your task is to view the input image and generate a single, precise text instruction to edit the image. Logic Requirements: Analyze Identity: Identify the main subject. You must explicitly mention their specific visual traits (species/breed for animals; age, hair color, facial features for humans) in the output to preserve their identity. Determine Posture: If Animal: The instruction must make the animal stand in an anthropomorphic posture, standing fully upright on its hind legs with a vertical torso and forelimbs hanging naturally at its sides. If Human: The instruction must make the human stand naturally, with arms hanging naturally at their sides. Apply Scene & Costume: Wearing a cute top, with the background being the scenery of an amusement park. Style: High-definition realistic, cinematic texture，high-end photography style, the aesthetic charm of fashion photography."

Glam Dance

"Create an AI-generated image based on the provided reference image. The subject's appearance (facial features, hairstyle, clothing, and overall temperament) should remain unchanged, as provided by the user, and the background must stay identical to the one in the reference image without modification. The posture of the subject should closely resemble the gesture in reference image 2, with the following detailed description: both hands are fully open, raised to shoulder height, with the palms facing forward and fingers spread out towards the screen. The left hand is slightly raised, with fingers slightly curled, while the palm remains open. A small amount of yellow paint is applied, evenly spread across the palm and part of the fingertips. The right hand is positioned similarly to the left, slightly more parallel to the body, with less finger curvature, and the palm faces the screen. A small amount of red paint is applied, evenly spread across the palm and fingertips. The paint on both hands should be evenly applied and natural, without excess, maintaining a relaxed and natural gesture. The background should match the environment from the reference image. The resulting image should have a higher resolution and finer textures, ensuring the paint on the hands looks natural and not overdone, while maintaining an artistic and relaxed style."

Beach Sketch

Handheld Doll

[Your reference image URL] Keep the original subject's face, features, hairstyle/outfit (or fur/colors for animals) 100% unchanged, do not modify the real subject. Create a stylized big-head miniature version of the subject: slightly enlarged head, small body, sitting cross-legged on a large realistic human palm. Pose: hands crossed (or paws resting) with a slightly pouty/cute annoyed expression, looking directly at the camera. Add a second gentle human hand playfully pinching the subject's cheek. Add 5 small 3D mini chibi versions of the same subject around the palm, each doing different fun actions (reading, waving, holding signs, dancing, napping), no repeated poses, lively and cute. Add sketchy white hand-drawn doodle elements: simple outlines around the main chibi and mini figures, plus cute casual doodles like hearts, stars, sparkles, small flowers and scribbles scattered lightly around the scene, in a playful hand-drawn style. Lighting: Keep the original picture lighting unchanged, high detail, clean focus. No colored borders, no extra frames, keep the composition clean and whimsical.

Wig Diva

Recreate the reference as a humorous DOGUE glamour cover. The uploaded dog is centered in a tight head-and-neck portrait, wearing a smooth blonde bob wig with bangs, black plush cat-ear headband, pearl necklace and small black hair clip detail. Deep warm brown studio background, large white 'DOGUE' masthead across the top, glossy fashion magazine look. The dog looks directly at camera with a sweet slightly serious diva expression. Preserve the uploaded pet's actual species, breed, coat color, markings, ear shape, eye color, muzzle and facial structure. Do not assume a fixed breed, color, or species from the reference image; adapt the exact gameplay to the uploaded pet. Actively change pose, facial expression, gaze direction, camera angle, wardrobe and styling to match the gameplay reference. Match the gameplay reference very closely in composition, crop, texture, lighting, visible pet angle, props, background, graphic layout and image quality. Vertical 3:4. Keep only the text explicitly requested in the prompt; do not add unrelated logos, watermarks or random words.