CVMP

The ACM SIGGRAPH European Conference on Visual Media Production brings together postproduction and research communities.

[DECaDE @DECaDE_Hub]

DECaDE's @kar_balan presenting @CVMP. PDFed - a blockchain-based federated learning system for training diffusion models, addresses a key privacy challenge - it enables us to measure and minimize the memorization of training data in generative AI models. https://t.co/CsDOd1WKO4

[CVMP @CVMP]

Aaaaand... That's a wrap. Thank you so much to everyone who made #CVMP2024 possible: chairs, committee, volunteers, keynotes, speakers, attendees, sponsors, & all the staff at #BFIsouthbank
This edition of #CVMP was a success & we hope to see you all next year. Stay tuned! 👋🏼👋🏼👋🏼

[CVMP @CVMP]

Congratulations to the best paper and best short paper awards!! It was so close that we also had two runner-ups. Congratulations to everyone for their great work 👏🏼👏🏼👏🏼 #CVMP2024

[CVMP @CVMP]

He showed us the multiple media and art projects they've been working on. From XR experiences in museums and tourism, to an AI Jesus

[CVMP @CVMP]

For our last keynote and talk of the conference we count with Dr Aljosa Smolic @DrSmolic from Lucerne University of Applied Sciences and Arts, talking about AI-based volumetric content creation for immersive XR experiences and production workflows #CVMP2024

[CVMP @CVMP]

From chapter generation and summarisation, to AI tutors to help revising and learning engagement, deep fakes for anonymous contributions/testimonials, and tools for content production, editing, and much more!

[CVMP @CVMP]

To end the gen AI session we have Graham Thomas, head of applied research and production at @BBCRD talking about the work and impact of Generative AI in Public Service Media at the BBC. How BBC AI and editorial principles apply, and what cool projects they work on #CVMP2024

[CVMP @CVMP]

And you can participate in the development and request a developer kit!

[CVMP @CVMP]

Third speaker of the industrial session is Armen Avetisyan from @RealityLabs giving us an insight into their always-on sensing AI wearables projects with their Aria Glasses #CVMP2024

[CVMP @CVMP]

The use of AI brings a set of technical challenges, but also legal and ethical issues. Artists and creatives are scared of using AI that can potentially take their jobs, but also worried to not use AI that helps with tools and efficiency. Where is the balance?

[CVMP @CVMP]

Next up at the Generative AI in Media Production session we have Graham Jack CTO of BeloFX talking about the use of AI in VFX, what works and what do EEesn't, from a creative's perspective #CVMP2024

[CVMP @CVMP]

Using a combination of three techniques, metadata, fingerprints, and watermarks, with Blockchain for decentralised, robust, and durable provenance identifiers: cr - content credential. More info at the C2PA coalition https://t.co/VuIbFkpyit

[CVMP @CVMP]

To open the afternoon industrial session on Generative AI in Media Production we have @JCollomosse from @Adobe @DECaDE_Hub to talk about how to build safe and fair generative AI using content provenance techniques #CVMP2024

[CVMP @CVMP]

And demos!

[CVMP @CVMP]

Check out today's #CVMP2024 short papers, posters and demos at lunch time and coffee breaks in The Blue Room ⚡

[CVMP @CVMP]

More posters

[CVMP @CVMP]

Our first keynote of today is Shunsuke Saito @psyth91 from @RealityLabs showing us their research on foundational models for digital humans and the importance of human avatars to be relightable, animatable, and generalisable #CVMP2024

[CVMP @CVMP]

To end the human and motion session, we have Kalila Shapiro from @ucl to talk to us about the relationship between usability and presence in object detection at a distance in VR systems #CVMP2024

[CVMP @CVMP]

Next we have @kar_balan from @DECaDE_Hub @cvssp_research talking about using blockchain-based decentralised asynchronous federated learning for training privacy-preserving diffusion models with data provenance and image verification for a fairer more inclusive AI #CVMP2024

[CVMP @CVMP]

Our first session this morning is on Human Motion, Privacy and Usability 🤸🏼‍♀️ first we have David E. Moreno-Villamarin from @FraunhoferHHI talking to us about using generative models to represent human motion from limited data