View source for 18296938. GENERATING VIDEOS USING DIFFUSION MODELS simplified abstract (Google LLC)

=GENERATING VIDEOS USING DIFFUSION MODELS=


==Organization Name==

[[:Category:Google LLC|Google LLC]]


[[Category:Google LLC]]

==Inventor(s)==

[[:Category:Jonathan Ho of New York NY (US)|Jonathan Ho of New York NY (US)]][[Category:Jonathan Ho of New York NY (US)]]

[[:Category:Tim Salimans of Utrecht (NL)|Tim Salimans of Utrecht (NL)]][[Category:Tim Salimans of Utrecht (NL)]]

[[:Category:Alexey Alexeevich Gritsenko of Amsterdam (NL)|Alexey Alexeevich Gritsenko of Amsterdam (NL)]][[Category:Alexey Alexeevich Gritsenko of Amsterdam (NL)]]

[[:Category:William Chan of Toronto (CA)|William Chan of Toronto (CA)]][[Category:William Chan of Toronto (CA)]]

[[:Category:Mohammad Norouzi of Richmond Hill (CA)|Mohammad Norouzi of Richmond Hill (CA)]][[Category:Mohammad Norouzi of Richmond Hill (CA)]]

[[:Category:David James Fleet of Toronto (CA)|David James Fleet of Toronto (CA)]][[Category:David James Fleet of Toronto (CA)]]

==GENERATING VIDEOS USING DIFFUSION MODELS - A simplified explanation of the abstract==


This abstract first appeared for US patent application 18296938 titled 'GENERATING VIDEOS USING DIFFUSION MODELS

'''Simplified Explanation:'''
The patent application describes methods, systems, and apparatus for generating an output video based on an input by updating an intermediate representation using a diffusion model.

* The method involves receiving an input and initializing a current intermediate representation.
* An output video is generated by updating the intermediate representation at each iteration using a diffusion model to process the input and generate a noise output.
* The current intermediate representation is updated with the noise output at each iteration.

'''Key Features and Innovation:'''
* Utilizes a diffusion model to process the input and generate a noise output for updating the intermediate representation.
* Updates the intermediate representation at each iteration to generate an output video conditioned on the input.

'''Potential Applications:'''
* Video processing and editing software.
* Computer-generated imagery (CGI) in movies and animations.
* Virtual reality (VR) and augmented reality (AR) applications.

'''Problems Solved:'''
* Efficient generation of output videos based on inputs.
* Enhancing the quality and realism of generated videos.

'''Benefits:'''
* Improved video generation process.
* Enhanced visual effects in videos.
* Increased efficiency in video editing.

'''Commercial Applications:'''
The technology can be used in video editing software, CGI production for films and animations, and VR/AR applications, potentially revolutionizing the way videos are created and edited in various industries.

'''Questions about the Technology:'''
1. How does the diffusion model improve the generation of output videos?
2. What are the potential limitations of using this technology in video production?


==Original Abstract Submitted==

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating an output video conditioned on an input. In one aspect, a method comprises receiving the input; initializing a current intermediate representation; generating an output video by updating the current intermediate representation at each of a plurality of iterations, wherein the updating comprises, at each iteration: processing an intermediate input for the iteration comprising the current intermediate representation using a diffusion model that is configured to process the intermediate input to generate a noise output; and updating the current intermediate representation using the noise output for the iteration.



[[Category:G06V10/82]]
[[Category:G06V10/771]]
[[Category:H04N7/01]]
[[Category:CPC_G06V10/82]]