Invoke AI, a company specializing in AI-based image creation, has introduced two new features to its platform: the Model Trainer and Control Layers. These features provide users with refined controls in image generation and allow for greater customization and control over the images created. In addition, Invoke has achieved SOC 2 compliance, ensuring a high level of data security.
Enhancing Image Generation with Model Trainer:
The Model Trainer feature offered by Invoke AI allows companies to train custom image generation models using as few as twelve pieces of their own content. This results in more consistent images that align with a developer’s intellectual property (IP). By training the AI to understand specific language and styles, developers can create art that adheres to their studio’s unique style and design features.
Kent Keirsey, the CEO of Invoke AI, explains that the Model Trainer feature enables developers to provide clear instructions to the AI by providing images that represent their desired interpretation. This can be done for general art styles or specific intellectual properties. By training the AI to understand the specific style and subjects, developers can ensure that the generated images align with their vision.
Ensuring Data Security with SOC 2 Compliance:
Invoke AI places a strong emphasis on data security and has achieved SOC 2 compliance. This compliance indicates that the company has passed rigorous tests to ensure the highest level of data security. By implementing robust security measures, Invoke AI reduces the risk of unauthorized use of a developer’s images to create another studio’s intellectual property.
Controlling Individual Parts of an Image with Control Layers:
The Control Layers feature offered by Invoke AI allows users to partition specific areas of an image and assign prompts to those areas. This level of control enables creators to adjust the composition of their image and control individual parts without altering the overall image. For example, a user can paint the upper corner of an image and prompt the AI to add a celestial body specifically in that corner.
With Control Layers, users can refine the prompts attached to each layer and generate images accordingly. The effects of these prompts are localized to the specified part of the image, providing users with precise control over their creations. Additionally, users can upload images to specific layers and choose which aspects they want the AI to retain, such as style, composition, or color.
Integration into Game Development Workflow:
Keirsey acknowledges that many developers are cautious about integrating AI into their workflows due to copyright concerns. However, he believes that as developers can secure copyright for AI-generated content, the use of AI in games will become more prevalent. Invoke AI aims to address this concern by offering tools that demonstrate human expression and provide more ways for developers to exhibit their creativity while protecting their intellectual property.
Conclusion:
Invoke AI’s new features, the Model Trainer and Control Layers, offer developers greater control and refinement over image generation. The Model Trainer allows for the creation of custom image generation models that align with a developer’s IP, while Control Layers enables precise control over individual parts of an image. With SOC 2 compliance, Invoke AI ensures data security and protects against unauthorized use of developers’ images. By addressing copyright concerns and providing tools for human expression, Invoke AI aims to accelerate the integration of AI into game development workflows.
