The AI generation has evolved at a wild pace, and Google only threw another great competitor in this mix through Gemini Flash 2.0. You can play with the new Google’s photo creation tool Artificial Intelligence Studio.
Gemini Flash, as the name suggests, is very fast, especially faster than Dall-E3 and other photo creators. This speed may mean low -quality images, but this is not the case here, especially since all changes and promotions of the ability to produce a form image. However, if you really want good results, you should know how to speak to artificial intelligence. After a lot of experience and error, I collected five tips for the best art from Gemini Flash 2.0. Some of this post may seem similar to the advice of other creative creators of artificial intelligence, because they, but this does not make them less useful in this context.
Tell a story
The new most interesting feature of creating Gemini Flash images is that it is not only good for one -time illustrations, but it can actually help you create a visual story by creating a series of images related to a fixed style, settings and mood.
To start, you just have to ask you to tell you with a story and the number of times you want to give up the clarification with the procedure. The result will include those pictures accompanying the text.
For my project, I asked of artificial intelligence “creating a story from the heroic child’s dragon who protects a fairy queen from an evil processor in a 3D animation style. For each scene, create a picture.” I saw the above begin to appear. If there is a problem, you can rewrite any of the story and the form will renew the image accordingly.
Be specific
If you ask Gemini to make a “dog in a garden”, you may get a blurry golden recovery sitting in a mysterious place. But if you say, “a thin golden recovery sitting on a wooden bench in Central Park during the fall, with red and orange leaves scattered on the floor” – get exactly what you conceive.
Artificial intelligence models flourish on the details. The more you advance, the better your image. So for the above image, instead of just ordering a future city appearance, I asked “the view of the reactionary city at sunset, with neon signs glowing with pink and blue cars, in the sky, and people walking in reactionary style clothes.” Seven seconds, the result came.
Get the conversation
One of my favorite objects about the new Gmini Flash is that you can have a conversation with it without losing a lot of speed. This means that you do not have to get everything correctly. After creating an image, you can literally chat with artificial intelligence for adjustments. Want to change colors? Personal addition? Make mood lighting? Just ask.
In the above photo, I started asking for “a comfortable reading angle with a fireplace, book shelves full of novels, and a large comfortable chair.” Then I improved it by requesting “making it at night with soft and warm lighting”, then continued it by requesting “adding a sleeping cat on a arm chair”, and ended through the request of Amnesty International “to give the room the ancient aesthetic of Victorian.” The end result on the left looks just like what you imagined, and makes Gemini feel a technical assistant, able to adapt to what I want without starting zero every time.
Gemini Flash corresponds to ChatGPT
Google boasts that Gemini is full of real knowledge, which means that you can get historical accuracy, realistic cultural details and real images if you ask for. Of course, this requires to be specific. For example, if you ask for a “Viking Warrior”, you may get something more like game of thrones a personality. But if you say, “Viking warrior is historically accurate from the ninth century, wearing a detailed shield for the series, a round wooden shield, and a traditional helmet” – you will get something more accurate.
As a test, I asked artificial intelligence to make “the ancient Mayan city at sunrise, with tall stone pyramids, the fertile forest surroundings, and people who wear traditional Mayan clothes.” It is not perfect, but it is very similar to the real thing of previous versions, which sometimes return with almost an Egyptian pyramid.
Write quickly
Most artificial intelligence models have long been struggled with the presentation of the text, and the conversion of words into unread beads. Even the best models today that can do it somewhat to do this and get them properly can take a few attempts. However, Gemini Flash is shockingly good to integrate the text into the pictures quickly and buy. To be very specific can help.
Thus, the image above was created by asking artificial intelligence to “make an old -style travel label that says” London’s visit “in bold, retro -print, and includes a flowing clarification of the city.”