ShadeSmith

Inspiration

This project was inspired by our own personal interest in the arts. We have all experienced the difficulties of having to color match while painting and thus wanted a tool to ease the experience. We wanted the app to be able to use the colors that the user has available to best mimic real-life paint mixing. We utilized the phone’s camera to capture colors as accurately as possible and even allow users to review the uploaded color for further accuracy. Shadesmith simplifies the job of an artist by reducing the time and thought spent on paint mixing and color matching, utilizing AI to aid creators rather than replace them.

What it does

Shadesmith is an AI-powered mobile application designed to help artists discover and mix the exact paint colors they want. Users can snap a photo of any object, or input a target shade, and Shadesmith instantly generates a recipe for mixing that color using the paints they already own. If a pigment is missing, the app can also suggest suitable alternatives, saving artists both time and frustration. While its primary use case is for traditional painters, the underlying technology can extend to textile design, digital color matching, and even interior design. By combining practicality with creativity, Shadesmith transforms the way people interact with color in both professional and personal projects.

How we built it

We developed Shadesmith during ShellHacks 2025, aiming to showcase how AI and cloud tools could transform the way artists work with color. The frontend was built in Dart with Flutter, giving the app a clean and responsive interface tailored for creatives. Our backend, written in Python, handled the logic, data flow, and integration with cloud services. On the AI side, we combined Google Cloud Vision to analyze images with Gemini to generate intelligent color suggestions and mixing advice. A major part of our workflow relied on the Google Cloud ADK, which automated several critical steps: converting images into PNG format, transforming images into RGB values, and running color analysis through Cloud Vision. From there, we went further by converting RGB values into CMYK, since digital RGB colors don’t directly translate to the real-world RYB (red–yellow–blue) paint primaries that artists use. This conversion ensured that our app could output recipes that felt practical and usable in real paint mixing. Overall, we built and demonstrated a robust pipeline from image capture to actionable paint recipes, laying a strong foundation for future development.

Challenges we ran into

One of the biggest challenges we faced was achieving consistent color accuracy, since digital RGB and hex values don’t directly translate to physical paint mixtures. We had to experiment with mapping detected digital colors into real-world pigment combinations, including transforming RGB values into CMYK to better approximate traditional paint primaries. Another major challenge was that this was our first time working with Google Cloud services and the ADK, which meant we had to learn how to integrate tools like Cloud Vision, Gemini, and the ADK on the fly while managing authentication and data handling. We also ran into deployment difficulties: although we were able to set up Google Cloud Run and successfully generate a live endpoint, the site only displayed the default hello message, since we couldn’t get Google Cloud to fully clone and link our repository.

Accomplishments that we're proud of

We are proud that Shadesmith, which started as a simple idea at ShellHacks 2025, grew into a functional mobile application in such a short time frame. This was our first time using Google Cloud services and the ADK, and we were able to successfully connect multiple components—Cloud Vision, Gemini, and the ADK—into a working pipeline. One major accomplishment was getting the ADK to automate image conversions, extract RGB values, transform them into CMYK for real-world paint recipes, and return intelligent color suggestions. Another was building a clean, responsive Flutter frontend, which gave the app a polished feel and made it accessible for artists despite the complex AI running behind the scenes. Even though our deployment on Google Cloud Run only reached the default hello message, it was still a milestone in setting up a live environment and moving closer to a production-ready system. Above all, we’re proud that Shadesmith shows genuine value: helping artists save time, reduce wasted paint, and unleash creativity. Knowing that this project has the potential to benefit both hobbyists and professionals gives us confidence in its long-term impact.

What we learned

Throughout the development of Shadesmith at ShellHacks 2025, we discovered how powerful AI can be when applied to creative fields beyond its typical technical or business applications. This was our first time working with Google Cloud services and the ADK, and we gained hands-on experience integrating tools like Cloud Vision and Gemini into a practical workflow. We learned how the ADK could automate essential tasks such as image conversion, color extraction, and transformations between RGB and CMYK, which taught us a great deal about the challenges of mapping digital colors to physical paint systems. On the frontend, working with Flutter sharpened our skills in building user-friendly mobile interfaces for an audience that may not be technically inclined, while still delivering the detail artists need to mix paints accurately. We also gained valuable lessons from deployment—though our Google Cloud Run setup only displayed the default hello page, it gave us critical insight into how cloud deployment pipelines work. Perhaps most importantly, we learned how to translate complex AI reasoning into simple, actionable outputs and how to collaborate effectively across disciplines, bridging technical engineering with artistic needs.

What's next for ShadeSmith

The next step for Shadesmith is full deployment, transitioning from a hackathon prototype into a live, accessible application. We plan to complete Google Cloud Run integration so users can interact with the real pipeline rather than just the default hello page. A major feature on our roadmap is augmented reality (AR) capability, which will let artists preview how a mixed color looks on a canvas, wall, or object directly through their phone. We also want to expand beyond the current three-color mixing limit, allowing artists to blend multiple pigments at once for more accurate and realistic shades. Another critical feature is lighting and hue configuration, giving users the ability to see how colors will appear under different lighting conditions (natural daylight, studio light, warm/cool tones, etc.) and make more informed choices.

Built With

• dart
• gemini
• google-adk
• google-cloud
• google-cloud-vision
• python