Now that anybody can use DALL-E 2. It’s kind of scary, this could be dangerous.
Access to DALLE-2 has long been a prized perk for those eager to explore the cutting edge of AI art generation. DALL-E 2 has been more careful and conservative than other text-to-image AI picture generators, such Stable Diffusion, leading to a large waiting list of those seeking to acquire access.
The organisation that created it, OpenAI, has officially declared that the waiting list is no more. Beginning on September 28, anybody may use DALL-E 2 to create photographs using artificial intelligence. Pricing will continue to be the same as it has been since a credit-based system was implemented in July (see below for details).
View this post on Instagram
Since its beta release in April, DALLE-2 has been regarded as one of the most sophisticated AI picture producers, particularly when it comes to producing photorealistic images. Most likely, you’ve already seen examples of DALL-E 2 output circulating online with its distinctive watermark of coloured squares in the lower right. DALL-E 2 is usually a lot more convincing, however it still has some of the nightmarish problems that other AI image generators have.
We’ve removed the waitlist for DALL·E. Sign up and start creating immediately! https://t.co/TCVDl1iMeU
— OpenAI (@OpenAI) September 28, 2022
One of the most popular text-to-image generators is now available to all creatives, providing them with a new tool to test out new concepts and maybe speed up their processes. Open access seems fairer. But it’s also a little scary. DALL-E 2 can create incredibly realistic photos, and it recently lifted a restriction on modifying genuine human faces, which raises the possibility that people’s photographs might be changed without their consent in order to commit fraud or create deepfakes.
Stability AI’s open-source Stable Diffusion, has come under fire for allowing users to make violent and obscene content since it has no limits. DALL-E 2 will now test OpenAI’s own security measures, which the company claims are already in place and have been strengthened through “learning from real-world use.” These advancements, stated in a blog post, are what now provide open access.
The business claims that to stop banned content, a combination of automatic and human monitoring is used. The dataset used to train DALL-E 2 was filtered to exclude overtly violent, pornographic, or offensive pictures. Attempts to produce images of public figures are prohibited. According to OpenAI, “We have built new detection and response algorithms to stop misuse, and we have strengthened our filters to more effectively reject efforts to produce pornographic, violent, and other content that violates our content policy.”
Additionally, a lot of creators could be worried about their employment. Why would someone hire a designer or artist to generate a picture when they could theoretically make whatever image they wanted by just putting a description into a prompt box?
To make the most of the technology, however, someone with good creative judgement and a lot of guidance are still required. A brand-new career path is emerging: writing AI prompts. Marketplaces for AI prompts that sell prompts for certain image types already exist.
What is DALLE-2?
DALLE 2 is a text-based AI picture generator based on machine learning. The name is a mashup of “Dali” (as in Salvador) and Pixar’s “WALL-E.” It’s a generative tool, which means users can create art from nothing by utilising text prompts, and it was developed by the artificial intelligence startup Open AI.
DALL-E 2 will utilise its training on millions of existing photographs and descriptions to build what it believes matches the subject and style of the image that users say they want to acquire. You might include something like “anthropomorphic hedgehog wearing a suit and tie, studio lighting, award winning high quality photos,” for instance. Yes, someone actually tried it; the results are seen in the tweet below.
Additionally, DALL-E 2 may be used to modify or update already published material. Users may extend a photograph beyond its initial frame thanks to a recent feature called outpainting.
ancient egyptian hieroglyphics depicting a man sitting at a computer with the word Dalle on the screen#dalle2 pic.twitter.com/EBzR8ESZ70
— Dylan (@DylnRhyman) September 28, 2022
What can DALL-E 2 be used for?
By writing text prompts, DALL-E 2 may be used to generate graphics in a broad range of styles. Even well-known artists’ styles can be recreated through it. Although up to now solely as a gimmick, brands like Heinz have experimented with utilising it to create marketing and promotional materials.
Plastic surgeons and architects have been using the tool to envision what their work might look like. A German art institute is running an evolving AI art exhibition that uses DALL-E 2 to generate art based on trending topics on Twitter. The tool may be useful for speeding up the creation of backgrounds for gaming and other uses. However, it’s mostly used for creative experimentation, concept generation, and idea testing.
¿Puede la #IA fusionar obras de Van Gogh, Monet, Munch y Hokusai?
En @OrbAmsterdam se formularon esta pregunta y lanzaron el reto a @OpenAI. El resultado es impresionante (nunca mejor dicho).
Así funciona la nueva capacidad de #outpainting de #DALLE2👇🏼pic.twitter.com/WWeyRvea6x
— Xavi Gassó (@XaviGasso) September 28, 2022
How much does using DALL-E 2 cost?
The first two months of DALLE-2 were free, but in July OpenAI started charging for the credits required to create artwork on the site. All new users receive several free credits to utilise, but after that, only 15 free credits are given out each month.
Additional credits can be purchased by users for $15 for 115 credits (enough to generate 460 square images). Artists who require financial support are welcome to apply for subsidised access through OpenAI.