Image Generation with Tensor.Art
Tensor.Art lets you generate images from text prompts. While the results may vary, with some experimenting, you can create inspiring visuals quickly. With better prompts and workflows, it’s a useful tool for quick ideas and inspiration.
To begin, we go to https://tensor.art and register with just one click using a Google account.
On the main page, the interesting part starts with entering your prompt.
I used this text and experimented with different models without aiming for anything explicitly grandiose. I wanted to see how adequate it is.
The guy is in the snapback with a Los Angeles logo. On the table we can see microchips and sensors for the smart home. The laptop has big wide monitor. We see the guy from his back. The room is full of grass and waterfalls
After entering the query and clicking the generate button, you'll be directed to a page where you can choose the model and proportions. Additionally, in the top right corner, there will be information about available credits that were provided upon registration.
But don't worry, there will be enough for you to play around with.
Of course, I played around with both the models and the query a bit, but the number of successful images is quite low.
Here's an example of what it generated for me...
Examples of unsuccessfully generated images
If you look closely, the second and third images might still be usable, but what happened to the guys in the first, fourth, and last images? Why are they chopped by the table?
In the first photo, something went wrong with the cap, which now has two peaks.
The background fill seems better than the main scene. I see a lot of greenery, paintings, shelves with items, and even details like pencils, lamps, and flower pots on the tables. The laptops almost everywhere look realistic. Overall, the proportions of the guys relative to the interior seem fine.
There are also good examples like these...
They can be used as cover images for an article or simply to add some visual interest to the text. They indeed lack the detail for realism. But I didn't use more complex techniques. These images were created based on a simple description without tweaking.
With the use of other tools and complex workflows, much better results can be achieved. People share their work on Tensor.Art, which showcases how detailed the results can be and what complex chains are used for this.
Just as an example, take a look at these two posts. And thousands of others.
Caption Master
Podmike
Isn't it impressive? If it's not a replacement for social media photos, then at the very least, it's a tool for inspiration and developing new techniques and approaches in graphics. It can combine and mix things we couldn't even imagine. And the best part is that you can test your idea in seconds.
Need a new spaceship in the shape of a fish? Need a great angle for a photo in the mountains? Want to imagine how a car would look painted in salad green with purple discs?
Spaceship in form of fish close to the ice planet
mountains landscape selfie photo / with friends
light green color car with violet wheels
A well-formulated request and a bit of imagination can take you very far, and this can be the beginning of something very interesting. Don't limit yourself by rejecting the help of AI.
Like many others, I also love looking at artworks created by people and published on the platform https://www.artstation.com. There, creativity knows no bounds. Now imagine what such an artist could achieve using AI as tools, akin to brushes. Photobashing could reach an unreal level.
Intrigued? Now give it a try yourself!
AI Speaks
Generating images with artificial intelligence has numerous potential benefits and prospects. On one hand, it can significantly increase productivity in creative fields by helping to generate ideas and design variations more quickly. On the other hand, it can provide access to creating art and design for people who previously lacked the necessary skills or resources. However, it's important to consider ethical and legal aspects such as copyright and the use of generated images. Regarding achieving better results, with the development of technologies and improvement of algorithms, it's possible that image generation will become even more efficient and high-quality. In the future, these images may find applications in various fields, from art and advertising to science and education. However, it's important to maintain a balance between using technology and preserving human creative potential and uniqueness.
