Dolly 3: OpenAI's New Realistic Image Generator
Table of Contents
- What is Dolly 3?
- Comparing Dolly 2 and Dolly 3 Images
- When Will Dolly 3 Be Fully Available?
- Trying Out Dolly 3 Yourself
- Conclusion
What is Dolly 3? An Overview of OpenAI's Newest Image Generator
Dolly is an image generator created by OpenAI, the makers of ChatGPT. The original Dolly model was released in January 2021, followed by an upgraded Dolly 2 in April 2022. Now OpenAI has unveiled Dolly 3, the latest iteration of this AI art tool.
In this blog post, we'll take a close look at Dolly 3 - what it is, how it compares to previous versions, when it will be fully available, and how you can try it out yourself.
Dolly's History: The Previous Image Models
As mentioned above, Dolly's first image model launched in early 2021. However, it failed to gain much popularity at the time, as other AI art generators like Midjourney and Stable Diffusion quickly surpassed it in both quality and accessibility. Dolly 2 arrived in April 2022. While it showed some improvement in image generation capabilities, it still wasn't widely adopted. So now OpenAI is hoping this third iteration - Dolly 3 - will finally make Dolly a top contender in AI-generated art.
Accessing Dolly 3 Now Through Microsoft Co-Pilot
At the moment, Dolly 3 is only accessible through the new Microsoft Co-Pilot. Specifically, you can generate Dolly 3 images via the Microsoft Bing Image Creator. This early access was granted when Co-Pilot launched on September 26, 2022. So even though Dolly 3 isn't yet available directly through OpenAI, this integration with Bing allows us a sneak preview.
Comparing Dolly 2 and Dolly 3 Images Side-by-Side
To see the improvements Dolly 3 provides, let's compare some images generated by Dolly 2 vs Dolly 3. We'll use the same prompt for both - "a picnic basket with sandwiches, fruit, cheese and wine on a checkered blanket" - and examine how they differ in quality and realism.
Dolly 2 Picnic Image
Using the picnic prompt, Dolly 2 produces an image that, while containing all the requested picnic elements, lacks realism. The lighting seems off, and details are blurry rather than sharp. The people, food and blanket don't look natural or photograph-quality.
Dolly 3 Picnic Image
With the exact same prompt, Dolly 3 generates an image that looks much more realistic. Details are clearer, lighting is improved, and the overall quality mimics a high-res photo. Looking closely, you can still tell it's AI-generated. But the leap forward from Dolly 2 is massive.
Side-by-Side Comparison
Placing images from Dolly 2 and Dolly 3 next to each other highlights the differences. While Dolly 2 feels obviously computer-generated, Dolly 3 could almost pass as a real photo. Imperfections exist, but you have to inspect closely to spot them. It's a big upgrade in terms of photorealism and overall quality.
When Will Dolly 3 Be Fully Available?
At the moment, Dolly 3 access is limited to testing through Microsoft Co-Pilot and Bing. OpenAI has stated that Dolly 3 will roll out more widely "in the coming weeks", first through the OpenAI API.
After that, we expect Dolly 3 integration directly into ChatGPT, allowing users to generate images simply by describing them in text conversations. This will provide an incredibly intuitive interface for AI art creation through Dolly 3.
Trying Out Dolly 3 Yourself
If you want to experiment with Dolly 3 image generation right now, here are step-by-step instructions:
-
Go to Bing.com and make sure you're logged into your Microsoft account
-
Click on 'Bing Image Creator' on the homepage sidebar
-
Type or paste an image description prompt into the text box
-
Click 'Create' and wait for your Dolly 3 image to generate
Just keep in mind that wait times to access Dolly 3 are still quite long due to demand. So you may need multiple attempts and some patience.
Conclusion
Dolly 3 represents a major leap forward in AI-generated image quality from OpenAI. Early testing shows huge improvements in realism and detail over previous Dolly versions.
While full access still lies ahead in the coming weeks, Microsoft Co-Pilot integration offers an exciting sneak peek at what Dolly 3 can create. As this model rolls out more widely, it has the potential to challenge or even surpass other top contenders in AI art.
We'll be sure to provide updates and hands-on testing as Dolly 3 access expands. So subscribe to our newsletter using the form below to stay up-to-date on the latest with this powerful new AI image generator.
FAQ
Q: What is Dolly 3?
A: Dolly 3 is the latest image generation model from OpenAI that produces more realistic images than previous versions.
Q: How can I access Dolly 3 now?
A: You can currently access Dolly 3 through the Microsoft Bing image creator before the official launch on OpenAI.
Q: Is Dolly 3 better than Dolly 2?
A: Yes, Dolly 3 generates noticeably more realistic images compared to Dolly 2.
Q: When will OpenAI release Dolly 3?
A: Dolly 3 is expected to officially launch on the OpenAI platform in the next 1-2 weeks.
Q: Can I try Dolly 3 myself?
A: Yes, you can access Dolly 3 now through Bing's image creator, but beware of long wait times due to high demand.
Q: Will Dolly 3 be available in ChatGPT?
A: OpenAI plans to integrate Dolly 3 into ChatGPT after the initial launch, likely within a few weeks.
Q: Is a Microsoft account needed for Dolly 3?
A: Yes, accessing Dolly 3 via Bing currently requires a Microsoft account.
Q: Why does Dolly 3 take so long right now?
A: Dolly 3 is experiencing very high demand after launching on Bing, so you may encounter long wait times when generating images.
Q: What other AI image models compete with Dolly 3?
A: Top competitors are Midjourney, Stable Diffusion, and DALL-E 2 which also create highly realistic AI art.
Q: Where can I learn more about AI image generation?
A: Check out the Everyday AI newsletter and podcast for the latest on using AI tools like Dolly 3.