Samsung's latest deepfake technique requires just a single source image [www.techspot.com]
Samsung AI makes painting move [www.pocket-lint.com]
Mona Lisa frown: Machine learning brings old paintings and photos to life [techcrunch.com]
It's Getting Way Too Easy to Create Fake Videos of People's Faces [www.vice.com]
Deepfakes are getting easier than ever to make, new research paper shows [www.theverge.com]
'Deepfake' AI can turn the Mona Lisa into a convincing real person [www.telegraph.co.uk]
Another great paper from Samsung AI lab! @egorzakharovdl et al. animate heads using only few shots of target person (or even 1 shot). Keypoints, adaptive instance norms and GANs, no 3D face modelling at all.
— Dmitry Ulyanov (@DmitryUlyanovML) May 22, 2019
▶️ https://t.co/Xk5D4WccpD
? https://t.co/SxnVfY72TT pic.twitter.com/GjVrJbejT0
Seeing is believing.
— Jeremiah Grossman (@jeremiahg) May 22, 2019
Video or it didn’t happen.
In the very near future, we won’t say or believe those things anymore. https://t.co/Ux8wz0VPr7
Impressive and disturbing from Samsung's AI lab. Take a still image or two, record someone talking and voila you've got a talking head with no 3D modelling involved. Go to 5:08 to see the Mona Lisa having a bit of a chat https://t.co/EH2X0GgQYW
— Chris Moran (@chrismoranuk) May 23, 2019
Wow. This is fascinating research. How has no one turned this into a selfie / impersonation app yet? — Few-Shot Adversarial Learning of Realistic Neural Talking Head Models https://t.co/1T0TiuwGkh
— Ryan Sarver (@rsarver) May 23, 2019
Videos of people from one photo ?
— Andreas Klinger ✌️ (@andreasklinger) May 23, 2019
Also works with photos from the past and even paintings.
This is crazy ?https://t.co/4WaIF3L1Fp
This is not a good thing. Deep fakes are possible with a frighteningly limited data set. Yet again technology has far outstripped our legislative ability and regulatory processes. Liberal democracy will suffer. Watch this space. https://t.co/Q0Pt7UGrCO
— Vonny LeClerc (@vonny_bravo) May 23, 2019
Prediction: The 2024 election will be consumed with a scandalous video of the Democratic candidate saying something shocking, and it will only be discovered it was a deep fake after there was blanket coverage for 72 hours. https://t.co/Ae7cmZmpsp
— Amanda Marcotte (@AmandaMarcotte) May 23, 2019
Its not ‘perfectly realistic’ it’s definitely though uncanny valley. I won’t hide I am excited what could be done with this by artists but equally (if not more) terrified by what will certainly be done with this by certain nefarious parties. Issa new world. https://t.co/TVA3pezfQX
— monika bielskyte (@monikabielskyte) May 23, 2019
Terrifying to think of all the bad ways this will be used. https://t.co/bjGR5lnn3N
— Cory Barlog ? #Raising Kratos (@corybarlog) May 23, 2019
Mona Lisa frown: Machine learning brings old paintings and photos to life https://t.co/bmBwT4MXnP pic.twitter.com/ub0Y5QPzQF
— TechCrunch (@TechCrunch) May 22, 2019
Samsung researchers have released a model that can generate faces in new poses from just a single image/frame (for each of face, pose). Done by building a well-trained landmark model in advance & one-shotting from that.
— Mark McD ☠ (@m4rkmc) May 23, 2019
This gif is my favourite...
https://t.co/lJrRQAGoSM pic.twitter.com/YzA9HBgUfN
AI algorithms have gotten eerily good at making fake videos based on large photo or video datasets. But a new preprint reports few-shot capability that needs only a few – or sometimes just one – photo to make realistic talking head animated portraits. https://t.co/k3FralAHNp pic.twitter.com/VlIhlxSefI
— Aaron Dy ??? (@AaronJDy) May 23, 2019
this is really, really something: researchers developed a way to make "perfectly realistic" fake videos of faces, using only one source image https://t.co/qDYSa5EQxb pic.twitter.com/bVc2eDSlhh
— Samantha Cole (@samleecole) May 23, 2019
Samsung researchers developed an algorithm that only needs one source image to create fake videos.https://t.co/uu6imcjUa9 pic.twitter.com/sn19CnSmv1
— VICE (@VICE) May 23, 2019
Samsung researchers developed an algorithm that only needs one source image to create fake videos. @vice https://t.co/u00oNUgDlS pic.twitter.com/mueEiofcZ9
— Boback Ziaeian ??♂️ (@boback) May 23, 2019
It's getting way too easy to create fake videos of people's faces: https://t.co/OR5mtZrzrM pic.twitter.com/kQVFUyDpSn
— Motherboard (@motherboard) May 23, 2019
It's getting way too easy to create fake videos of people's faces:https://t.co/uu6imcjUa9 pic.twitter.com/FbETeEBPv8
— VICE (@VICE) May 23, 2019
Not worried at all. https://t.co/bX3yYgFyZn
— Arielle Duhaime-Ross (@adrs) May 23, 2019
Deepfakes are getting easier than ever to make, new research paper shows https://t.co/f53fnDoL8M pic.twitter.com/59nH9BvFuT
— The Verge (@verge) May 23, 2019
Another great paper from Samsung AI lab! @egorzakharovdl et al. animate heads using only few shots of target person (or even 1 shot). Keypoints, adaptive instance norms and GANs, no 3D face modelling at all.
— Dmitry Ulyanov (@DmitryUlyanovML) May 22, 2019
▶️ https://t.co/Xk5D4WccpD
? https://t.co/SxnVfY72TT pic.twitter.com/GjVrJbejT0
This #AI brings portraits to life.?
— René Schulte (@rschu) May 23, 2019
Amazing new paper introducing a GAN-based #DeepLearning model that synthesizes videos from just a single photo! ?
Can you imagine a talking museum gallery. You hold up your phone and the painting talks to you. ?https://t.co/69mrU3lhVW pic.twitter.com/ssnhdrxGSK
We are sooooo not prepared for this. Ironically, @AriMelber raised this point while interviewing Adam Schiff right before Pelosi took the stage at this event.
— Alex Seitz-Wald (@aseitzwald) May 23, 2019
And ystdy, Samsung announced it can turn a photo into a realistic video of a person speaking. https://t.co/xB8oJVMV3l https://t.co/w074UVJ7MT
The scenarios in which this technology could be misused are frightening. https://t.co/Jk9Ei5SCZc #deepfakes pic.twitter.com/HeoVbgoKPe
— nwsltr (@nwsltrMe) May 24, 2019
Amazing work by the @Samsung AI Center animating from works of art and single frame faces! It's amazing (and a little scary) where synthetic video is going! https://t.co/E5bXt2TGV0
— Atul Butte (@atulbutte) May 23, 2019
Motherboard It's Getting Way Too Easy to Create Fake Videos of People's Faces: Samsung researchers developed an algorithm that only needs one source image to create fake videos. https://t.co/OKlwNGa5Ib #AI #programming #algorithms #machinelearning Via @motherboard pic.twitter.com/DT4b4AiemR
— Bradley Jon Eaglefeather (@bjeaglefeather) May 24, 2019
single-image deepfakes is a very crazy thing https://t.co/F4PaZ1TIt2
— Jason Koebler (@jason_koebler) May 23, 2019
This, in a way, is a leap beyond what even deepfakes and other algorithms using generative adversarial networks can accomplish. https://t.co/uu6imcjUa9
— VICE (@VICE) May 24, 2019
Samsung Deepfake AI can fabricate a video clip of you from a single photo, as AI animates photos without 3D modeling.
— DataScienceNigeria (@DataScienceNIG) May 24, 2019
Cautious! AI can make you appear to do or say things you never did.
Watch: https://t.co/6auRg1GSl2
Paper: https://t.co/5JtK3mpwWr…
Kudos @egorzakharovdl etal pic.twitter.com/UzyZG5Qtja
Animating old photos & paintings with #AI https://t.co/JPruApZFK4 pic.twitter.com/7FLsoxqWx9
— David McCandless (@mccandelish) May 23, 2019
Samsung recently revealed a new "deepfake AI" that can take photos and generate an animation of a moving character (https://t.co/Eu2WDtjipv). 4/7
— Kalle Alm (@kallewoof) May 24, 2019
As predicted, those interests who know there will be VERY incriminating evidence of them offered up publicly, are desperate to continue ginning-up the emerging deepfake technology in upcoming attempts to claim "it's not me".
— OutlawJoseyWales⚒️ (@outlawjw) May 24, 2019
It's gonna happen. https://t.co/YKR8NTqrZ7
'#Deepfake' AI can turn the Mona Lisa into a convincing real person https://t.co/Zr1uzBAJNj
— Jon Carroll (@jondcarroll) May 23, 2019