Marta Mrak retweeted
Gen-3 Alpha Video to Video is now available on web for all paid plans. Video to Video represents a new control mechanism for precise movement, expressiveness and intent within generations. To use Video to Video, simply upload your input video, prompt in any aesthetic direction you like, or, choose from a collection of preset styles.
See examples below.
(1/6)
Marta Mrak retweeted
Introducing Sora, our text-to-video model.
Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions.
openai.com/sora
Prompt: “Beautiful, snowy Tokyo city is bustling. The camera moves through the bustling city street, following several people enjoying the beautiful snowy weather and shopping at nearby stalls. Gorgeous sakura petals are flying through the wind along with snowflakes.”
PeQuENet - Very effective way to reduce compression artefacts and make video quality better using our attention modules in a deep network, as previously featured in a colourisation approach bbc.co.uk/rd/blog/2021-05-vi…
PeQuENet: Perceptual Quality Enhancement of Compressed Video with Adaptation- and Attention-based Network
deepai.org/publication/peque…
by Saiping Zhang et al. including @MartaMrak
#GenerativeAdversarialNetwork #ComputerVision
5
VIPs of @IEEEsps welcomed us today to @ieeeICASSP with energizing performance of local dragons 🎇🇸🇬
And @ieeeICASSP is finally back in person, after a couple of years of virtual experiences. Great to see my @ieeeMMSP and @IEEEsps colleagues
10
Marta Mrak retweeted
Continual developments in compression are needed to deliver bandwidth-intensive video formats such as ultra-high definition or panoramic video. We are exploring new approaches which use artificial intelligence techniques.
bbc.co.uk/rd/blog/2021-11-dv…
This year @ieeeMMSP workshop was held as a hybrid event but I had to attend it virtually, still missing out on many benefits and fun of being there in person. Many thanks to the organisers for sending a nice conf. package by post, including USB proceedings @IEEEsps @CIVIT_Finland
How to quickly create visuals for a story when lots of footage is available? Our colleagues from @FCTNOVA - Joao Magalhaes @jmc_mag and @davidfsemedo led this work and came up with a number of useful strategies to automate the process 💡💡 A summary was presented at @ACMMM_2021
The extra perspectives from user generated content can be a great addition to a news story, but it can be hard to curate an overwhelming amount of material that comes in from a live event.
We are developing tools to help editors and automate the process:
bbc.co.uk/rd/blog/2021-11-au…
3
📢 IEEE ICME 2022 deadlines are approaching fast
- special sessions, panels, grand challenges, workshops and tutorials: 22 Nov '21
- regular papers: 12 Dec '21
See more at 2022.ieeeicme.org/
Organised by @IEEEsps, @ieeecassociety, @ComputerSociety and @ComSoc
Great experience of hybrid @ieeeMMSP 2021 thanks to the amazing work lead by Atanas Gotchev @TampereUni and his team 👏👏
The photo shows how my colleague @marcgorriz presented his work today. The recording of the whole session is available on the virtual platform.
📢 Applied AI & ML for Video Coding and Streaming, the first special issue in IEEE OJSP @IEEEsps, is now published thanks to my fellow guest editors Mahmoud Reza Hashemi, Shervin Shirmohammadi, Ying Chen and Moncef Gabbouj @MoncefGabbouj
ieeexplore.ieee.org/document…
Paper 7/8
“Efficient bitrate ladder construction for content-optimized adaptive video streaming,” dx.doi.org/10.1109/OJSP.2021….
by Angeliki V. Katsenou @angiekatsenou; Joel Sole; David R. Bull @davrull; @BristolUni and @netflix
1
1
Paper 8/8
“Forecasting video QoE with deep learning from multivariate time-series,” dx.doi.org/10.1109/OJSP.2021…
by Hossein Ebrahimi Dinaki; Shervin Shirmohammadi; Emil Janulewicz; David Côté, @uOttawa and @CienaBluePlanet
1













