Scale visual production using Stability AI Image Services in Amazon Bedrock

TL;DR

• Stability AI Image Services are now available in Amazon Bedrock, delivering ready-to-use media editing capabilities via the Bedrock API and extending Stability AI’s SD3.5 and Stable Image Core/Ultra models already in Bedrock.
• The nine tools span Edit and Control categories, enabling tasks like Erase Object, Remove Background, Inpaint, Search and Recolor, Search and Replace, Sketch, Structure, Style Guide, and Style Transfer—all accessible within the Bedrock experience.
• A Jupyter notebook walkthrough demonstrates how to run Stability AI Image Services in Bedrock; prerequisites include creating a SageMaker AI notebook instance and verifying the associated execution role permissions.
• The new capabilities are designed to help enterprise teams scale professional-grade visual content production across industries, with potential time and cost savings in media, marketing, retail, gaming, architecture, and education.
• To get started, explore Stability AI models in Amazon Bedrock and the AWS Samples GitHub repo referenced in the post.

Context and background

Stability AI Image Services are now available in Amazon Bedrock, offering ready-to-use media editing capabilities delivered through the Bedrock API. These tools extend on Stability AI’s Stable Diffusion 3.5 models (SD3.5) and Stable Image Core and Ultra models, which are already available in Amazon Bedrock and have set new standards in image generation. The professional creative production process often involves multiple editing steps to achieve the exact output needed. With Stability AI Image Services in Bedrock, users can modify, enhance, and transform existing images without jumping between disparate systems or sending files to external services; everything runs through the same Bedrock experience you’re already using. This post was written with Alex Gnibus of Stability AI. The authors include Isha Dua, Fabio Branco, and Suleman Patel from AWS, and the piece discusses how these tools enable precise creative control to accelerate professional-grade visual content. The business impact can be immediate for teams that produce visual content at scale. The stability AI image services span 9 tools across two categories: Edit and Control. Each tool handles editing tasks that typically require specialized software or manual intervention. The Edit tools simplify complex editing tasks, starting with retouching and background operations. The Erase Object tool removes unwanted elements from images while maintaining background consistency. The Remove Background tool isolates subjects with precision to create clean product listings or varied lifestyle settings. The Search and Recolor and Search and Replace tools target specific image elements for modification, enabling colorway changes or object swaps without new photoshoots. The Inpaint tool fills in or replaces areas based on a mask, providing precise manipulation of image structure and style. The Control category includes Sketch, Structure, Style Guide, and Style Transfer tools to transform concepts into photorealistic visuals, preserve composition while changing subjects, and align imagery with brand styles. To demonstrate Stability AI Image Services in Amazon Bedrock, the post walks through a Jupyter notebook example found in a GitHub repository. Prerequisites include creating a SageMaker AI notebook instance, verifying the associated execution role permissions, running the notebook, and cleaning up resources to avoid ongoing charges. The notebook walkthrough and related materials show how to leverage Bedrock for image editing tasks end-to-end. The availability of Stability AI Image Services in Amazon Bedrock represents an exciting step forward for visual content creation and manipulation, with significant implications for professional teams at enterprises. Media and entertainment creators can rapidly enhance scenes and apply effects; marketing teams can generate campaign variations quickly; retail and ecommerce teams can streamline product photography and digital catalogs; gaming developers can prototype environments more efficiently; architecture firms can visualize designs instantly; and educational institutions can develop more engaging visual content. With these tools, organizations of varying sizes can produce professional-grade visuals more efficiently, expanding creative possibilities while potentially reducing costs and turnaround times. To get started, check out Stability AI models in Amazon Bedrock and the AWS Samples GitHub repo referenced in the post. Alex Gnibus is a Product Marketing Manager at Stability AI; Isha Dua is a Senior Solutions Architect at AWS; Fabio Branco is a Senior Customer Solutions Manager at AWS; and Suleman Patel is a Senior Solutions Architect at AWS. Their bios accompany the article to provide context on the authors and contributors.

What’s new

Stability AI Image Services are now available in Amazon Bedrock, delivering a set of ready-to-use media editing capabilities through the Bedrock API. These tools extend the Bedrock image generation capabilities by enabling editing, modification, and transformation of existing images without leaving the Bedrock workflow. The nine tools span two categories:

• Edit: Erase Object, Remove Background, Search and Recolor, Search and Replace, Inpaint
• Control: Sketch, Structure, Style Guide, Style Transfer Each tool is designed to address specific editing tasks that would normally require separate software or manual effort. The Erase Object tool removes unwanted elements while preserving background continuity. The Remove Background tool isolates subjects to create clean product imagery or adaptable lifestyle scenes. Search and Recolor changes colors on targeted elements, useful for producing colorway variations without reshoots. Search and Replace can swap objects to update seasonal elements or create virtual try-on experiences. Inpaint fills in content in designated areas based on a mask. The Sketch tool converts sketch-style renderings into photorealistic concepts, enabling architects to visualize ideas or apparel brands to generate product mockups from drawings. The Structure tool preserves layout, composition, and spatial relationships while altering subjects or styles. The Style Guide tool derives artistic styles and colors from a reference image to generate new images aligned with brand guidelines. The Style Transfer tool uses features from reference images to transform existing imagery while preserving composition, enabling variations such as transforming modern product photography into traditional or other stylistic expressions. A practical demonstration accompanies the announcement via a Jupyter notebook example in the associated GitHub repository. The walkthrough provides concrete steps to run the sample notebook, illustrating how to apply Stability AI Image Services to real-world scenarios within Bedrock.

Why it matters (impact for developers/enterprises)

The integration of Stability AI Image Services into Amazon Bedrock represents a meaningful shift for teams responsible for visual content at scale. Key implications include:

• Faster, centralized workflows: Creators can modify, enhance, and transform images within Bedrock without exporting assets to external tools or workflows, reducing handoffs and latency.
• Consistent brand and style: Style Guide and Style Transfer tools help marketing and design teams apply brand-aligned visuals across campaigns and catalogs, maintaining visual consistency at scale.
• Expanded creative possibilities: Sketch-to-photorealistic conversions and structure-preserving edits enable rapid prototyping of scenes, product visualizations, and concept art while preserving layout integrity.
• Industry applicability: Media and entertainment workflows, marketing teams, ecommerce product photography, gaming environments, architectural visualizations, and educational content creation can benefit from accelerated iterations and varied visual outcomes.
• Operational efficiency: By consolidating editing tasks in a single Bedrock experience, organizations can streamline production pipelines, potentially lowering costs and reducing turnaround times for visual assets.

Technical details or Implementation

The available tools span two categories, with a total of nine capabilities. A concise overview follows, and a table summarizes tool names and categories for quick reference.

• Create a SageMaker AI notebook instance to run the sample notebook.
• After a few minutes, verify that the notebook instance status is InService and ensure the SageMaker AI execution role has the correct permissions.
• Run the notebook to execute the sample workflow.
• To avoid ongoing charges, stop the ai-images-notebook-instance SageMaker AI notebook instance after completing the walkthrough.
• After a few minutes, the notebook instance transitions from Stopping to Stopped, and SageMaker AI deletes the notebook instance.
• For more details, refer to the Clean up Amazon SageMaker notebook resources guidance in the source. The broader takeaway is that Stability AI Image Services in Bedrock enable enterprises to enhance and edit visuals at scale within a unified cloud workflow, potentially reducing complexity and time-to-delivery for marketing materials, product imagery, and other visual assets.

Key takeaways

• Stability AI Image Services integrate directly into Amazon Bedrock, expanding editing capabilities in the Bedrock ecosystem.
• The nine tools cover both editing and control tasks, enabling end-to-end image refinement without external tools.
• A notebook-based walkthrough demonstrates practical usage, with explicit guidance on provisioning and cleanup of SageMaker resources.
• The solution targets scalable visual production across multiple industries and business functions, offering potential efficiency gains.
• Start by reviewing Stability AI models in Bedrock and the AWS Samples GitHub repository to experiment with the sample workflow.

References

• https://aws.amazon.com/blogs/machine-learning/scale-visual-production-using-stability AI-image-services-in-amazon-bedrock/