Today in Generative Media

Deepfakes threaten elections; Synthesia's avatars emote; Adobe's AI video upscaling

• AI deepfakes threaten to upend global elections. No one can stop them. (Washington Post)

• Nvidia-backed startup Synthesia unveils AI avatars that can convey human emotions (CNBC)

• Adobe’s impressive AI upscaling project makes blurry videos look HD (The Verge)

• The largest AI film conference. One day of learning, sharing, & collaborating. (AI on the Lot)

• Will Generative AI Suppliers Last Long Enough for Games to Ship? (AI and Games)

• The “it” in AI models is the dataset. (Non_interactive)

• Rethinking Inductive Biases for Surface Normal Estimation (CVPR 2024, project page) Code on GitHub. Real-time demo on X.

• FlowMap: High-Quality Camera Poses, Intrinsics, and Depth via Gradient Descent (project page)

• Align Your Steps: Optimizing Sampling Schedules in Diffusion Models (project page)

• FlashSpeech: Efficient Zero-Shot Speech Synthesis (arXiv)

• 💡Training a Motion LoRA from CG Render (LinkedIn)

• Generate a video that morphs between subjects, with an optional style (Replicate)

• 💡 HiDiffusion: Unlocking Higher-Resolution Creativity and Efficiency in Pretrained Diffusion Models (GitHub)

• You can re-build the set design for a Gucci Bloom ad on Flair AI in 6 steps. (X)

• Introducing Morph Studio, an all-in one AI video workflow. Ignite your creativity with our story playground 🧩 Releasing access to waitlisted users starting today. (X)

• I've updated my sticker-maker model on Replicate. It now uses LayerDiffuse rather than BRIA-RMBG – the stickers have better transparency and the model can now be used commercially! (X)

• No real details on the workflow so hard to say if this is realtime (or real for that matter), but in any case, I wouldn't be surprised if this is possible in real time in a year or less given the trajectory of tech we see now. (X)

• I'm here today to introduce the latest in wearable computing technology. Introducing the AI SmartNose (X)