Can Google Bard generate images?

While Google has touted its AI tech for years, particularly on its Pixel phones, the rise of generative AI systems like ChatGPT seems to have caught it by surprise, forcing it to play catch-up. The spearhead of that effort is Google Bard, which connects to a number of other Google apps and services. How powerful is it, though? Does Bard have any support for image generation?

Can Google Bard generate images?

No, at least not yet. For the moment Bard is a ChatGPT-style chatbot, meaning you can ask it knowledge questions, have it summarize emails, articles, and documents, and assign it some other limited tasks.

The chief advantage of Bard is its previously-mentioned links with the Google ecosystem. If you ask it to plan a group trip to Austin, New York, or Saskatoon, for instance, it can fetch open dates from Gmail, pull flight information from Google Flights, and get airport directions from Google Maps. You might even get relevant YouTube videos.

Google does offer image generation, but only through SGE, its generative AI search tech. Even then this is an opt-in experiment and not widely available, so you’ll probably have to turn to one of the more established image platforms.

What are some of the best AI image generators?

Midjourney

Midjourney has become a go-to choice for many, since its V5 iteration has overtaken most of the competition. In some cases the results it produces can look photorealistic, albeit with some minor mistakes that give away their AI origin. You can go in the opposite direction if you like, even creating anime-style content.

You can’t access Midjourney through a dedicated app or website. Instead you need to use Discord, a real-time chat service most commonly associated with gaming. Discord does run on every platform you might care about however, including the web, and the chat commands you need to know aren’t that complex.

The real drawback to Midjourney is its cost. It costs at least $10 per month or $96 per year, and you’re encouraged to spend more to get a higher share of fast rendering.

DALL-E

Next to Midjourney, DALL-E is perhaps the best-recognized name in AI image generation, since it broke into the mainstream way back in 2021. It doesn’t hurt that the company behind it is OpenAI, best known for ChatGPT.

In essential aspects it’s similar to Midjourney, but the real advantage is pricing. If you access it through Microsoft’s Bing Image Creator, it’s completely free, although you’re assigned a fixed number of credits to use that only replenish weekly. If you want to escape the confines of Bing you can buy credits directly from OpenAI, but you’ll have fork over $15 at a time. That does get you 115 images, so it might be worth it if you expect to do a sizable chunk of AI work.

Stable Diffusion

Stable Diffusion is completely free and open-source. If you own a powerful computer with a dedicated graphics card, you can even download and run the AI model yourself. That translates to a GPU with 8GB of VRAM, so you’ll probably want an NVIDIA RTX 3060 or better.

You can try Stable Diffusion online at stablediffusionweb.com. Just be aware that while you don’t need an account for this, anything generated will be visible to others, and you won’t be able to browse your own history. Thankfully that also means that people can’t tie an image to you personally, and you’ll be able to browse other recent prompts.