Très beau prompt pour chatgpt : Prompt :
Une boule de cristal [TYPE] posée sur une corniche en pierre, reflétant une image inversée de [INSÉRER UN MONUMENT OU UNE SCÈNE], tandis que le véritable [INSÉRER MONUMENT OU SCÈNE] apparaît en arrière-plan avec un léger flou.
La boule de
MULTIMODAL AI
-
Crystal Ball Prompt for ChatGPT Image Generation Scenes
By
–
-
Gemma 3n E4B: Lightweight Multimodal AI Model
By
–
And ofc, there is a recently added Gemma 3n E4B. A SOTA, lightweight, open-source multimodal AI model, designed for efficient execution on low-resource devices.
-
Veo 3 Struggles with Split-Screen Foley Artist Prompt
By
–
This was a prompt idea from @steveruizok, but I don't think it's one that Veo 3 can get right, this was the closest:
— fofr (@fofrAI) 25 mai 2025
> a video is split 50/50, on the left is a scene, on the right is a foley artist producing the sound effects for that scene pic.twitter.com/Sx4kTFbePwThis was a prompt idea from @steveruizok
, but I don't think it's one that Veo 3 can get right, this was the closest: > a video is split 50/50, on the left is a scene, on the right is a foley artist producing the sound effects for that scene -
AI Model Fails to Generate Accurate Europe Map
By
–
How is this possible? Based on everything i understand of from how those models work a map of Europe should be flawless. It calls England Spalnd and Ireland Relad At least it got Shittyjania correct.
-
100% AI-Generated Car Show Video Fools Viewers as Real
By
–
On continue :
— VISION IA (@vision_ia) 25 mai 2025
Je suis tombé sur cette vidéo et j’ai cru à un vrai salon automobile.
Ambiance, dialogues, mouvements de foule… tout sonnait juste.Mais tout est faux. 100 % généré par IA. Et le pire ?
Si vous la passiez au JT, personne ne sourcillerait.On vient de franchir un… https://t.co/d2MFmhEWuD pic.twitter.com/dkatMmOW8tOn continue : Je suis tombé sur cette vidéo et j’ai cru à un vrai salon automobile. Ambiance, dialogues, mouvements de foule… tout sonnait juste.Mais tout est faux. 100 % généré par IA. Et le pire ? Si vous la passiez au JT, personne ne sourcillerait.On vient de franchir un
-
GRIT: Teaching MLLMs Grounded Visual Reasoning with Images
By
–
10. Teaching MLLMs to Think with Images GRIT is a new method that enables MLLMs to perform grounded visual reasoning by interleaving natural language with bounding box references.
-
Top AI Research Papers of the Week: May 19-25
By
–
Here are the top AI Papers of the Week (May 19 – 25): – ARC-AGI-2
– AdaptThink
– EfficientLLM
– Visual Planning
– The Pitfalls of Reasoning
– Teaching MLLMs to Think with Images Here is the full list: -
Visual Planning: Image-Based Reasoning Replaces Language Planning
By
–
1. Visual Planning Proposes a novel reasoning paradigm that replaces language-based planning with image-based reasoning.
-
Visual Turing Test Repeatedly Surpassed by AI Systems
By
–
The visual Turing test is getting shattered again and again
-
Veo 3 generates Burning Man interviews from latent space
By
–
Conjuring burning man interviews out of latent space. Veo 3 is too good! pic.twitter.com/jDHs64nshw
— Bilawal Sidhu (@bilawalsidhu) 25 mai 2025Conjuring burning man interviews out of latent space. Veo 3 is too good!