首页|Patent Application Titled 'Generative Model For Multi-Modality Outputs From A Si ngle Input' Published Online (USPTO 20240233318)
Patent Application Titled 'Generative Model For Multi-Modality Outputs From A Si ngle Input' Published Online (USPTO 20240233318)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
Reporters obtained the following quote from the background information supplied by the inventors: “Acommon task in digital image generation is generating multi ple images that depict visually similar scenesor environments, expressed in dif ferent modalities. For instance, some approaches for digitally rendering athree -dimensional digital environment rely on multiple digital images that each expre ss the environmentin a different modality to do so, such as one image that depi cts visual characteristics (e.g., colors) ofthe environment as perceived by a h uman viewer, another image that represents depth characteristics ofdifferent po ints in the environment, another image that identifies segmentation boundaries f or differentobjects in the environment, and so forth. Conventional image genera tion systems, however, are unable togenerate these multiple modality images for the environment using a single generative model.”
NETL
NSTL
