
Adobe’s Firefly custom models, now in public beta, promise consistent brand-aligned generative AI outputs. They aim to preserve style details and enhance creative control.
Adobe has introduced custom models for its Firefly generative AI platform, now available in public beta. These models allow creators and brands to generate content aligned with their unique visual identity, addressing concerns about unpredictable AI outputs.
The custom models can preserve details such as stroke weight, color palette, and lighting, making them particularly effective for illustration styles, character designs, and photographic aesthetics. Adobe emphasized that a creator’s style is their signature, and maintaining consistency across campaigns and platforms is essential for brand growth.
Once trained, the models can produce new ideas that remain faithful to a brand’s aesthetic. Adobe noted that this consistency offers a competitive advantage for teams producing high volumes of content. Importantly, the models are private by default, ensuring that generated assets remain proprietary.
Several major organizations, including Tapestry and Deloitte Digital, have already tested the models to scale on-brand creativity. While their long-term effectiveness remains to be fully assessed, the initiative represents a significant step toward giving creatives more control over generative AI outputs.
