Skip to main content
MindStudio
Pricing
Blog About
My Workspace
Image Generation Model

Amazon Nova Canvas

Amazon Nova Canvas is a state-of-the-art image generation model that creates professional-grade images from text or images provided in prompts.

Publisher Amazon
Type Image
Context Window 1,024 tokens
Training Data Late 2024
Price $0.06/image

Text and image-based professional image generation

Amazon Nova Canvas is an image generation model developed by Amazon, released in late 2024. It accepts both text and image inputs to produce images, and is available through Amazon Web Services. The model is designed for generating professional-grade images suitable for commercial and creative workflows.

Unlike large language models, Nova Canvas does not interpret instructions through reasoning. Instead, it maps prompts directly to visual outputs based on how accurately the prompt describes the desired image. This makes prompt construction an important factor in achieving consistent results, and Amazon provides dedicated prompt engineering guidance for the model.

What Amazon Nova Canvas supports

Text-to-Image Generation

Generates images from natural language text prompts. The model maps descriptive text directly to visual output rather than interpreting commands.

Image-to-Image Input

Accepts existing images as part of the prompt input to guide or condition the generated output.

Seed Control

Supports a seed input parameter that allows reproducible image outputs by fixing the random generation state.

Resolution Configuration

Accepts numeric width and height parameters, allowing control over output image dimensions up to supported limits.

Sampling Configuration

Exposes a select input for sampling method and a numeric parameter for controlling generation steps or guidance scale.

Negative Prompting

Supports negative text prompts to specify visual elements that should be excluded from the generated image.

Ready to build with Amazon Nova Canvas?

Get Started Free

Common questions about Amazon Nova Canvas

What is the context window for Amazon Nova Canvas?

Amazon Nova Canvas has a context window of 1024 tokens, which applies to the text prompt input used to describe the desired image.

When was Amazon Nova Canvas trained?

According to the available metadata, Amazon Nova Canvas has a training date of late 2024.

How does prompting for Nova Canvas differ from prompting a language model?

Nova Canvas does not reason or interpret command-based instructions. It transforms prompts into images based on how well the prompt captures the visual essence of the desired output, so descriptive and precise language tends to produce better results.

What input types does Amazon Nova Canvas accept?

Nova Canvas accepts text prompts, image inputs, numeric parameters for dimensions and generation settings, a select input for sampling options, and a seed value for reproducibility.

Where is Amazon Nova Canvas available?

Amazon Nova Canvas is available through Amazon Web Services and can be accessed via AWS services such as Amazon Bedrock.

Parameters & options

Width Number

Each side must be evenly divisible by 16. The aspect ratio must be between 1:4 and 4:1.

Default: 1024 Range: 320–4096 (step 1)
Height Number

Each side must be evenly divisible by 16. The aspect ratio must be between 1:4 and 4:1.

Default: 1024 Range: 320–4096 (step 1)
Quality Select
Default: standard
standardpremium
CFG Scale Number

Specifies how strongly the generated image should adhere to the prompt. Use a lower value to introduce more randomness in the generation.

Default: 6.5 Range: 1.1–10 (step 0.1)
Negative Prompt Text

Description of what to exclude from an image.

Seed Seed

A specific value that is used to guide the 'randomness' of the generation.

Range: 0–858993459

Start building with Amazon Nova Canvas

No API keys required. Create AI-powered workflows with Amazon Nova Canvas in minutes — free.