Skip to main content

Generate Videos

Veo 3.1​

Veo 3.1 is Google's state-of-the-art video generation model, available on FlyMyAI for fast inference.

Try Veo 3.1 on FlyMy.AI

Example: Generate a Video with Veo 3.1​

from flymyai import client, FlyMyAIPredictException

apikey = "fly-***"
model = "flymyai/veo31-fast-generate"

payload = {
    "prompt": "A cinematic shot of a spaceship flying through a nebula, dramatic lighting",
}

fma_client = client(apikey=apikey)
try:
    response = fma_client.predict(
        model=model,
        payload=payload
    )
    print(f"Output: {response.output_data}")
except FlyMyAIPredictException as e:
    print(e)
    raise e
caution

Video generation models can take significantly longer than image models. For long-running operations, consider using predict_async_task with polling instead of synchronous predict.

wan2-img-to-video-lora​

wan2-img-to-video-lora transforms static images into dynamic videos with optional LoRA support.

Try wan2-img-to-video-lora on FlyMy.AI

Example: Convert Image to Video​

from flymyai import client, FlyMyAIPredictException
import pathlib

apikey = "fly-***"
model = "flymyai/wan2-img-to-video-lora"

payload = {
    "prompt": "Morphing into plushtoy",
    "negative_prompt": "bright colors, overexposed, static, blurry details, low quality",
    "width": 1280,
    "height": 720,
    "num_frames": 81,
    "num_inference_steps": 30,
    "guidance_scale": "5.0",
    "input_image": pathlib.Path("/path/to/your/file.jpg"),
    "lora_url": "None",
    "fps": 16,
    "acceleration_factor": "2.5"
}

fma_client = client(apikey=apikey)
try:
    response = fma_client.predict(
        model=model,
        payload=payload
    )
    print(f"Output: {response.output_data['output']}")
except FlyMyAIPredictException as e:
    print(e)
    raise e