Meitu Inc. has announced that its flagship model, MiracleVision, has achieved another upgrade in image generation capabilities, marking further enhancements in the model's overall strength. Concurrently, Meitu has introduced MOKI, a one-stop AI short film creation tool, which is expected to gradually integrate into Meitu's product ecosystem, offering users a new visual experience.
This upgrade marks the second update to MiracleVision within a month. On September 23, the model's video generation capabilities were enhanced, supporting the creation of ultra-long videos up to 1 minute in length, with a frame rate of 24 FPS and a resolution of 1080P. This latest upgrade particularly focuses on strengthening image generation, leveraging MiracleVision's semantic understanding and text expansion capabilities to achieve precise visual expressions and atmospheric renderings with a sense of story.
The MiracleVision team stated that this upgrade deeply integrates Meitu's long-term accumulations in the fields of imaging, aesthetics, and art, incorporating diverse aesthetic concepts into the model training process. This results in images that are more delicate and natural, capable of presenting full-spectrum cinematic visuals from contours, lighting, to composition. The upgraded MiracleVision supports content input from ultra-short to ultra-long texts, accurately understands diverse styles and complex scenes, and showcases imaginative and creative visuals.
It is worth noting that this upgrade specifically extracts Chinese cultural elements for targeted optimization, achieving precise understanding and content expression of Eastern semantics and even the fusion of Eastern and Western semantics, offering users a visually rich style with an Eastern flavor.