We’ve had AI image generators for years now, but they’ve gotten significantly better in recent months. ChatGPT, Gemini, Firefly, Midjourney, and other services now offer high-quality images that are nearly indistinguishable from real photos. Since most AI-generated images lack visible watermarks, the untrained eye might not be able to tell the difference between real photography and AI-generated content.
ChatGPT went viral a few months ago thanks to its new 4o image generation tool. Everyone wanted to try it, and OpenAI saw a big jump in new ChatGPT users. But if you choose Gemini instead of ChatGPT, you’ll get the same spectacular experience. In fact, Google just introduced Imagen 4 and Imagen 4 Ultra, which bring even more powerful image generation features to Gemini.
No sooner had Imagen launched than Soul became the talk of the town. It’s a new model from an AI startup called Higgsfield, and it’s going viral on social media. Like tools from OpenAI and Google, Soul can deliver images that look like real photos taken with a smartphone. But Soul is a “high-aesthetic photo model” designed to replicate the quality of real-life photography.
Soul promises “fashion-grade realism” that might make you “throw away your iPhone,” according to an announcement from Higgsfield on social media.
There’s another big difference between Soul and other AI tools for high-end image generation. Soul includes more than 50 presets to help you get the image you want faster.
These 50+ presets are available at Higgsfield and save you the trouble of describing camera settings to the AI. You can pick from options like “iPhone,” “Tokyo Streetstyle,” “Overexposed,” or “0.5 Selfie,” and the AI will understand the kind of image you’re aiming for.
You then use natural language to tell Soul what kind of image you want, without having to get technical about photo specs. Just describe what you want to see in the picture.
Soul has plenty of jaw-dropping examples on its website, some of which you can see above and below alongside the prompts used to generate them.
I’ve been following AI-generated images and videos for more than a year now, but the realism in these Higgsfield Soul images is unmatched.
As some people pointed out on social media, Soul removes the plasticky look often seen in images from ChatGPT and other AI generators. Everything looks authentic, from people to buildings to everyday objects.
The images don’t have visible watermarks, so they could easily fool unsuspecting users. With Soul, you can create almost anything instead of relying on your iPhone.
One thing Soul doesn’t support is image uploads. You can’t upload a photo of yourself and have the AI insert you into a variety of scenes.
To use Soul, you’ll need a subscription. For $9/month, you get 150 credits and watermark-free access. You can get started at this link. The only other thing to note is that Soul runs in your browser, not as a standalone app.
