|
Today, we are announcing the vague useability of the Amazon Titan Image Generator v2 model with recent capabilities in Amazon Bedrock. With Amazon Titan Image Generator v2, you can guide image creation using reference images, edit existing visuals, erase backgrounds, produce image variations, and safely customize the model to carry on brand style and subject consistency. This strong tool streamlines laborflows, increases productivity, and transports creative visions to life.
Amazon Titan Image Generator v2 transports a number of recent features in insertition to all features of Amazon Titan Image Generator v1, including:
- Image conditioning – Provide a reference image alengthy with a text prompt, resulting in outputs that chase the layout and structure of the engager-supplied reference.
- Image guidance with color palette – Control exactly the color palette of produced images by providing a catalog of hex codes alengthy with the text prompt.
- Background removal – Automaticassociate erase background from images compriseing multiple objects.
- Subject consistency – Fine-tune the model to carry on a definite subject (for example, a particular dog, shoe, or handbag) in the produced images.
New features in Amazon Titan Image Generator v2
Before getting begined, if you are recent to using Amazon Titan models, go to the Amazon Bedrock console and pick Model access on the bottom left pane. To access the procrastinateedst Amazon Titan models from Amazon, seek access splitly for Amazon Titan Image Generator G1 v2.
Here are details of the Amazon Titan Image Generator v2 in Amazon Bedrock:
Image conditioning
You can engage the image conditioning feature to shape your creations with precision and intention. By providing a reference image (that is, a conditioning image), you can teach the model to center on definite visual characteristics, such as edges, object portrays, and structural elements, or segmentation maps that clarify distinct regions and objects wilean the reference image.
We help two types of image conditioning: Canny edge and segmentation.
- The Canny edge algorithm is engaged to reshift the famous edges wilean the reference image, creating a map that the Amazon Titan Image Generator can then engage to guide the generation process. You can “draw” the set upations of your desired image, and the model will then fill in the details, textures, and final aesthetic based on your guidance.
- Segmentation supplys an even more granular level of regulate. By supplying the reference image, you can clarify definite areas or objects wilean the image and teach the Amazon Titan Image Generator to produce satisfied that aligns with those clarifyd regions. You can exactly regulate the placement and rendering of characters, objects, and other key elements.
Here are generation examples that engage image conditioning.
To engage the image conditioning feature, you can engage Amazon Bedrock API, AWS SDK, or AWS Command Line Interface (AWS CLI) and pick CANNY_EDGE or SEGMENTATION for regulateMode of textToImageParams with your reference image.
"taskType": "TEXT_IMAGE",
"textToImageParams": {
"text": "a cartoon deer in a unprejudicedy world.",
"conditionImage": input_image, # Optional
"regulateMode": "CANNY_EDGE" # Optional: CANNY_EDGE | SEGMENTATION
"regulateStrength": 0.7 # Optional: weight given to the condition image. Default: 0.7
}The chaseing a Python code example using AWS SDK for Python (Boto3) shows how to request Amazon Titan Image Generator v2 on Amazon Bedrock to engage image conditioning.
begin base64
begin io
begin json
begin logging
begin boto3
from PIL begin Image
from botocore.exceptions begin ClientError
def main():
"""
Entrypoint for Amazon Titan Image Generator V2 example.
"""
try:
logging.basicConfig(level=logging.INFO,
establishat="%(levelname)s: %(message)s")
model_id = 'amazon.titan-image-generator-v2:0'
# Read image from file and encode it as base64 string.
with uncover("/path/to/image", "rb") as image_file:
input_image = base64.b64encode(image_file.read()).decode('utf8')
body = json.dumps({
"taskType": "TEXT_IMAGE",
"textToImageParams": {
"text": "a cartoon deer in a unprejudicedy world",
"conditionImage": input_image,
"regulateMode": "CANNY_EDGE",
"regulateStrength": 0.7
},
"imageGenerationConfig": {
"numberOfImages": 1,
"height": 512,
"width": 512,
"cfgScale": 8.0
}
})
image_bytes = produce_image(model_id=model_id,
body=body)
image = Image.uncover(io.BytesIO(image_bytes))
image.show()
except ClientError as err:
message = err.response["Error"]["Message"]
logger.error("A client error occurred: %s", message)
print("A client error ocremedyd: " +
establishat(message))
except ImageError as err:
logger.error(err.message)
print(err.message)
else:
print(
f"Finished generating image with Amazon Titan Image Generator V2 model {model_id}.")
def produce_image(model_id, body):
"""
Generate an image using Amazon Titan Image Generator V2 model on demand.
Args:
model_id (str): The model ID to engage.
body (str) : The seek body to engage.
Returns:
image_bytes (bytes): The image produced by the model.
"""
logger.info(
"Generating image with Amazon Titan Image Generator V2 model %s", model_id)
bedrock = boto3.client(service_name="bedrock-runtime")
hug = "application/json"
satisfied_type = "application/json"
response = bedrock.request_model(
body=body, modelId=model_id, hug=hug, satisfiedType=satisfied_type
)
response_body = json.loads(response.get("body").read())
base64_image = response_body.get("images")[0]
base64_bytes = base64_image.encode('ascii')
image_bytes = base64.b64decode(base64_bytes)
finish_reason = response_body.get("error")
if finish_reason is not None:
lift ImageError(f"Image generation error. Error is {finish_reason}")
logger.info(
"Successbrimmingy produced image with Amazon Titan Image Generator V2 model %s", model_id)
return image_bytes
class ImageError(Exception):
"Custom exception for errors returned by Amazon Titan Image Generator V2"
def __init__(self, message):
self.message = message
logger = logging.getLogger(__name__)
logging.basicConfig(level=logging.INFO)
if __name__ == "__main__":
main()Color conditioning
Most scheduleers want to produce images adhering to color branding guidelines so they seek regulate over color palette in the produced images.
With the Amazon Titan Image Generator v2, you can produce color-conditioned images based on a color palette—a catalog of hex colors supplyd as part of the inputs adhering to color branding guidelines. You can also supply a reference image as input (voluntary) to produce an image with supplyd hex colors while inheriting style from the reference image.
In this example, the prompt depicts:a jar of salad dressing in a rustic kitchen surrounded by recent vegetables with studio weightlessing
The produced image echos both the satisfied of the text prompt and the specified color scheme to align with the brand’s color guidelines.
To engage color conditioning feature, you can set taskType to COLOR_GUIDED_GENERATION with your prompt and hex codes.
"taskType": "COLOR_GUIDED_GENERATION",
"colorGuidedGenerationParam": {
"text": "a jar of salad dressing in a rustic kitchen surrounded by recent vegetables with studio weightlessing",
"colors": ['#ff8080', '#ffb280', '#ffe680', '#e5ff80'], # Optional: catalog of color hex codes
"referenceImage": input_image, #Optional
}Background removal
Whether you’re seeing to composite an image onto a constant color backdrop or layer it over another scene, the ability to immacuprocrastinateedly and exactly erase the background is an vital tool in the creative laborflow. You can instantly erase the background from your images with a one step. Amazon Titan Image Generator v2 can intelligently distinguish and segment multiple foreground objects, ensuring that even intricate scenes with overlapping elements are immacuprocrastinateedly isoprocrastinateedd.
The example shows an image of an iguana sitting on a tree in a forest. The model was able to acunderstandledge the iguana as the main object and erase the forest background, replacing it with a see-thcoarse background. This lets the iguana stand out evidently without the sidetracking forest around it.
To engage background removal feature, you can set taskType to BACKGROUND_REMOVAL with your input image.
"taskType": "BACKGROUND_REMOVAL",
"backgroundRemovalParams": {
"image": input_image,
}Subject consistency with fine-tuning
You can now seamlessly integrate definite subjects into visuassociate captivating scenes. Whether it’s a brand’s product, a company logo, or a beadored family pet, you can fine-tune the Amazon Titan model using reference images to lobtain the distinct characteristics of the chosen subject.
Once the model is fine-tuned, you can spropose supply a text prompt, and the Amazon Titan Generator will produce images that carry on a reliable depiction of the subject, placing it naturassociate wilean diverse, produceive contexts. This uncovers up a world of possibilities for labeleting, advertising, and visual storytelling.
For example, you could engage an image with the caption Ron the dog during fine-tuning, give the prompt as Ron the dog wearing a superhero cape during inference with the fine-tuned model, and get a distinct image in response.
To lobtain, visit model inference parameters and code examples for Amazon Titan Image Generator in the AWS write downation.
Now useable
The Amazon Titan Generator v2 model is useable today in Amazon Bedrock in the US East (N. Virginia) and US West (Oregon) Regions. Check the brimming Region catalog for future refreshs. To lobtain more, verify out the Amazon Titan product page and the Amazon Bedrock pricing page.
Give Amazon Titan Image Generator v2 a try in Amazon Bedrock today, and send feedback to AWS re:Post for Amazon Bedrock or thraw your common AWS Support reach outs.
Visit our community.aws site to discover proset up-dive technical satisfied and to discover how our Builder communities are using Amazon Bedrock in their solutions.
— Channy
