This media is not supported in your browser
VIEW IN TELEGRAM
Google's new neuron creates a 3D model from a single image, a set of images or by promt in 1 minute.
CAT3D shows that it's getting easier to create 3D models from ordinary pictures.
The neural network uses a multiview diffusion model to create new views and simply inputs them into NeRF/3DGS.
This is a research paper by DeepMind and it is likely that these techniques will be used in Google applications.
Project page.
#service
CAT3D shows that it's getting easier to create 3D models from ordinary pictures.
The neural network uses a multiview diffusion model to create new views and simply inputs them into NeRF/3DGS.
This is a research paper by DeepMind and it is likely that these techniques will be used in Google applications.
Project page.
#service
👍7❤1
This media is not supported in your browser
VIEW IN TELEGRAM
All you need is a sketch and a detailed promo.
There are dozens of different styles on Scribble SDXL that you can freely experiment with. The bot is launched very simply from anywhere, even from a phone.
⭐️ Pumping skills here.
#Network⚡️ #AI⚡️ #Feature
There are dozens of different styles on Scribble SDXL that you can freely experiment with. The bot is launched very simply from anywhere, even from a phone.
⭐️ Pumping skills here.
#Network⚡️ #AI⚡️ #Feature
👍7⚡2
This media is not supported in your browser
VIEW IN TELEGRAM
ElevenLabs now generates any sounds for free
Type in promt and you get literally anything from rain noise to moans to nuclear sirens.
#Ai #ElevenLabs #future #audiogeneration #techupdate
Type in promt and you get literally anything from rain noise to moans to nuclear sirens.
#Ai #ElevenLabs #future #audiogeneration #techupdate
👍6
This media is not supported in your browser
VIEW IN TELEGRAM
ByteDance has introduced InstaDrag AI 🫣
It can change objects in a picture. For example, head rotation as in the example above.
InstaDrag is trained on video clips, so it understands how objects should change when changing their position.
Project page. The code will be available soon
#service #aitool
It can change objects in a picture. For example, head rotation as in the example above.
InstaDrag is trained on video clips, so it understands how objects should change when changing their position.
Project page. The code will be available soon
#service #aitool
👍3❤1
❗️ Breaking news: OpenAI former employee published a report with insights about the future of AI
Most importantly:
- AGI by 2027 is a reality;
- AGI is a key geopolitical resource right now. Forget about nuclear weapons - that's obsolete. Countries will do everything they can get AGI first, just as they once did with the atomic bomb;
- AGI would require a trillion-dollar computing cluster. Microsoft is already building a similar one for OpenAI;
- It will consume more electricity than the entire U.S. generates;
- The money for AGI will come from bigtech - Nvidia, Microsoft, Amazon and Google are alreadt laying out $100 billion in spending per quarter just for AI;
- By 2030, $8 trillion dollars each year will be invested in AI;
- If by this point you think you have a delusion in front of you - you don't. Comparable spending and global change occurred during the Manhattan Project and the Industrial Revolution
#news #ai #agi #breakingnews
Most importantly:
- AGI by 2027 is a reality;
- AGI is a key geopolitical resource right now. Forget about nuclear weapons - that's obsolete. Countries will do everything they can get AGI first, just as they once did with the atomic bomb;
- AGI would require a trillion-dollar computing cluster. Microsoft is already building a similar one for OpenAI;
- It will consume more electricity than the entire U.S. generates;
- The money for AGI will come from bigtech - Nvidia, Microsoft, Amazon and Google are alreadt laying out $100 billion in spending per quarter just for AI;
- By 2030, $8 trillion dollars each year will be invested in AI;
- If by this point you think you have a delusion in front of you - you don't. Comparable spending and global change occurred during the Manhattan Project and the Industrial Revolution
#news #ai #agi #breakingnews
👍6😱2
This media is not supported in your browser
VIEW IN TELEGRAM
SignLLM is the first multilingual sign language model that can generate gestures based on input text! 🔥
AI was trained on public sign language data, including American Sign Language (ASL) and seven others.
Project page ✅
Does anyone speak sign language? 🫰
#service #aitool #signlanguage #LLM
AI was trained on public sign language data, including American Sign Language (ASL) and seven others.
Project page ✅
Does anyone speak sign language? 🫰
#service #aitool #signlanguage #LLM
👍7❤🔥2
This media is not supported in your browser
VIEW IN TELEGRAM
Nvidia CEO Jensen Huang announced a number of new AI technologies 🤖
During a speech at the Computex conference, including next-generation "Rubin" chips, a new AI gaming assistant and AI tools for creating realistic avatars.
Nvidia's Rubin platform is due in 2026, with Rubin Ultra a year later, as part of what Huang called a "new industrial revolution."
Nvidia also showed off Project G-Assist, an artificial intelligence gaming assistant that provides context-aware assistance and personalized responses in PC games.
The company also unveiled ACE, a suite of AI services that simplifies the creation of digital avatars for fields such as customer service and healthcare.
Nvidia continues to be an AI leader, both in terms of the company's dominance in accelerating AI chips and new tools.
#news #ainews
During a speech at the Computex conference, including next-generation "Rubin" chips, a new AI gaming assistant and AI tools for creating realistic avatars.
Nvidia's Rubin platform is due in 2026, with Rubin Ultra a year later, as part of what Huang called a "new industrial revolution."
Nvidia also showed off Project G-Assist, an artificial intelligence gaming assistant that provides context-aware assistance and personalized responses in PC games.
The company also unveiled ACE, a suite of AI services that simplifies the creation of digital avatars for fields such as customer service and healthcare.
Nvidia continues to be an AI leader, both in terms of the company's dominance in accelerating AI chips and new tools.
#news #ainews
👍2❤1
This media is not supported in your browser
VIEW IN TELEGRAM
😳 The most viral video right now:
an AI artist showed what the last seconds of life of people who like to drive on the oncoming traffic look like.
In 7 days, the video has gained about 200 million views.
#aigeneration #aiart #video
an AI artist showed what the last seconds of life of people who like to drive on the oncoming traffic look like.
In 7 days, the video has gained about 200 million views.
#aigeneration #aiart #video
👍14🤯6❤🔥2❤1😁1
Apple announced "Apple Intelligence" at the WWDC 2024 conference — a new set of artificial intelligence features for iPhone, Mac, and other platforms.
Among the presented features:
▫️ Text processing: concise summaries, translations from any language, email sending, and error detection;
▫️ Image generation in three styles (sketch, illustration, animation) and emoji by text prompt. One-tap removal of unnecessary objects;
▫️ Searching for photos with specific people/animals/places in the gallery and messages;
▫️ Siri will be able to record or explain information from the screen, find needed information in almost any app by voice request;
▫️ Sorting notifications, automatic reminders, and much more.
The company claims that its AI features will be processed on-device to ensure data privacy.
🔻However, to take advantage of this benefit, users will need the A17 Pro chip — which is the whole point, to sell them all.💰
🔻Additionally, some AI features will still be transferred to the cloud, and to smooth over this nuance, Apple claims it will be a "private cloud" and Apple will never be able to access it.😁
Apple Intelligence will be available in English only in 2024. Support for other languages will come next year.
Among the presented features:
▫️ Text processing: concise summaries, translations from any language, email sending, and error detection;
▫️ Image generation in three styles (sketch, illustration, animation) and emoji by text prompt. One-tap removal of unnecessary objects;
▫️ Searching for photos with specific people/animals/places in the gallery and messages;
▫️ Siri will be able to record or explain information from the screen, find needed information in almost any app by voice request;
▫️ Sorting notifications, automatic reminders, and much more.
The company claims that its AI features will be processed on-device to ensure data privacy.
🔻However, to take advantage of this benefit, users will need the A17 Pro chip — which is the whole point, to sell them all.
🔻Additionally, some AI features will still be transferred to the cloud, and to smooth over this nuance, Apple claims it will be a "private cloud" and Apple will never be able to access it.
Apple Intelligence will be available in English only in 2024. Support for other languages will come next year.
Please open Telegram to view this post
VIEW IN TELEGRAM
The Verge
Every “Apple Intelligence” feature announced
Apple makes its AI pitch.
👍2
🤖Learn and grow with AI ( Artifical intelligence) 👽
Apple announced "Apple Intelligence" at the WWDC 2024 conference — a new set of artificial intelligence features for iPhone, Mac, and other platforms. Among the presented features: ▫️ Text processing: concise summaries, translations from any language, email…
Please open Telegram to view this post
VIEW IN TELEGRAM
Users have uncovered a comparison of Apple's neural networks with other models — the results are disappointing.
The models that work on-device provide answers of the same quality as the smallest models on the market (almost always junk) — only unquestionably outperforming Gemma-2B. The models running from the server are better: roughly on the level of GPT-4o, which is not surprising — OpenAI has just announced a partnership.
We are looking forward to trying out iOS 18 with its neural features to evaluate everything firsthand.
The models that work on-device provide answers of the same quality as the smallest models on the market (almost always junk) — only unquestionably outperforming Gemma-2B. The models running from the server are better: roughly on the level of GPT-4o, which is not surprising — OpenAI has just announced a partnership.
We are looking forward to trying out iOS 18 with its neural features to evaluate everything firsthand.
This media is not supported in your browser
VIEW IN TELEGRAM
This is a 4—hour lecture on creating GPT-2 from scratch in Python from OpenAI co-founder Andrey Karpati. The programmer literally starts with an empty file and builds a model.
You can try to create your own artificial intelligence here. 🔥
#ChatGPT⚡️ #Prompts⚡️ #AInews
You can try to create your own artificial intelligence here. 🔥
#ChatGPT⚡️ #Prompts⚡️ #AInews
🆒5👍2
This media is not supported in your browser
VIEW IN TELEGRAM
SD3 Medium with 2 billion parameters works even on laptops. The generation quality is stunning, the prompt understanding is amazing, and it generates images with text without errors!
Please open Telegram to view this post
VIEW IN TELEGRAM
👍1
We've had it: a user sent Google's AI Gemini a photo of a mushroom to see if it was edible 😱
Gemini said that this "yummy" mushroom is edible. In fact, it's one of the deadliest mushrooms on the planet
When first symptoms appear, it's ready too late to save a person who tried this shroom.
Conclusion:don't let AI tell you what to eat
#news
Gemini said that this "yummy" mushroom is edible. In fact, it's one of the deadliest mushrooms on the planet
When first symptoms appear, it's ready too late to save a person who tried this shroom.
Conclusion:
#news
❤3
Microsoft Generative AI Training Course ✍️
Its a great repository that will make you a pro in prompt engineering.
Inside there are 18 tutorials on different models, their features and integration into all systems. At the end you can design your own and get tons of experience.
You can become Sam Altman here
#guide #Course #useful
Its a great repository that will make you a pro in prompt engineering.
Inside there are 18 tutorials on different models, their features and integration into all systems. At the end you can design your own and get tons of experience.
You can become Sam Altman here
#guide #Course #useful
❤8👍7⚡2