If you’ve tried at least some of the text-to-image AI art generation services that have been launched in recent years, such as MidJourney or OpenAI’s various versions of DALL-E, you’ve probably noticed some similarities. Chief among them: they all look pretty much like a chat interface. There is usually room for the user to enter text prompts, and the application responds with an image embedded in a message.
While this interface works well for many users and application developers, some people believe it is limiting and ultimately not what established artists and designers want when using AI at work. But now based in San Francisco Visually electric is here to provide a different approach. One that is the new startup – which one emerges from stealth today after a seed round last year from Sequoia, BoxGroup and Designer Fund in an undisclosed amount, he says it’s better suited to visual creativity than texting back and forth with an AI model.
Dunn previously led product design and brand at mobile website building company Universe, and before that was head of design at Playspace, an acquisition of Google.
Visual Electric aims to be a “much better product” for AI art, visual design, and creativity for business users, such as independent designers, in-house designers at major brands, and even “pro-sumers.”
The company is deliberately not launching its own underlying AI image generator machine learning (ML) model. Instead, it is built on top of the open source Stable Diffusion XL model (currently a lawsuit is being filed in the form of a copyright lawsuit by artists against the individual company who developed it, Stability AI, as well as Midjourney and other AI art generators).
This is because Dunn and his two co-founders – Visual Electric chief product officer Adam Menges, former co-founder of Microsoft acquisition Lobe; and chief technology officer Zach Stiggelbout, also formerly of Lobe, believe that AI models for image generation are becoming commonplace, and that it is the front-end user interface that will most differentiate companies and drive the successes of the failures will separate.
“We just want to create the best product experience,” says Dunn. “We are truly model agnostic and happy to exchange any model that gives users the best results. Our product can easily accommodate multiple models or the next model that comes out.”
How Visual Electric differs from Midjourney, DALL-E 3 and other AI art apps
What is the biggest deviation of Visual Electric compared to previous image generators? It allows users to generate and drag and move their images across an infinite virtual ‘canvas’, allowing them to compare images side by side rather than the top-down ‘linear’ form factor of other chat-based AI art generators apps, the latter of which forces users to scroll back up to see their previous generations. Users can continue to generate new sets of four images at a time and move them anywhere on this canvas.
“Creativity is a non-linear process,” Dunn said. “You want to discover; you want to go down different paths and then go back to an idea you were looking at before and take that in a new direction. Chat forces you into this very linear flow where it’s like you have a starting point and an ending point. And that’s just not how creativity works.”
While there is of course room to enter text prompts, this box has been moved to the top of the screen instead of the bottom as with many chat interfaces.
To help overcome the initial hurdle some users face (not knowing exactly what to type to get the AI to produce the image they have in their mind), Visual Electric provides a drop-down list of autocomplete suggestions, similar to what a user finds when typing a search query on Google. These are all recommendations based on what Visual Electric has seen from early users, resulting in the highest quality images. But a user is also free to deviate from this entirely and type in a custom prompt as well.
Additionally, Visual Electric’s web-based AI art generator offers a range of useful additional tools for customizing the prompt and style of the resulting images, including preset styles that mimic common styles in the pre-AI digital and print art worlds, including ‘marker ‘, ‘classic animation’, ‘3D rendering’, ‘airbrush’, ‘risograph’, ‘stained glass’ and many others – with new styles added regularly.
Instead of the user having to specify the aspect ratio of their image – 16×9 or 5×4 are two common examples – within the prompt text, they can select it as an option from the buttons in the drop-down list or from a handy sidebar at the right side, putting it in more direct competition with Adobe’s Firefly 2 AI art interface that offers similar functionality.
This sidebar also allows the user to specify dominant colors and elements they want to exclude from the resulting AI-generated image, also entered via text.
Furthermore, the user can click a button to ‘remix’ or regenerate their images based on the initial prompt, or to ‘retouch’ selected areas of the image and have AI regenerate only those areas they highlight with a digital brush whose size the user can use. can vary, keeping the rest of the image and adding to it in the same style. For example, if you don’t like the hair on your AI-generated subject, you can “retouch” the underlying Stable Diffusion XL model and instruct it to redo only that part of the image.
There is also a built-in upscaler to improve the resolution and details of images.
“These are the tools that represent what we think of as the AI-native workflow in the order you use them,” says Dunn.
Pricing, community and early success stories
As Visual Electric launches publicly today, the company has been quietly alpha testing with several dozen designers, who Dunn says have already provided valuable feedback to improve the product, as well as promising results on how Visual Electric has been used to to help improve the product. practical situations on the work floor of companies.
Dunn mentioned one client in particular – withholding the name for confidentiality reasons – who had a small team of designers working on creating menus and other visual materials for more than 600 universities.
In the past, this team would have spent a lot of time sifting through stock photos and finding images that matched but also accurately represented the items on a school dining hall menu, and had to manually edit the stock photos to make it. more accurate.
With Visual Electric, they can now generate entirely new images that meet menu requirements and edit parts of them without having to use Adobe Photoshop or other competing tools.
“They are now able to take a non-creative task and turn it into something that is very creative and much more satisfying, and they can do it in a tenth of the time,” Dunn claimed.
Another key differentiator that Visual Electric offers is an “Inspiration” feed consisting of AI-generated images created by other users on the platform. This feed, a grid of different sized images reminiscent of Pinterest, allows the user to hover over the images and see their clues. They can also grab and ‘remix’ any images on the public feed and import them to their private canvas.
“This was an early decision that we made, which is that we think with generative AI there is an opportunity to bring the network into the tool,” Dunn explains. “Right now you have inspiration sites like Pinterest and designer-specific sites like Dribble, and then you have tools like Photoshop, Creative Suite and Figma. I’ve always found it strange that these things aren’t connected somehow, because they’re so connected.”
Visual Electric’s pricing, which launches publicly in the US today, is as follows: a free plan offering 40 generations per day at slower speeds and a license limited to personal use (so no sale or use of said images as marketing material); a standard plan of $20 per month or $16/month, paid annually upfront, which allows community sharing, unlimited generations at 2x faster speeds, and a royalty-free commercial use license; as well as a pro plan for $60 per month or $48 per month, paid annually in advance, which offers everything the last two plans offer, but also even higher resolution images, and, critically, privatized generations.
Frequently Asked Questions – FAQs
Visual Electric offers a non-linear canvas, allowing users to drag, move, and compare images creatively.
The platform provides autocomplete suggestions and a drop-down list based on user preferences.
Visual Electric provides a free plan, a standard plan at $20/month, and a pro plan at $60/month.
Yes, the standard and pro plans include a royalty-free commercial use license.
The “Inspiration” feed allows users to view, remix, and import AI-generated images from others on the platform.
Clients report significant time savings and increased creative satisfaction, as seen in a case study with a university design team.
As Visual Electric emerges from stealth, it brings a refreshing breeze to the realm of AI art generation, challenging the status quo set by chat interfaces. By adopting a non-linear canvas, this startup liberates creativity, enabling users to seamlessly drag, move, and compare images in an infinite space. The decision to build upon the open-source Stable Diffusion XL model reflects Visual Electric’s commitment to prioritizing user experience over underlying AI models. With a suite of intuitive tools, from autocomplete suggestions to diverse style presets, Visual Electric empowers designers and creators with unprecedented flexibility. As the platform launches publicly, the success stories of design teams attaining both efficiency and creative satisfaction underscore the tangible impact of this innovative approach. Visual Electric not only envisions a new era for AI art but actively shapes it, where artistic exploration knows no bounds.
Article Internal Images Credit: VentureBeat