The Israeli video personalization company is unveiling Strata, claiming it produces separate, editable layers for text, animation, footage, and actors instead of a single flat file. This challenges the limits of diffusion-based video generators.
Currently, AI video models create a flat file that limits editing, forcing users to start over for any changes. This has kept AI video away from professional production, where videos are built in layers for seamless editing until the final render.
Idomoo, specializing in video personalization, introduces Strata, a generative AI model for layered video output. Unlike pixel generation, as explained by co-founder Danny Kalish, Strata generates structure—offering independent layers with typography, animation, motion paths, and audio, creating a “production-ready video blueprint.”
Standard diffusion models condense everything into a single tensor during generation, embedding relationships in pixels. Strata tackles a different problem, designing compositions with placement, contrast, movement, timing, and brand compliance across all layers simultaneously. The output can be edited at the layer level, akin to Adobe After Effects.
Strata integrates with Lucas, Idomoo’s AI video agent on its Next Generation Video Platform, incorporating brand awareness. Lucas extracts “Brand DNA” from approved content, ensuring consistent typography, motion cues, color, and tone across generated videos, moving beyond template restrictions.
Idomoo argues templates compromise visuals, whereas Strata designs custom blueprints for each video. Its layered output allows real-time personalization with individual data fields.
Idomoo’s platform serves companies like JPMorganChase, Verizon, and American Airlines, producing personalized videos for communications and marketing. Strata enhances personalization at the composition level.
Currently, Strata is in early access testing with large customers via the Lucas AI Video Agent. However, Idomoo hasn’t disclosed customer details, benchmarks, or comparisons with other models. The “first foundation model for layered video” claim hasn’t been independently assessed, and Strata’s technology is patent pending.
Idomoo’s documentation previously indicated reliance on existing AI models, but Strata signals a shift towards developing foundational AI for video. Its effectiveness will be clearer as enterprises transition from early access to full implementation.
Established in 2007 by Yaron Kalish, Danny Kalish, and Assaf Fogel, Idomoo has raised $27 million, including $9 million in Series A and $18 million in Series B funding. The company has spent nearly two decades refining its personalized video platform for enterprise clients, providing ample structured data for training.