Marc

Computer Vision, Generative AI & Software Engineering.

[Marc @superr_marc]

More magic wands for an RPG. Prompt on the ALT.
#Midjourney #AIart #GenerativeAI #Games #Assets #wand #UnrealEngine #Unity

[Amine 😞🌧 @AmineMaTue]

DEUX CLIOS 6 NEUVES À GAGNER 😎

La 1ère voiture : RT ce tweet et je tire au sort une personne qui gagne la Clio parmi les retweets.

La 2ème voiture : ce soir 20h je live quelque part dehors en Île De France. Le premier qui me trouve je cut le live et il gagne la voiture.

[Marc @superr_marc]

@MyKpopIdolAI https://t.co/gGe6exOySb

[Marc @superr_marc]

@dreamina_ai Let me in 🥳🥳🥳

[Dreamina AI @dreamina_ai]

less than 5 days left. €40,000 prize pool + a trip to Annecy — the "Oscars of Animation" — is waiting for you. Submit your AI animated work now.

-->RT + Comment within 24h to get 1000 extra credits via DM! First 200 users ONLY

[Anthropic @AnthropicAI]

New research from Anthropic Fellows Program: Selective GradienT Masking (SGTM).

We study how to train models so that high-risk knowledge (e.g. about dangerous weapons) is isolated in a small, separate set of parameters that can be removed without broadly affecting the model.

[Claire Silver 🌸 @ClaireSilver]

gm

So if you ask Nano Banana to make an image into a Depth Map, you get a grayscale image where white=close & black=far

Then ask Google AI Studio for an app that turns depth maps into point clouds & lets you control them with hand gestures from webcam

blam you’re Tony Stark

[Marc @superr_marc]

But Kiro isn't just fast, it's smart.

I spent 8 hours writing specs (requirements, design, tasks). Kiro referenced them throughout development, always knowing what to build next and why.

Result: 40+ hours saved in rework.

Planning became my superpower.

[Marc @superr_marc]

I just built a production-ready AI app in 90 hours with @kirodotdev

DreamWeen transforms dreams into cinematic videos using Gemini, Veo, and OpenAI.

Here's how Kiro changed the way I approach development 🧵

#hookedonkiro #Kiroween

[Marco @ai_artworkgen]

OK, Sora. This is cool! 🔥

I gave Sora some timeline-specific instructions and it nailed them every time.

→ Examples + prompts ... 👇

[OpenAI Developers @OpenAIDevs]

Introducing AgentKit—build, deploy, and optimize agentic workflows.

💬 ChatKit: Embeddable, customizable chat UI
👷 Agent Builder: WYSIWYG workflow creator
🛤️ Guardrails: Safety screening for inputs/outputs
⚖️ Evals: Datasets, trace grading, auto-prompt optimization

[Google DeepMind @GoogleDeepMind]

Software vulnerabilities can be notoriously time-consuming for developers to find and fix.

Today, we’re sharing details about CodeMender: our new AI agent that uses Gemini Deep Think to automatically patch critical software vulnerabilities. 🧵

[DeepSeek @deepseek_ai]

🚀 Introducing DeepSeek-V3.2-Exp — our latest experimental model!

✨ Built on V3.1-Terminus, it debuts DeepSeek Sparse Attention(DSA) for faster, more efficient training & inference on long context.
👉 Now live on App, Web, and API.
💰 API prices cut by 50%+!

1/n

[Luma @LumaLabsAI]

This is Ray3. The world’s first reasoning video model, and the first to generate studio-grade HDR. Now with an all-new Draft Mode for rapid iteration in creative workflows, and state of the art physics and consistency. Available now for free in Dream Machine.

[Chris @ChrisGPT]

🚨 OpenAI Claims to have found the cause of hallucinations

Maybe this is why, all of those employees posted about AGI just before this paper came out

[Charly Wargnier ♨️ @DataChaz]

This is wild.

A real-time webcam demo using SmolVLM from @huggingface and llama.cpp! 🤯

Running fully local on a MacBook M3.

[Tencent Hy @TencentHunyuan]

Today we're announcing the open-source release of HunyuanVideo-Foley, our new end-to-end Text-Video-to-Audio (TV2A) framework for generating high-fidelity audio.🚀

This tool empowers creators in video production, filmmaking, and game development to generate professional-grade audio that precisely aligns with visual dynamics and semantic context, addressing key challenges in V2A generation.🔊

Key Innovations:

🔹Exceptional Generalization: Trained on a massive 100k-hour multimodal dataset, the model generates contextually-aware soundscapes for a wide range of scenes, from natural landscapes to animated shorts.

🔹Balanced Multimodal Response: Our innovative multimodal diffusion transformer (MMDiT) architecture ensures the model balances video and text cues, generating rich, layered sound effects that capture every detail—from the main subject to subtle background elements.

🔹High-Fidelity Audio: Using a Representation Alignment (REPA) loss function and a powerful Audio VAE, we've improved generation stability and producing professional-grade audio, free of noise and inconsistencies.

HunyuanVideo-Foley achieves SOTA on multiple benchmarks, surpassing all open-source models in audio quality, visual-semantic alignment, and temporal alignment.

👉Try it now: https://t.co/jdxOMEu3Uf
🌐Project Page: https://t.co/hh3G5NGZbT
🔗Code: https://t.co/fpaYnXUVJ3
📄Technical Report: https://t.co/R3GqVaOVmy
🤗Hugging Face: https://t.co/kF0dcPh2Ns

[Xenova @xenovacom]

NEW: Apple releases FastVLM and MobileCLIP2 on Hugging Face! 🤗

The models are up to 85x faster and 3.4x smaller than previous work, enabling real-time VLM applications! 🤯

It can even do live video captioning 100% locally in your browser (zero install). Huge for accessibility!

[EyeingAI @EyeingAI]

WTF... Nano-Banana is disrupting everything.

You can now generate visuals directly on canvas, auto-place them in your site, keep styles consistent & publish in one click.

Goodbye Figma, Midjourney & Photoshop.

Tutorial + some crazy examples: 🧵

[The Daily Ai @The_DailyAi]

Google just build the craziest AI photo editor ever

Nano Banana in Gemini is wild

People are already dropping insane use cases

10 wild examples :

1. Model pose like the sketch