Google Debuts Imagen 4 AI Image Generation Tech

Google has recently announced its newest text-to-image model, Imagen 4. This updated version promises significantly improved text rendering capabilities compared to its predecessor, Imagen 3. Alongside Imagen 4, Google also launched Imagen 4 Ultra, a premium model designed for users who need highly precise image generation from their text prompts, albeit at an additional cost.

Diving Deeper into Imagen 4 and Imagen 4 Ultra

Google positions the standard Imagen 4 model as the ideal choice "for most tasks." For those requiring even greater fidelity and adherence to complex instructions, Imagen 4 Ultra is presented as the deluxe option. This version is engineered to follow more precise text prompts, with Google promising "strong" output results that aim to rival other leading image generators like Dall-E and Midjourney.

Access and Pricing: What You Need to Know

Both Imagen 4 and Imagen 4 Ultra are currently available in a paid preview through the Gemini API. Additionally, users can experiment with the models via limited free testing in Google AI Studio.

Regarding cost, the standard Imagen 4 model is priced at $0.04 per image. For the enhanced capabilities of Imagen 4 Ultra, the price increases by 50 percent to $0.06 per image.

A Look at Imagen 4's Output Quality

Google showcased Imagen 4 Ultra's abilities with various examples. One such demonstration was a three-panel comic strip generated from a prompt describing a small spaceship being attacked by a giant blue space lizard, complete with sound effects like "Crunch!" and, somewhat curiously, "Had!!" The resulting image reportedly followed the prompt accurately and achieved an acceptable visual quality, resembling a toon rendering from a 3D application.

Another example involved a prompt for a "front of a vintage travel postcard for Kyoto: iconic pagoda under cherry blossoms, snow-capped mountains in distance, clear blue sky, vibrant colors." Imagen 4 successfully translated this into an image that matched the description, though the style was noted as somewhat generic and lacking distinct artistic charm. Other showcased images included a hiking couple waving from a rock summit and a mock "avant garde" fashion shoot. While the generated images were generally of good quality and accurately reflected the input prompts, they reportedly retained a noticeable machine-generated appearance.

Critical Perspective: How Does Imagen 4 Stack Up?

Overall, Imagen 4 is considered a fine and mild improvement over previous iterations. However, it may not be groundbreaking enough to significantly impress users, especially when compared to established market leaders such as Dall-E 3 and Midjourney 7.

Furthermore, there's an observation that public enthusiasm for AI-generated art might be waning. The primary use case for such technology currently appears to be in creating spammy advertisements commonly found on social media platforms and at the bottom of online articles. This context suggests that while Google continues to advance its technology, the broader reception and utility of AI art tools like Imagen 4 are still evolving.