짧은 답변
두 사람의 얼굴이 인식되는 조화로운 코스튬 장면.
짧은 답변
두 사람의 얼굴이 인식되는 조화로운 코스튬 장면.
Local look lab
Upload a photo you own. The canvas, grade, grain, and caption stay in this browser.
Choose a photo to start. Nothing leaves this browser.
AI 스튜디오
단체 사진보다 두 개의 참조 이미지가 안정적입니다. 정확한 인원 수를 꼭 적으세요.
두 사람의 얼굴이 인식되는 조화로운 코스튬 장면.
AI 스튜디오
Prompt lab
Useful for comparing how different models handle two references.
Use the two uploaded photos as separate identity references. Preserve both people's recognizable faces, skin tones, approximate ages, body proportions, and natural anatomy. Create a coordinated Halloween couple portrait: one person in a weathered western coat, dark hat, and dusty boots; the other in a soft ivory angel costume with feathered wings and a simple silver halo. Keep exactly two people in the image. Place them in a moonlit desert motel courtyard with one warm sign and a dark blue sky. Use photorealistic editorial photography, realistic fabric, and cinematic side lighting. Do not swap faces, merge the two faces, add a third person, duplicate limbs, or add text and watermarks.
Avoid: face swap, merged faces, third person, duplicated limbs, text, watermark
Fast iteration helps when you are testing pose and costume pairing.
Edit the two uploaded photos into a coordinated 1970s disco Halloween portrait. Preserve both subjects' faces, skin tones, ages, and body proportions. Dress one person in a green-and-red sequined jacket with a wide collar; dress the other in a flowing gold jumpsuit with a dark belt. Keep exactly two people. Place them on a small dance floor with a mirrored ball, warm practical lights, and a dark background. Use photorealistic party photography with realistic sequin reflections and gentle film grain. Do not exchange faces, add a third dancer, distort hands, or add readable logos and watermarks.
Avoid: exchanged faces, third dancer, distorted hands, logos, watermark
Character-focused routes are a good fit for coordinated costumes.
Use both uploaded photos as identity references. Preserve both people's recognizable faces, natural skin texture, body proportions, and the original number of people. Create a coordinated sci-fi Halloween portrait: one person in a midnight-blue flight suit with silver piping, the other in a deep-violet observatory coat with a star-map embroidery. Place them in a glass observatory above a dark ocean, with a large moon and faint aurora outside. Keep exactly two people and keep their faces unchanged. Use cinematic science-fiction photography, realistic fabric, and cool rim light. Avoid face swapping, a third person, extra fingers, floating heads, text, and watermarks.
Avoid: face swapping, third person, extra fingers, floating heads, text, watermark
Model routing
| Provider | Best for | Typical route | Docs |
|---|---|---|---|
| ReplicateExperiment-first | Broad catalogue for experiments and batch comparisons | Community and first-party image/video models | Docs |
| falSpeed-first | Fast hosted queues and strong model coverage | FLUX, Kling, and other hosted media models | Docs |
| KieCharacter-first | Nano Banana image edits and Kling video routes | Nano Banana, Kling video | Docs |
| WaveSpeedBudget-first | Low-cost image edits and image-to-video drafts | Seedream edit, FLUX Kontext, Wan image-to-video | Docs |
| SeeAPIRouting layer | One API surface for several model families and fallback routes | Image and video inference endpoints | Docs |
AI 스튜디오
English
FAQ
아니요. 로컬 에디터는 브라우저 안에서 동작합니다. 공급자에게 프롬프트를 붙여 넣으면 해당 서비스 약관이 적용됩니다.
원본 사진에 대한 권리가 있고, 모델 라이선스가 상업적 사용을 허용하며, 필요한 곳에 AI 사용을 밝히는 경우에만 가능합니다. 타인의 얼굴은 동의 없이 쓰지 마세요.
확인한 출처