Text-to-Animation: A Lost Race for Some

Stability AI claims to make some groundbreaking developments in animations. With the release of an SDK for Stable Diffusion, the platform now allows users to create animated videos with text, image or video prompts.?

The SDK has a variety of parameters, such as, animation prompts, style presets, cadence, and?FPS (frames per second), among others. It also allows users to fine-tune things like colours, 3D depths, and even post-processing and is like having the ultimate control panel for your creative vision.?

Sounds cool! But is it as revolutionary? What if these developments have already happened and Stability AI is lagging in the race? Deforum, an open-source software for animations, which uses Stable Diffusion’s image-to-image function to create animations, has already launched a notebook for text-to-animation.

RunwayML, however, poses true competition to Stable Animation. Its soon-to-be launched Gen-2, which goes beyond the capability of Gen-1’s style transfer and video modifications, can create video from just a text prompt. The most important difference is that the product can be used by non-technical users with ease.?

The power and capability of RunwayML can be compared to OpenAI’s DALL-E, which once captured the imagination of people with its powerful text-to-image generation capabilities. With the evolution of text-to-anything technology, chatbots made the most hype and OpenAI focussed all its attention towards it.?

Currently, Midjourney, Stable Diffusion,?Imagen , and others have zoomed far ahead in the text-to-image generation race with DALL-E 2 lurking far behind. Meanwhile, the evolution of DALL-E 2’s successors have also slowed down and not gone far beyond image.

It will be interesting to see emerging platforms like RunwayML and a bunch of others carve their niche in the text-to-animation space.

Read the full story?here .

Capgem-AI

If you listen to the speech recordings of the CEOs of IT giants such as TCS, Infosys, Wipro and others, (keep the speed at 2x, if you will), you would be amazed at the number of times you hear the word “generative AI”. But there’s little evidence of these companies actually working on any such generative AI products.?

However, there’s Capgemini, which utilises pre-built models and frameworks like TensorFlow and PyTorch, and customises them for specific use cases. They also have in-house tools like AI Class Box and IDEA to accelerate MLOps and aid clients in cloud migration. Additionally, they prioritise AI-driven automation and containerization for model reusability.

Read the full story?here .

How to Overcome Hardware Issues in AI

The biggest hurdle in building a ChatGPT like model is the availability of outdated hardware — GPUs. American computer scientist and Turing award winner Jack Dongarra, in an exclusive interview with?AIM?gave insights into how the hardware technology can be improved to meet the current AI demand from people.?

Dongarra envisions a multidimensional approach to computing, where various technologies, including CPU, GPU, machine learning, neuromorphic, optical, and quantum computing, converge within high-performance computers.?

Read the full story?here .

Prompt Engineer, a Hot Job

Have you ever regretted not making a career out of computer science? It’s time to shed the sinking feeling because the latest chatbot-based job called ‘prompt engineering’ has left even computer engineers behind in terms of pay.

Prompt engineers are being paid much more than Python developers. Anthropic, a Google-supported AI startup, has announced salaries of up to $335,000 for the role of ‘Prompt Engineer and Librarian’. Similarly, Klarity, an automated document reviewer, has put up a job for an ML Engineer, who is capable of masterfully coaxing AI tools to deliver optimal outcomes, and is willing to pay up to $230,000! It seems the quest for the perfect prompt is on, and the rewards are nothing short of remarkable.

Read the full story?here .