4K Seedance 2.0 — референс в видео
Генерируйте кинематографичные видео в 4K по референсным изображениям и текстовым промптам. Сохраняйте идентичность объекта и композицию, добавляя выразительное движение с синхронизацией аудио. Управляйте движением камеры и светом через подробные промпты.
Внутри рабочего процесса
Предварительный просмотр исходного шаблона только для чтения. Откройте его в Cloud, чтобы изучить ноды и внести изменения.
Loading the graph
Open full-size workflow preview- Выход
- Видео
- Автор
- Purz
Вызовите эту модель из своего кода
Приведённый ниже запрос использует тот же подготовленный граф и настройки, что и Playground. Каждая отправка может запустить платный запуск; не повторяйте автоматически отправку, результат которой неизвестен.
What you need
- Runs on
- Comfy Cloud
- Ключ
- COMFY_API_KEY
- Ваши файлы
- Загружаются кодом по их URL
Требуются платный тариф Cloud и кредиты.
# Python 3.10+: pip install comfy-sdk==0.4.0
import os
from comfy_sdk import Comfy
api_key = os.environ.get("COMFY_API_KEY", "").strip()
if not api_key: raise RuntimeError("Set COMFY_API_KEY")
os.environ.setdefault("COMFY_BASE_URL", "https://cloud.comfy.org")
client = Comfy(api_key=api_key)
workflow = client.workflows.from_json({
"344": {
"class_type": "SaveVideo",
"inputs": {
"filename_prefix": "video/Seedance2.0_r2v",
"format": "auto",
"format.codec": "auto",
"video": [
"360",
0
]
},
"_meta": {
"title": "Save Video"
}
},
"352": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_image_2_FILENAME"
},
"_meta": {
"title": "Load Image"
}
},
"353": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_image_3_FILENAME"
},
"_meta": {
"title": "Load Image"
}
},
"354": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_image_4_FILENAME"
},
"_meta": {
"title": "Load Image"
}
},
"356": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_image_1_FILENAME"
},
"_meta": {
"title": "Load Image"
}
},
"360": {
"class_type": "ByteDance2ReferenceNode",
"inputs": {
"model": "Seedance 2.0",
"model.auto_downscale": True,
"model.auto_upscale": False,
"model.duration": 7,
"model.generate_audio": True,
"model.prompt": "Fast-paced fashion showcase filmed as a first-person selfie with a wide-angle lens. The woman from the first reference image holds the camera at arm's length on a windswept coastal bluff at dawn, talking to the lens and laughing as sea mist blows past. [0-2s] She wears the light henley top from the third reference with the wide-leg trousers from the fourth, steps back and turns so both are fully visible. [2-4s] Glitch transition: she now wears the slate-blue utility jacket from the second reference over the same henley, tugging the collar and leaning close to the camera. [4-7s] Glitch transition: she twirls in the full look, jacket open, trousers flaring, then waves goodbye. Garments match their references exactly, consistent in every frame. Quick whip-pans, heather and cliffs behind her, wind, waves and an upbeat electronic beat.",
"model.ratio": "16:9",
"model.reference_images.image_1": [
"356",
0
],
"model.reference_images.image_2": [
"352",
0
],
"model.reference_images.image_3": [
"353",
0
],
"model.reference_images.image_4": [
"354",
0
],
"model.resolution": "4k",
"seed": 838593524,
"watermark": False
},
"_meta": {
"title": "ByteDance Seedance 2.5 Reference to Video (Legacy)"
}
}
})
workflow.set_input("356", "image", client.assets.from_url("https://comfy.org/catalog-media/df5a1664-68f6-4c3c-8b34-02895d86db16/54a073ef-1f26-4eec-b764-94bfd82eed0b"))
workflow.set_input("352", "image", client.assets.from_url("https://comfy.org/catalog-media/7d8b9701-bf08-4fb7-a7df-47f64b430460/a25eac7f-d83f-437d-8035-cb999f16c66f"))
workflow.set_input("353", "image", client.assets.from_url("https://comfy.org/catalog-media/8119a491-8843-40ea-8c25-f777b5191fef/0383b1d3-632a-414d-933a-8b67b5f3314e"))
workflow.set_input("354", "image", client.assets.from_url("https://comfy.org/catalog-media/1ab3d685-99d1-4974-8f60-da41339a13fa/21931d99-5c1f-48b2-9b2e-2c8a6a02d53b"))
# run() also passes api_key to the partner nodes in this workflow.
job = client.run(workflow, api_key=api_key)
for output in job.get_outputs("344"):
output.to_file(output.name)Попробовать пример
Об этом рабочем процессе
Generate cinematic videos from reference images and text prompts in 4K. Preserve subject identity and composition while adding expressive motion with synchronized audio. Control camera movement and lighting through detailed prompts.