Invoke has introduced an innovative tool designed for game companies to harness AI for image generation.
Since the release of OpenAI’s ChatGPT-3.5 in November 2022, a variety of image generation tools have emerged. However, Invoke CEO Kent Keirsey emphasizes that their solution is specifically tailored for the gaming sector, prioritizing ethical technology adoption through artist-centric tools, safety and security measures, and accessible entry points.
At Devcom in Cologne, Germany, ahead of the Gamescom expo, I spoke with Keirsey about the intersection of AI and gaming.
Invoke CEO Kent Keirsey
Disclosure: Devcom covered my expenses in Cologne, where I moderated two panels.
GamesBeat: What’s your focus at Invoke?
Kent Keirsey: We specialize in generative AI for game development, particularly in image generation. Our scope covers everything from concept art to marketing assets, supporting the entire pipeline of image creation. Our primary focus is on controllability and customization, allowing artists to sketch and compose their ideas, with AI assisting them to finish the work, rather than relying on a "push button" approach.
We’re actively collaborating with leading game publishers, moving beyond pilot projects to full production deployments. We're implementing tools for managing intellectual property (IP) within our platform.
Our main priority is the artist; our tool is designed to empower rather than replace them. We partnered with the Linux Foundation for the Open Model Initiative, making open-source models available, enabling indie developers and individuals to use and own their assets without competitive disadvantages from AI.
GamesBeat: What type of art does your tool generate?
Keirsey: Currently, it focuses on 2D art. While outputs from 3D models can integrate images or text, the resulting meshes typically require extensive refinement from 3D artists. We aim to bridge the gap between 2D concept art and 3D modeling, allowing 2D artists to preview and iterate before passing their work to 3D modelers.
GamesBeat: How does your tool compare to Pixar’s workflow?
Keirsey: In some ways, yes. Our tool prioritizes individual image creation rather than narrative sequences. We’re model agnostic, enabling clients to train their own models and use them within our platform. Essentially, we allow artists to input their vision into a canvas that AI then interprets into the final asset.
This collaborative process lets artists redefine the interpretation of their prompts, dramatically enhancing control over style and outcomes.
GamesBeat: Do artists prefer drawing first or using text prompts?
Keirsey: Most artists find it easier to express themselves through drawing. Our tool supports hybrid workflows, whether they sketch directly within the platform or upload external sketches. We plan to expand our canvas capabilities soon, enhancing those familiar compositing features.
GamesBeat: How much time do artists save using Invoke?
Keirsey: One artist mentioned that a project that typically takes five to seven business days now takes only four to six hours, thanks to our tool's efficiency. While the time savings are significant, the creative process remains integral, requiring artistic decision-making and iterative work.
GamesBeat: How large is your team currently?
Keirsey: We have nine employees, having launched the company last February, secured $3.7 million in seed funding by June, and released our enterprise product in January. Although gaming is our core focus, we’re exploring opportunities in various industries.
GamesBeat: How competitive is your space?
Keirsey: The market is crowded with image generators. However, what sets us apart is our scalability and our self-hosted, open-source product. We prioritize enterprise clients, ensuring security and compliance, particularly regarding IP protection.
GamesBeat: What legal questions do you face?
Keirsey: The main concerns revolve around data ownership and how we handle customer content. Our clear policy states that we do not train on customer data. Legal reviews often involve copyright issues concerning generated outputs. We believe that with the right parameters, clients can claim copyright over AI-assisted outputs generated through our tools.
GamesBeat: What’s your view on all-in-one AI solution platforms?
Keirsey: I’m skeptical of platforms that claim broad functionality. Each model architecture requires intricate side components for effective control, which we offer. Our tool provides a comprehensive suite that enables seamless workflows tailored to specific project needs, unlike generic options.
GamesBeat: What is your roadmap moving forward?
Keirsey: We’ve successfully launched and are iterating our product based on user feedback. While many artists support our tool, some hesitate to advocate for it publicly due to industry sensitivities. Our focus is on demonstrating how this technology can benefit rather than replace artists, encouraging a more optimistic perception.
GamesBeat: Where is Invoke based?
Keirsey: We operate remotely, with team members in Atlanta, Toronto, and Australia, fostering a diverse and distributed workforce.
Disclosure: Devcom covered my expenses in Cologne, where I moderated two panels.