Stable Diffusion XL: An Overview of New Features and Access
Written on
Chapter 1: Introduction to Stable Diffusion XL
Recently, Emad Mostaque, the Founder and CEO of Stability AI, shared on Twitter that public testing for Stable Diffusion XL has commenced. Here are the key points to note:
- "XL" is a temporary label for the new AI model; a formal name will be disclosed later.
- Image quality has seen notable enhancements compared to previous iterations.
- The speed of image generation has significantly improved over earlier versions.
Let’s delve deeper into the distinctions between the new and previous AI models.
Section 1.1: Examples of Image Generation
Prompt: Luxury sports car with aerodynamic curves, captured in high contrast and high-key lighting with shallow depth of field, exotic and detailed.
Prompt: Bowl of steaming ramen with a perfectly cooked egg at the center, garnished with thin meat slices, green onions, and nori, all immersed in flavorful broth.
Prompt: Serene waterfall nestled in a lush jungle, surrounded by exotic flora and fauna, showcasing a tranquil tropical landscape.
While some images appear more refined, I feel that the improvements aren’t groundbreaking. The new model produces visually appealing results, yet further adjustments might be necessary for truly outstanding outputs. Speed enhancements are noticeable, but they may not be as substantial as anticipated. What are your thoughts on Stable Diffusion XL?
Section 1.2: How to Access Stable Diffusion XL
Stable Diffusion XL (beta) can be accessed via the new Dreamstudio website.
You can also utilize the model on the legacy website.
Section 1.3: Pricing Structure
Charges are based on the number of generation credits utilized. The pricing model is as follows:
- $10 for 1000 generation credits.
Depending on your image settings, credit usage may vary. Keeping the settings at their default will yield about 5000 images for 1000 credits.
Chapter 2: Final Thoughts
Although I’m not overly impressed by the initial results from this new preview model, competitors like MidJourney, Leonardo AI, and Microsoft Image Generator still deliver superior outputs. Despite this, Stable Diffusion XL represents progress, and there is ample opportunity for enhancement. I’m eager to see what final features Stability AI will unveil in their official announcement, and I anticipate the finished product will surpass the current preview.
Stay informed on the latest developments in the creative AI realm by following the Generative AI publication. If you appreciate my work on Medium, consider becoming a member through my referral link for unlimited access. Wishing you a wonderful day!
In this video titled "Stable Diffusion XL Is Here!", you can explore the exciting new features and improvements introduced with Stable Diffusion XL.
The second video, "Stable Diffusion XL 0.9 IS HERE! Better than Midjourney?", dives deeper into comparisons and user experiences with the latest updates.