Build a consistent storyboard of characters using Amazon Nova on Amazon Bedrock – Part 1

The art of storyboarding is the cornerstone of modern content creation, and weaves its key role through filmmaking, animation, advertising and UX design. Traditionally, creators have relied on hand-drawn sequential illustrations to map their stories, but today's AI Foundation models (FMSs) are changing this landscape. FMSs like Amazon Nova Canvas and Amazon Nova Reel offer the ability to convert text and image input into professional-grade visuals, providing short clips that promise to revolutionize your pre-implementation workflow.

However, this technological leap presents a unique set of challenges. These models excel at quickly generating diverse concepts, a boon for creative exploration, but maintain consistent character design and stylistic consistency throughout the scene. Subtle changes to the structure of the prompt or model can dramatically differ in visual output, potentially destroying narrative continuity, and creating additional work for content creators.

To address these challenges, we have developed this two-part series that explores practical solutions to achieve visual consistency. In Part 1, we share tested prompt patterns that deliver reliable and consistent results with Amazon Nova Canvas and Amazon Nova Reel, digging deep into the rapid engineering and character development pipeline. Part 2 examines techniques such as the finely tuned Amazon Nova Canvas for exceptional visual consistency and accurate character control.

Consistent character design using Amazon Nova Canvas

The foundation for an effective storyboard begins with the establishment of a well-defined character design. Amazon Nova Canvas offers several powerful techniques to create and maintain character consistency throughout the visual story. To help implement these techniques in your own projects, we provided comprehensive code examples and resources to our GitHub repository. We recommend following each step in detail. If you are new to Amazon Nova Canvas, we recommend that you first check the generated images in Amazon Nova to get used to the basic concepts.

Basic text prompts

Amazon Nova Canvas converts textual descriptions into visual representations. Unlike large-scale language models (LLM), image generation models do not interpret commands or engage in inference. It responds best to descriptive captions. Including certain details in the prompt, such as physical attributes, clothing, and styling elements, directly affects the generated output.

for example, “A seven-year-old Peruvian girl with dark hair in two low braids wearing school uniforms.” The model provides clear visual elements for generating early character concepts, as shown in the image in the following example:

Implementing visual styles

Storyboard consistency requires both character functionality and a unified visual style. Our approach separates style information into two key components of the prompt.

• Style description – Opening phrases that define the visual medium (for example, “Graphic novel style illustration”))
• Style details – Closing phrases specifying artistic elements (for example, “Bold linework, dramatic shadows, flat color palette”))

This structured technique allows you to maintain character consistency throughout the storyboard while exploring a variety of artistic styles, including graphic novels, sketches, and 3D illustrations. Below is an example prompt template and some style information you can experiment with.

{style_description} A 7 year old peruvian girl with dark hair in two low braids wearing a
    school uniform. {style_details}
styles = [
    {
        "name": "graphic-novel",
        "description": "A graphic novel style illustation of",
        "details": "Bold linework, dramatic shadows, and flat color palettes. Use
            high contrast lighting and cinematic composition typical of comic book
            panels. Include expressive line work to convey emotion and movement.",
    },
    {
        "name": "sketch",
        "description": "A simple black and white line sketch of",
        "details": "Rough, sketch-like lines create a storyboard aesthetic. High
            contrast. No color",
    },
    {
        "name": "digital-illustration",
        "description": "A 3D digital drawing of",
        "details": "High contrast. Rounded character design. Smooth rendering.
            Soft texture. Luminous lighting",
    },
]

Character variations by seed value

seed The parameters act as a tool to generate character variations while adhering to the same prompt. By keeping and changing the text description constant seed Value, creators can explore multiple interpretations of character design without starting from scratch, as shown in the image in the example below.

Seed = 1	Seed = 20	Seed = 57	Seed = 139	Seed = 12222

Fast compliance control using CFGScale

cfgScale Parameters are another tool for maintaining character consistency and control how strictly Amazon Nova Canvas follows prompts. Working on a scale of 1.1-10, lower values ​​will increase the model's more creative freedom, while higher values ​​will force strict and quick compliance. The default value for 6.5 usually provides the best balance, but it is important to find the right settings, as shown in the following image. A too low value can lead to inconsistent character representations, while a too high value can overemphasize quick elements at the expense of natural constructs.

Seed = 57, CFGScale = 1.1	Seed = 57, CFGScale = 3.5	Seed = 57, CFGScale = 6.5	Seed = 57, CFGScale = 8.0	Seed = 57, CFGScale = 10

Scene integration with consistent parameters

These techniques can be combined to test character consistency in different narrative contexts, as shown in the following example. Maintains consistent input for styles, seedand cfgScaleonly the scene description is changed so that the characters can be recognized throughout the scene sequence.

Seed = 57, CFG_Scale: 6.5	Seed = 57, CFG_Scale: 6.5	Seed = 57, CFG_Scale: 6.5
Graphic novel style illustration of a 7-year-old Peruvian girl with dark hair in two low braids in school uniform Ride your bike on the Mountain Pass Bold linework, dramatic shadows, flat color palette. Use the typical high contrast lighting and cinematic composition for comic book panels. Include expressive line work to convey emotions and movements.	Graphic novel style illustration of a 7-year-old Peruvian girl with dark hair in two low braids in school uniform Walking the path through the tall grass of the Andeans Bold linework, dramatic shadows, flat color palette. Use the typical high contrast lighting and cinematic composition for comic book panels. Include expressive line work to convey emotions and movements.	Graphic novel style illustration of a 7-year-old Peruvian girl with dark hair in two low braids in school uniform Eating ice cream on the beach Bold linework, dramatic shadows, flat color palette. Use the typical high contrast lighting and cinematic composition for comic book panels. Include expressive line work to convey emotions and movements.

Storyboard Development Pipeline

Based on the character consistency techniques we discussed, we can now implement an end-to-end storyboard development pipeline that transforms written scenes and character descriptions into visually coherent storyboards. This systematic approach uses established parameters to describe styles. seed Values, and cfgScale A value that provides character consistency while adapting to various narrative contexts. Below is an example of an explanation of the scene and the character.

"scenes":[
    {
        "description": "Mayu stands at the edge of a mountainous path, clutching
            a book. Her mother, Maya, kneels beside her, offering words of encouragement
            and handing her the book. Mayu looks nervous but determined as she prepares
            to start her journey."
    },
    {
        "description": "Mayu encounters a 'danger' sign with a drawing of a
            snake. She looks scared, but then remembers her mother's words. She takes a
            deep breath, looks at her book for reassurance, and then searches for a stick
            on the ground."
    },
    {
        "description": "Mayu bravely makes her way through tall grass, swinging
            her stick and making noise to scare off potential snakes. Her face shows a
            mix of fear and courage as she pushes forward on her journey."
    }
],
"characters":{
    "Mayu":  "A 7-year-old Peruvian girl with dark hair in two low braids wearing a
        school uniform",
    "Maya":  "An older Peruvian woman with long dark hair tied back in a bun, wearing
        traditional Peruvian clothing"
}

Our pipeline incorporates established best practices using Amazon Nova Lite, using the first craft-optimized image prompt, and is passed to Amazon Nova Canvas for image generation. Depending on the settings numberOfImages Higher (usually 3 variations) while maintaining consistency seed and cfgScale Provides authors with multiple options to maintain consistency of values ​​and characters. I generated an optimized image prompt using the following prompt for Amazon Nova Lite:

Describe an image that best represents the scene described. Here are some examples:
scene: Rosa is in the kitchen, rummaging through the pantry, looking for a snack. She
    hears a strange noise coming from the back of the pantry and becomes startled.
imagery: A dimly lit pantry with shelves stocked with various food items, and Rosa
    peering inside, her face expressing curiosity and a hint of fear.
scene: Rosa says goodbye to her mother, Maya. Maya offers her words of encouragement.
imagery: A wide shot of Rosa's determined face, facing Maya and receiving a small wrapped
    gift.
Only describe the imagery. Use no more than 60 words.
scene: {scene_description}
imagery:

Our pipeline generated the following storyboard panels:

	Mayu stands on the edge of a mountainous road, clutching a book. Her mother, Maya, kneeled beside her, offering words of encouragement and handing her the book. Mayu appears nervous, but he decides that she is preparing to start her journey.
	Mayu encounters signs of “danger” in the snake painting. She looks scary, but then she remembers what her mother said. She takes a deep breath, looks at the book for peace of mind, then looks for a stick on the ground.
	Mayu bravely passes through the tall grass, shaking a stick to make noise and scaring potential snakes. Her face shows a mixture of fear and courage as she advances her journey.

These techniques significantly improve character consistency, but they are not perfect. A thorough examination shows that even images within the same scene show variation in character consistency. Use consistency seed Values ​​help to control these variations, and the techniques outlined in this post significantly improve consistency compared to basic rapid engineering. However, if your use case requires near perfect character consistency, I recommend moving on to part 2. Part 2 will explore advanced fine-tuning techniques.

Video generation for animated storyboards

If you want to convert your storyboards to shorter, animated video clips beyond static scene images, you can use Amazon Nova Reel. Use Amazon Nova Lite to convert image prompts to video prompts, adding subtle and camera movements optimized for your Amazon Nova Reel models. These prompts, along with the original image, act as a creative constraint for Amazon Nova Reel, generating the final animation sequence. Below is an example prompt, with the resulting animated scene in GIF format:

A sunlit forest path with a 'Danger' sign featuring a snake. A 7-year-old Peruvian girl
    stands, visibly scared but resolute. Bold linework, dramatic shadows, and flat color
    palettes. High contrast lighting and cinematic composition. Mist slowly drifting.
    Camera dolly in.

Input image	Output video

Conclusion

This first part of the series explored basic techniques for achieving character and style consistency using Amazon Nova Canvas, from structured prompt engineering to building an end-to-end storyboard pipeline. We have shown how to combine style descriptions. seed Values, and be careful cfgScale Parameter control can greatly improve character consistency in a variety of scenes. We also demonstrated how to integrate Amazon Nova Lite and Amazon Nova Reel to enhance your storyboard workflow and enable both optimized prompt generation and animation sequences.

These techniques provide a solid foundation for consistent storyboard generation, but are not perfect. Variations of the suble can still occur. I recommend continuing with part 2. Here we explore advanced model tweaking techniques that will help you achieve near-perfect character consistency and visual fidelity.

About the author

Alex Barklee I am AWS Senior AI/ML Specialist Solution Architect. She helps customers use AI services to build media solutions using generated AI. Her industry experience includes over the top video, database management systems and reliability engineering.

James Woo He is AWS Senior AI/ML Specialist Solutions Architect, helping customers design and build AI/ML solutions. James' work covers a wide range of ML use cases with a large interest in computer vision, deep learning, and scaling ML across the enterprise. Prior to joining AWS, James was an architect, developer and technology leader for over six years. This included four years in the engineering and marketing and advertising industries.

Vladimir Budyrov He is a leading solution architect at AWS, focusing on agent and generation AI, and software architecture. He leads large Genai implementations, bridging cutting-edge AI capabilities in production-ready business solutions, while optimizing cost and solution resilience.

Nora Shannon Johnson I'm an Amazon Music solution architect focusing on discovery and growth with Ai/ML. In the past, she supported AWS through the development of generation AI prototypes and tools for developers such as financial services, healthcare, and retail. She is an engineer and consultant in a variety of industries, including Devops, Fintech, Industrial AI/ML, and Edtech in the US, Europe and Latin America.

Ehsan Shokrgozar I am a senior solution architect specializing in AWS media and entertainment. He is passionate about helping M&E customers build more efficient workflows. He combines his previous experience as a technical director and pipeline engineer for various animation/VFX studios with his knowledge of building M&E workflows in the cloud to help his customers achieve their business goals.