Congrats! Big props unifying vision language datasets and making it easy to access them. We also recently released a massive VQA dataset of 30M samples(22M filtered), spanning 42 countries and 39 languages. The dataset contains factual visual questions and answers about world
MULTIMODAL AI
-
DeepSeek OCR Paper and Multimodal LLM Article Resources
By
–
Link to the paper: https://
github.com/deepseek-ai/De
epSeek-OCR/blob/main/DeepSeek_OCR_paper.pdf
… My "Understanding Multimodal LLMs" article with more info on how images are fed to LLMs, how cross-attention works, etc: https://
magazine.sebastianraschka.com/p/understandin
g-multimodal-llms?utm_source=publication-search
… -
DeepSeek-OCR Release: Vision Model Capabilities Explained
By
–
DeepSeek finally released a new model and paper. And because this DeepSeek-OCR release is a bit different from what everyone expected, and DeepSeek releases are generally a big deal, I wanted to do a brief explainer of what it is all about. In short, they explore how vision
-
Visual Tokenization Challenges: Aspect Ratios and Image Preprocessing Complexity
By
–
I know it’s popular to hate tokenizers, but visual representations (which are also tokenized) bring a lot of messiness as well. Aspect ratios, cropping, resolution, brightness, etc. Sure, models learn to deal with that but it requires lots of data to make them robust wrt these.
-
World simulation models like Genie 3 move beyond pixel prediction
By
–
the solution? it’s not better video models. it’s world simulation models like google’s genie 3. genie doesn’t just predict pixels – it simulates physics rules. gravity, collision, momentum. it understands that when you climb a ladder, your hand grips, your foot lifts, your
-
Google VEO 3.1 Introduces AI-Powered Video Editing Feature
By
–
La nouvelle fonction de Google VEO 3.1 : l'édition vidéo par IA. Il est possible de rajouter facilement des objets pic.twitter.com/XA74WQkvf3
— VISION IA (@vision_ia) 21 octobre 2025La nouvelle fonction de Google VEO 3.1 : l'édition vidéo par IA. Il est possible de rajouter facilement des objets
-
Veo 3.1 Shows Major Quality Improvements in Video Generation
By
–
Great to see the major jump on quality between our Veo 3.0 and Veo 3.1 models. High quality video generation will unlock all kinds of creative uses!
-
Veo 3.1 tops video generation leaderboards with major improvements
By
–
Awesome to see Veo 3.1 top the LMArena video leaderboards by a large distance with big improvements over Veo 3.0 for text-to-video (+30) and image-to-video (+70)! Huge congrats to the team! Try it for yourself in http://
flow.google and the @GeminiApp -
Optimizing First and Last Frame AI Video Generation Techniques
By
–
Here are some really tactical ways to optimize the First and last frame capability:
— Google AI (@GoogleAI) 20 octobre 2025
— Make sure your prompt includes precise camera motion descriptions for smooth and creative transitions between the first and last frames.
— Or, you can simply include the word “transform” in… pic.twitter.com/pGGtgZMjWzHere are some really tactical ways to optimize the First and last frame capability: — Make sure your prompt includes precise camera motion descriptions for smooth and creative transitions between the first and last frames.
— Or, you can simply include the word “transform” in -
Scene Extension in Generative Video AI: Chaining Frames
By
–
Scene extension works by chaining: the last frame of the first video acts as the visual starting point for the next. This means that if your first video ends in music or background noise, the second video will maintain that audio consistency throughout.
— Google AI (@GoogleAI) 20 octobre 2025
Also, you can extend… pic.twitter.com/q7n2FXNfB0Scene extension works by chaining: the last frame of the first video acts as the visual starting point for the next. This means that if your first video ends in music or background noise, the second video will maintain that audio consistency throughout. Also, you can extend