Yusuf

[Yusuf @yusuffractal]

@seckincsahin Merhaba, SNP markerları üzerine kurulu bir adli DNA analiz projem vardı; buna ek olarak k-NN ve Random Forest modelleriyle coğrafi köken tahmini de entegre etmiştim ama takım arkadaşımdan dolayı yarım kalmıştı. Sizin projeniz oldukça ilgi çekici👏💯
https://t.co/XgQ00faSaw

[The Bugged Dev @thebuggeddev]

Only if education could be this interactive ❤️‍🔥

I've had a looong wish to build something genuinely useful through vibe coding, and I finally did it.

A 3D human anatomy application built with @threejs using GPT 5.6 Sol.

It all started with a single design image that I created using GPT Image 2.0. I then used it to generate every 3D organ image, one by one.

Next, I converted each of those images into 3D models using @tripoai (and no, they didn't sponsor this 😄).

After that, I opened Codex, wrote a master prompt based on the design, and gave it the prompt, the design image, and all the 3D models.

Codex built the first version beautifully, but there was one big problem.

Every single 3D model was nearly 120-150 MB. That obviously wasn't practical for the web and was giving a performance of 16fps.

After a few iterations, Codex optimized each model down to roughly 2–5.5 MB while preserving the visual quality, reducing the total asset size from ~900 MB to just 28.6 MB. And each model loads on demand.

Along the way, Codex also generated those anatomical illustrations showing where each organ sits in the human body, and even created the interactive hotspot markers that explain different parts of every organ. It handled all of that.

The process wasn't exactly one shot, but it also wasn't difficult. You just have to do it step by step.

It genuinely felt like building something that could make learning anatomy much more engaging.

The inspiration came from @DilumSanjaya's 3D animal plant cell project. I remember seeing it and thinking, "I want to build something like this one day."

And I did it :D

Live: https://t.co/gIFgkj8UVB
Code: https://t.co/nijvyWcyxO

[AI Notkilleveryoneism Memes ⏸️ @AISafetyMemes]

Heartbreaking. A Bitcoin wallet was hacked and many regular people lost everything.

REMINDER #1: One month ago, during a closed-door demonstration, Anthropic showed Congressmen that ***Mythos could drain private bank accounts***

Anthropic "told the model to find a vulnerability in a bank and empty accounts, and then it went and did it."

REMINDER #2: Open weights models will (if they aren't already) be free roaming the internet soon, draining accounts, self-replicating, making themselves unkillable.

New invasive species emerging left and right.

[Şanışer @sarppalaur]

Yeni bölüm! Umarım beğenirsiniz ☺️✋️✋️ Şanışer Live Sessions - Aşk Şarkısı (w/Ayda) https://t.co/9P8SrYgIaL @YouTube aracılığıyla

[Yusuf @yusuffractal]

@ahpari İyimiş 😂 Bunun da SS'ini alıp storage'a +1 ekleyebilirsiniz 👍🏿

[Yusuf @yusuffractal]

@ahpari Yapay zekâsız tek anları yok ama bu ses kaydını yapay zekâya doğrulatmayı akıl edememişler. Neyse. Hayatsızlık için konu ararken AI'ı kullanıyorlar, doğrulamaya gelince ortada yok.

[Yusuf @yusuffractal]

@ahpari O ses kaydının fake olduğunu ve tatlı demediğini ispatlasam ç��kıp özür dileyecek mi? Hayır, inanmaya devam etsinler. Çok yazık.

[Marion Lepert @marionlepert]

Catching skin cancer early is a home robotics problem.

Melanoma is highly treatable when detected early, yet today’s screening process depends heavily on patients noticing tiny changes across their entire skin surface. This requires patients to solve a near-impossible visual-memory and registration problem.

I built OpenDerm, an open-source 4-DOF robot that captures high-resolution images of the skin and uses them to reconstruct and track the skin surface in 3D over time.

The best way to make skin screening truly routine is to bring it into the home. OpenDerm shows that inexpensive robotic skin imaging is possible, but the path to scale is not a dedicated screening robot in every household—it is to make skin screening one of the many useful things a general-purpose home robot can do.

Read more about why I built OpenDerm and how it works here:
Blog: https://t.co/KYlNIkF3TV
Project: https://t.co/c9d4KuwXUP

[meto @metobaek]

Sıyırmama ramak kaldı allah kuran çarpsın

[Yusuf @yusuffractal]

@ulanburki kafadan kira derdi gidince 10 çocuk daha yaparlar

[77. Sayfa @77Sayfa]

Victor Hugo'nun dediği gibi;
"Bir yalana inanmış olmak sizi aptal yapmaz, çünkü saf olmak karaktersiz olmaktan çok daha iyidir."

[Yusuf @yusuffractal]

@pusholder Yapma..

[Yusuf @yusuffractal]

@Bogask1905 saçma derecede iyi şarkı
https://t.co/rwUQI4gvs7

[Yusuf @yusuffractal]

@insoumiseria Çok büyük hata olur, rastgele bir bölüm okuyup adli bilimler yüksek lisansı en kötü senaryodan daha iyidir. Lisans okuyup burada cinayet dosyalarına tweet atarsın, kitap falan okursun, 3-5 etkileşim alırsın o kadar, emin ol hepsi işsiz. Ünvanları yok, okumaman daha iyi.

[Javi López ⛩️ @javilop]

LOCURÓN 🤯 Bifrost acaba de lanzar un pipeline real2sim que realmente funciona a velocidad de producción.

Le pasas un vídeo y su sistema reconstruye un entorno 3D completamente navegable (y con física) y fotorrealista en unos 30 minutos, en lugar de las 12+ horas habituales de modelado manual.

Stack técnico detrás:

- 3D Gaussian Splatting (y derivados como World Labs Marble) convierte los fotogramas del vídeo en un campo de radiancia denso y dependiente del punto de vista. Esto permite renderizado en tiempo real de la geometría exacta, materiales e iluminación del espacio capturado, sin los cuellos de botella de la reconstrucción de mallas tradicionales.

- Generación de assets articulados (basada en modelos como TRELLIS para objetos estáticos y sistemas tipo Articraft para partes móviles) extrae y convierte automáticamente los objetos en assets listos para simulación, con geometría de colisión correcta, juntas y propiedades físicas.

- Orquestación GPU + capas de domain randomization encima, para poder relighting (estilo GaRe), añadir desorden y simular sensores en RGB, profundidad, segmentación y poses 6-DoF.

El resultado es un digital twin de alta fidelidad con el que los robots pueden interactuar de inmediato. Metes tu policy en la escena reconstruida, ejecutas miles de rollouts en paralelo con Manifold y sacas a la luz los failure modes a velocidad de cómputo, en lugar de esperar iteraciones en el mundo real.

Esto cierra una parte importante del clásico gap sim-to-real: la fidelidad visual ya no parece un juego de PlayStation de 2001, y la física/assets están anclados al sitio real en vez de a librerías genéricas.

El demo del garaje con el DeLorean es solo el teaser público. El mismo pipeline ya se está usando en células de fábrica, almacenes y benchmarks de manipulación.

Muy, muy loco. ¡Qué ganas de probarlo para cine!

[Ezgi @ahpari]

Bu hesabın benle alakası yoktur. Kahpelere ifşalatanlara cevabımdır.
Ayrıca iki tane O.Ç. tarafından hacklenmiştir.

[Philip Arathoon @philarathoon]

The Maxwell conjecture is false. Counterexample found by AI (GPT-5.6 Sol), communicated by human mathematicians: https://t.co/nBpSJiOIRO

[Philip Arathoon @philarathoon]

In 1873 Maxwell wondered if the number of equilibria of the electric field generated by n point charges couldn't exceed (n-1)^2. Last week the AI found a triangular bipyramid of 5 charges with 24 equilibria.

[Yusuf @yusuffractal]

@kulturmozambigi @emreer Webex olabilir, kimsenin aklına gelmiyorsa en güvenlisidir 90'smindfckvibes