Finally, DALL-E 2, OpenAI’s image-generating AI system, is available as an API, meaning developers can integrate the system into their apps, websites, and services. In a blog post today, OpenAI announced that any developer can start harnessing the power of DALL-E 2 – which over three million people now use to produce over four million images per day – once that he created an OpenAI API account as part of the public beta.
DALL-E 2 API price varies by resolution. For 1024×1024 images, the cost is $0.02 per image; 512×512 images cost $0.018 per image; and 256×256 images cost $0.016 per image. Volume discounts are available for companies working with the OpenAI Enterprise Team.
As with the DALL-E 2 beta, the API will allow users to generate new images from text prompts (e.g. “a fluffy bunny hopping in a field of flowers”) or modify existing images . Microsoft, a close OpenAI partner, uses it in Bing and Microsoft Edge with its Image Creator tool, which lets users create images if web results don’t return what they’re looking for. Fashion design app CALA uses the DALL-E 2 API for a tool that lets clients refine design ideas from text descriptions or images, while photo startup Mixtiles brings it to a flow of artwork creation for its users.
Not much changes in terms of policy with the launch of the API, which may disappoint those who fear that generative AI systems like DALL-E 2 will be launched without sufficient consideration of the ethical and legal issues they face. pose. As before, users are bound by the OpenAI Terms of Service, which prohibits using DALL-E 2 to generate overtly violent, sexual, or hateful content. OpenAI also continues to prevent users from uploading photos of people without their consent or images they don’t have the rights to, using a mix of automated and human monitoring systems to enforce this.
A slight tweak is that images generated with the API will not be required to contain a watermark. OpenAI introduced the watermark during the DALL-E 2 beta as a way to indicate which images came from the system, but opted to make it optional with the launch of the API.
“We encourage developers to disclose that images are AI-generated, but do not require that they include the DALL-E 2 signature,” said Luke Miller, product manager at OpenAI overseeing DALL-E development. E 2, to TechCrunch via email.
OpenAI also uses prompt-level and image-level filters with DALL-E 2, although the filters that some customers have complained about are overzealous and inaccurate. And the company has focused some of its research efforts on diversifying the types of images generated by DALL-E 2, with the aim of combating the biases that text-to-image AI systems are known to fall victim to ( for example, mostly generating images of white men when asked for text like “examples of CEOs”).
But these measures have not appeased all criticism. In August, Getty Images banned the downloading and sale of artwork generated using DALL-E 2 and other similar tools, following similar rulings by sites including Newgrounds, PurplePort and FurAffinity. Getty Images CEO Craig Peters told The Verge the ban was prompted by concerns about “unresolved legal issues” as training datasets for systems such as DALL-E 2 contain copyrighted images retrieved from the web.
Trying to find common ground, Getty Images’ rival Shutterstock recently announced that it would start using DALL-E 2 to generate content, but simultaneously launch a “contribution fund” to reimburse creators when the company sells work to train text-to-image AI systems. . It also prohibits AI art uploaded by third parties to minimize the risk of copyrighted works ending up on the platform.
Technologists Mat Dryhurst and Holly Herndon are leading an effort called Source+ to allow people to ban the use of their work or likeness for AI training purposes. But it’s voluntary. OpenAI hasn’t said whether it will participate — or even whether it will ever introduce a self-service tool to allow rights holders to exclude their work from training or content generation.
In an interview, Miller revealed few details regarding the new mitigations except that OpenAI has improved its techniques to prevent the system from generating biased, toxic, and otherwise offensive content that customers might find objectionable. He described the Open API beta as an “iterative” process, which will involve working with “users and artists” over the coming months as OpenAI evolves the infrastructure powering DALL-E 2.
Certainly, if the DALL-E 2 beta is any indication, the API program will evolve over time. At first, OpenAI disabled the ability to change people’s faces with DALL-E 2, but later enabled this ability after making improvements to its security system.
“We’ve done a lot of work on that side of things – both through the images you upload and the prompts you send to aligning with our content policy and cooking in different attenuations to filter at the level from the prompt and at the image level to ensure it complies with our content policy, so for example if someone uploads an image that contains hateful or gory symbols – as very , very, very violent — that would be rejected,” Miller said. “We’re still thinking about how we can make the system better.
But while OpenAI seems keen to avoid the controversy surrounding Stable Diffusion, the open source equivalent of DALL-E 2 that has been used to create porn, gore and celebrity deepfakes, it leaves API users the care of choosing exactly how and where to deploy its technology. Some, like Microsoft, will no doubt take a measured approach, slowly rolling out DALL-E 2 powered products to garner feedback. Others will dive in headfirst, embracing both the technology and the ethical dilemmas that come with it.
If there’s one thing for certain, it’s that there’s pent-up demand for generative AI – the consequences are doomed. Even before the API was officially released, developers were releasing workarounds to integrate DALL-E 2 into apps, services, websites, and even video games. With the launch of the public beta, fueled by the formidable marketing power of OpenAI, synthetic images are on the verge of truly entering the mainstream.
#build #applications #DALLE #generate #images