Amazon Nova Reel 1.1: That includes as much as 2-minutes multi-shot movies

At re:Invent 2024, we introduced Amazon Nova fashions, a brand new era of basis fashions (FMs), together with Amazon Nova Reel, a video era mannequin that creates quick movies from textual content descriptions and non-obligatory reference photographs (collectively, the “immediate”).

At this time, we introduce Amazon Nova Reel 1.1, which gives high quality and latency enhancements in 6-second single-shot video era, in comparison with Amazon Nova Reel 1.0. This replace allows you to generate multi-shot movies as much as 2-minutes in size with constant model throughout pictures. You’ll be able to both present a single immediate for as much as a 2-minute video composed of 6-second pictures, or design every shot individually with customized prompts. This provides you new methods to create video content material by Amazon Bedrock.

Amazon Nova Reel enhances inventive productiveness, whereas serving to to scale back the time and value of video manufacturing utilizing generative AI. You need to use Amazon Nova Reel to create compelling movies in your advertising and marketing campaigns, product designs, and social media content material with elevated effectivity and artistic management. For instance, in promoting campaigns, you possibly can produce high-quality video commercials with constant visuals and timing utilizing pure language.

To get began with Amazon Nova Reel 1.1 
In case you’re new to utilizing Amazon Nova Reel fashions, go to the Amazon Bedrock console, select Mannequin entry within the navigation panel and request entry to the Amazon Nova Reel mannequin. If you get entry to Amazon Nova Reel, it applies each to 1.0 and 1.1.

After gaining entry, you possibly can attempt Amazon Nova Reel 1.1 straight from the Amazon Bedrock console, AWS SDK, or AWS Command Line Interface (AWS CLI).

To check the Amazon Nova Reel 1.1 mannequin within the console, select Picture/Video beneath Playgrounds within the left menu pane. Then select Nova Reel 1.1 because the mannequin and enter your immediate to generate video.

Amazon Nova Reel 1.1 presents two modes:

• Multishot Automated – On this mode, Amazon Nova Reel 1.1 accepts a single immediate of as much as 4,000 characters and produces a multi-shot video that displays that immediate. This mode doesn’t settle for an enter picture.
• Multishot Handbook – For many who want extra direct management over a video’s shot composition, with handbook mode (additionally known as storyboard mode), you possibly can specify a novel immediate for every particular person shot. This mode does settle for an non-obligatory beginning picture for every shot. Photos should have a decision of 1280×720. You’ll be able to present photographs in base64 format or from an Amazon Easy Storage Service (Amazon S3) location.

For this demo, I take advantage of the AWS SDK for Python (Boto3) to invoke the mannequin utilizing the Amazon Bedrock API and StartAsyncInvoke operation to begin an asynchronous invocation and generate the video. I used GetAsyncInvoke to verify on the progress of a video era job.

This Python script creates a 120-second video utilizing MULTI_SHOT_AUTOMATED mode as TaskType parameter from this textual content immediate, created by Nitin Eusebius.

import random
import time

import boto3

AWS_REGION = "us-east-1"
MODEL_ID = "amazon.nova-reel-v1:1"
SLEEP_SECONDS = 15  # Interval at which to verify video gen progress
S3_DESTINATION_BUCKET = "s3://<your bucket right here>"

video_prompt_automated = "Norwegian fjord with nonetheless water reflecting mountains in good symmetry. Uninhabited wilderness of Big sequoia forest with daylight filtering between huge trunks. Sahara desert sand dunes with good ripple patterns. Alpine lake with crystal clear water and mountain reflection. Historical redwood tree with detailed bark texture. Arctic ice cave with blue ice partitions and ceiling. Bioluminescent plankton on seaside shore at evening. Bolivian salt flats with good sky reflection. Bamboo forest with tall stalks in filtered mild. Cherry blossom grove towards blue sky. Lavender discipline with purple rows to horizon. Autumn forest with pink and gold leaves. Tropical coral reef with fish and colourful coral. Antelope Canyon with mild beams by slender passages. Banff lake with turquoise water and mountain backdrop. Joshua Tree desert at sundown with silhouetted bushes. Iceland moss- coated lava discipline. Amazon lily pads with good symmetry. Hawaiian volcanic panorama with lava rock. New Zealand glowworm cave with blue ceiling lights. 8K nature pictures, skilled panorama lighting, no motion transitions, good publicity for every surroundings, pure coloration grading"

bedrock_runtime = boto3.consumer("bedrock-runtime", region_name=AWS_REGION)
model_input = {
    "taskType": "MULTI_SHOT_AUTOMATED",
    "multiShotAutomatedParams": {"textual content": video_prompt_automated},
    "videoGenerationConfig": {
        "durationSeconds": 120,  # Have to be a a number of of 6 in vary [12, 120]
        "fps": 24,
        "dimension": "1280x720",
        "seed": random.randint(0, 2147483648),
    },
}

invocation = bedrock_runtime.start_async_invoke(
    modelId=MODEL_ID,
    modelInput=model_input,
    outputDataConfig={"s3OutputDataConfig": {"s3Uri": S3_DESTINATION_BUCKET}},
)

invocation_arn = invocation["invocationArn"]
job_id = invocation_arn.cut up("/")[-1]
s3_location = f"{S3_DESTINATION_BUCKET}/{job_id}"
print(f"nMonitoring job folder: {s3_location}")

whereas True:
    response = bedrock_runtime.get_async_invoke(invocationArn=invocation_arn)
    standing = response["status"]
    print(f"Standing: {standing}")
    if standing != "InProgress":
        break
    time.sleep(SLEEP_SECONDS)

if standing == "Accomplished":
    print(f"nVideo is prepared at {s3_location}/output.mp4")
else:
    print(f"nVideo era standing: {standing}")

After the primary invocation, the script periodically checks the standing till the creation of the video has been accomplished. I cross a random seed to get a unique consequence every time the code runs.

I run the script:

Standing: InProgress
. . .
Standing: Accomplished
Video is prepared at s3://<your bucket right here>/<job_id>/output.mp4

After a couple of minutes, the script is accomplished and prints the output Amazon S3 location. I obtain the output video utilizing the AWS CLI:

aws s3 cp s3://<your bucket right here>/<job_id>/output.mp4 output_automated.mp4

That is the video that this immediate generated:

Within the case of MULTI_SHOT_MANUAL mode as TaskType parameter, with a immediate for multiples pictures and an outline for every shot, it isn’t mandatory so as to add the variable durationSeconds.

Utilizing the immediate for multiples pictures, created by Sanju Sunny.

I run Python script:

import random
import time

import boto3


def image_to_base64(image_path: str):
    """
    Helper perform which converts a picture file to a base64 encoded string.
    """
    import base64

    with open(image_path, "rb") as image_file:
        encoded_string = base64.b64encode(image_file.learn())
        return encoded_string.decode("utf-8")


AWS_REGION = "us-east-1"
MODEL_ID = "amazon.nova-reel-v1:1"
SLEEP_SECONDS = 15  # Interval at which to verify video gen progress
S3_DESTINATION_BUCKET = "s3://<your bucket right here>"

video_shot_prompts = [
    # Example of using an S3 image in a shot.
    {
        "text": "Epic aerial rise revealing the landscape, dramatic documentary style with dark atmospheric mood",
        "image": {
            "format": "png",
            "source": {
                "s3Location": {"uri": "s3://<your bucket here>/images/arctic_1.png"}
            },
        },
    },
    # Example of using a locally saved image in a shot
    {
        "text": "Sweeping drone shot across surface, cracks forming in ice, morning sunlight casting long shadows, documentary style",
        "image": {
            "format": "png",
            "source": {"bytes": image_to_base64("arctic_2.png")},
        },
    },
    {
        "text": "Epic aerial shot slowly soaring forward over the glacier's surface, revealing vast ice formations, cinematic drone perspective",
        "image": {
            "format": "png",
            "source": {"bytes": image_to_base64("arctic_3.png")},
        },
    },
    {
        "text": "Aerial shot slowly descending from high above, revealing the lone penguin's journey through the stark ice landscape, artic smoke washes over the land, nature documentary styled",
        "image": {
            "format": "png",
            "source": {"bytes": image_to_base64("arctic_4.png")},
        },
    },
    {
        "text": "Colossal wide shot of half the glacier face catastrophically collapsing, enormous wall of ice breaking away and crashing into the ocean. Slow motion, camera dramatically pulling back to reveal the massive scale. Monumental waves erupting from impact.",
        "image": {
            "format": "png",
            "source": {"bytes": image_to_base64("arctic_5.png")},
        },
    },
    {
        "text": "Slow motion tracking shot moving parallel to the penguin, with snow and mist swirling dramatically in the foreground and background",
        "image": {
            "format": "png",
            "source": {"bytes": image_to_base64("arctic_6.png")},
        },
    },
    {
        "text": "High-altitude drone descent over pristine glacier, capturing violent fracture chasing the camera, crystalline patterns shattering in slow motion across mirror-like ice, camera smoothly aligning with surface.",
        "image": {
            "format": "png",
            "source": {"bytes": image_to_base64("arctic_7.png")},
        },
    },
    {
        "text": "Epic aerial drone shot slowly pulling back and rising higher, revealing the vast endless ocean surrounding the solitary penguin on the ice float, cinematic reveal",
        "image": {
            "format": "png",
            "source": {"bytes": image_to_base64("arctic_8.png")},
        },
    },
]

bedrock_runtime = boto3.consumer("bedrock-runtime", region_name=AWS_REGION)
model_input = {
    "taskType": "MULTI_SHOT_MANUAL",
    "multiShotManualParams": {"pictures": video_shot_prompts},
    "videoGenerationConfig": {
        "fps": 24,
        "dimension": "1280x720",
        "seed": random.randint(0, 2147483648),
    },
}

invocation = bedrock_runtime.start_async_invoke(
    modelId=MODEL_ID,
    modelInput=model_input,
    outputDataConfig={"s3OutputDataConfig": {"s3Uri": S3_DESTINATION_BUCKET}},
)

invocation_arn = invocation["invocationArn"]
job_id = invocation_arn.cut up("/")[-1]
s3_location = f"{S3_DESTINATION_BUCKET}/{job_id}"
print(f"nMonitoring job folder: {s3_location}")

whereas True:
    response = bedrock_runtime.get_async_invoke(invocationArn=invocation_arn)
    standing = response["status"]
    print(f"Standing: {standing}")
    if standing != "InProgress":
        break
    time.sleep(SLEEP_SECONDS)

if standing == "Accomplished":
    print(f"nVideo is prepared at {s3_location}/output.mp4")
else:
    print(f"nVideo era standing: {standing}")

immediate = "A easy pink triangle transforms by geometric shapes in a journey of self-discovery. Clear vector graphics towards white background. The triangle slides throughout damaging house, morphing easily right into a circle. Pan left because it encounters a blue sq., they carry out a geometrical dance of shapes. Monitoring shot as shapes mix and separate in mathematical precision. Zoom out to disclose a sample shaped by their actions. Restricted coloration palette of major colours. Exact, mechanical actions with good geometric alignments. Transitions use easy wipes and geometric form reveals. Flat design aesthetic with sharp edges and stable colours. Closing scene exhibits all shapes combining into a fancy mandala sample."