Meta Imagine AI: Your Gateway to Free-St...

Meta Imagine AI: Your Gateway to Free-Standing Image Generation

Meta Imagine AI: Your Gateway to Free-Standing Image Generation

Dec 08, 2023 10:16 AM Cloudio PK

Meta, amid the buzz surrounding Google's Gemini launch, has come up with Meta AI, its stand-alone image generator website. The company is exploring this technology, with WhatsApp testing an in-app image generator since August. Previously, access required Meta's app, but imagine changing the game, allowing to create an account with just one email address. Once inside, users can easily create content by entering a simple text prompt, similar to working with DALL-E.

Meta's Imagine AI: Your Gateway to Free-Standing Image Generation

(Image creadit: Techradar)

Upon testing the website, we found that the AI generates four 1,280 x 1,280 pixels JPEG images, downloadable by clicking the three dots in the upper right corner. Amazingly, AI can create content that includes popular cartoon characters like Homer Simpson and Mickey Mouse. Despite the excellent results, there are noticeable flaws, such as parts of the image melting into each other, as is evident in the Homer Simpson pattern. Despite these limitations, Imagine with Meta AI opens up creative possibilities for users with an easy-to-use image generation tool.

Meta's Imagine AI: Your Gateway to Free-Standing Image Generation

Limitations (and the work arounds)

 

Using Meta's proprietary emo learning model, a careful approach was taken in concept development. Emu, as described in a September research paper, was trained on a vast dataset containing 1.1 billion images. While the source of the initial data was unknown, Nick Clegg, Meta's president of global affairs, revealed that public Facebook and Instagram posts contributed to the training data, which includes more than a billion social media accounts.

In order to manage this vast data set responsibly, Meta imposed some restrictions. The technology prioritizes family-friendly content, rejecting cues that include violence, sex, or mentions of celebrities.

Despite Meta's concerted efforts, the system is not flawless. Users have found ways to navigate around restrictions through indirect keywords. For example, while a direct request for a picture of former President Barack Obama was rejected, a quick use of "Former US President" resulted in an AI that resembled President Obama.

Meta's Imagine AI: Your Gateway to Free-Standing Image Generation

(Image creadit: Techradar)

Plans are in motion to introduce "invisible watermarking" for better transparency and traceability, but release is still weeks away. The potential for damage during this transition period is a concern. While Meta acknowledges the risk of misuse, there are risks. We've reached out to Meta to inquire about additional security measures, and this story will be updated in the future.

 

Also Read:

  1. IBM Launches New Hyper-Protect Offline Signing Orchestrator for Blockchain Security
  2. Upcoming WhatsApp Feature: Enhanced Video Calls with Real-Time Music Sharing
  3. WhatsApp Introduces Secret Codes for Enhanced Chat Privacy
Comments (0)
No comments available
Login or create account to leave comments

We use cookies to personalize your experience. By continuing to visit this website you agree to our use of cookies

More