One of the standard complaints about AI image generators is that the ‘photographic’ images they create are often very obviously not photographic. Improvements to various models have started to change that stereotype, and one of the best examples I’ve seen is Leonardo.AI.
Like many AI image generators, Leonardo.AI works by transforming descriptive text prompts as well as other sources and transforming them into vivid, detailed visuals. What’s notable is that even with a free account, you can get some truly amazingly realistic images.
Creating photographs with Leonardo.AI
You need to sign up for a free account at Leonardo.AI, but when you do, you can create many images daily. To make the kind of photographic realism seen above, it’s best to use the model/preset options the platform offers. They all have their benefits, but the Cinematic Kino option is the one I found best for the kind of photographic images I wanted to see.
On the free tier, you automatically get four images for a prompt, but I’ve picked out my favorites from the prompts I tested. You can write out whatever prompt you wish, but I experimented by using the option to randomly generate a prompt to come up with a couple, and the details are extensive.
The image above is “An ancient, battle-hardened knight stands tall in gleaming armor, his weathered face revealing tales of countless victories and defeats. This breathtaking portrait, whether captured in a painting or photograph, showcases every intricate detail of his protective gear, shimmering under the intense play of light and shadow. The quality of the image is impeccable, with every piece of armor reflecting a sense of timelessness and resilience.”
The one below comes from “In the dimly lit alley, a mesmerizingly enigmatic sorcerer stands, exuding an ethereal aura. The photograph, reminiscent of a neo-noir film still, captures the essence of mystery and magic. The sorcerer’s flowing robes shimmer with iridescent patterns, reflecting the dim streetlights, while a crown of glowing crystals hovers above their head. Their eyes, a piercing shade of emerald green, seem to penetrate the darkness with an otherworldly power. The image is a stunning masterpiece, blending fantasy with a touch of gritty realism.”
AI prompt help
Leonardo.AI won’t just come up with useful prompts, but you can even ask it for help refining your broad ideas. For example, I started by writing “A steampunk character enjoying New York City.”
When I clicked on the “improve prompt” button, the AI wrote: “A woman wearing a steampunk outfit smiles in a steampunk rendition of New York City. The photograph captures every intricate detail of her outfit, from the intricate gears to the billowing steam. Her outfit is a masterpiece of brass and leather, with goggles perched on her forehead adding a touch of adventure. The city is a mesmerizing blend of Victorian architecture and futuristic inventions, all brought to life with vibrant colors and impeccable detailing. This incredible scene draws viewers into a world of wonder and excitement.”
As far as expansions go, that’s a pretty comprehensive set of additions to my initial broad idea. The inclusion of so many extra adjectives and adverbs suggests they are the key to making a better image, as is the evocation of emotions the picture is supposed to make you feel. I don’t know if it’s a world of wonder and excitement exactly, but it’s a pretty great look for a steampunk costume.
For the next image, I started with “a group of people play dodgeball in a gym.” The AI returned with the following prompt and image: “A dynamic action photograph of a mixed group of athletic men and women playing an intense game of dodgeball in a brightly lit gymnasium. The players are captured mid-motion, with some dodging, some throwing, and others diving to avoid the ball. They wear vibrant team uniforms: one team in red tank tops with blue shorts, and the other in blue tank tops with red shorts.”
Leonardo.AI is also good at writing prompts based on uploaded images. As a test, I uploaded a picture of myself at a restaurant. The description it came back with was incredibly accurate in most elements. According to Leonardo.AI, the image showed a “portrait-style photograph featuring a man with a light to medium skin tone, smiling broadly. He has short, dark brown hair and a well-groomed beard. His eyes are dark, and he is wearing a black button-up shirt. The background is softly blurred, suggesting an indoor setting with warm lighting, possibly from overhead bulbs, creating a cozy ambiance. The man is positioned slightly off-center, and the lighting highlights his facial features, giving the image a warm and inviting feel.”
I decided to play with that description using one of the pre-sets, specifically the “soft retro-future” option. The result doesn’t look exactly like me, but it is not a million miles away. Notably, soft retro-future apparently translates to “leather jumpsuit with silver accents and standing in the rain.”
The free tier is useful but limited in some ways. Paying for a subscription starting at $10 a month gets you access to Fast Mode, which speeds up the image creation and provides more tokens for producing more pictures. Free accounts are fine for casual use, but if you’re planning to create a lot of visuals, a paid account ensures you won’t run out of credits halfway through your creative binge.
Another premium feature is custom model training, which uses your preferences to train the AI to create images in a specific style or theme. For example, if you’re obsessed with mid-century modern interiors, you can teach the AI to focus on that aesthetic. Paid users also get access to higher-resolution outputs that work better for printing. Still, even at the free level, it’s nice to find an AI image generator that can make people who look like they belong in the real world, or at least in a film about a princess.
You might also like
https://cdn.mos.cms.futurecdn.net/TmpzCqp4Nztz6mBjdXdevY-1200-80.png
Source link
erichs211@gmail.com (Eric Hal Schwartz)
