Dreamstime delivers 320 million images instantly with Fastly

The challenge

Dreamstime CEO and co-founder Serban Enache built a stock photography marketplace twenty-five years ago on a disruptive model: empower photographers to sell their work directly to designers. What started as an additional option for around 1,000 contributors of the stock photography industry in 2000 disrupted the model and ushered in 1.3 million contributors on Dreamstime's platform alone. As of November 2025, the company now manages approximately 320 million files, with billions of variations across different formats, serving a global community of 57 million registered designers.

As digital cameras evolved and mobile photography exploded, Dreamstime's library grew exponentially. The team faced a critical infrastructure challenge: their previous CDN provider couldn't support the new technical requirements needed to deliver this massive collection efficiently. Dreamstime needed intelligent, real-time image manipulation at the edge - something that simply wasn't possible with their existing setup. With an ever-growing library ingesting over 10 million new files monthly, pre-generating every possible size, crop, or format became expensive and inefficient.

The solution

Dreamstime evaluated CDN providers based on specific technical criteria that would support their vision for dynamic content delivery. "We wanted to have different image formats like WebP," said Elian Ionescu, Digital Experience Director at Dreamstime. "We wanted everything to be dynamically generated from the browser. This would allow us to have only one image uploaded on Fastly, which is generated in different sizes for its various placements across the website." The goal was clear: reduce storage costs while improving performance across every device.

Fastly met these requirements where other providers fell short. The platform's ability to support real-time image resizing and modern formats like WebP was exactly what Dreamstime needed. "With Fastly, we can generate optimized image sizes based on specific display criteria," Enache explained. We are now capable of delivering a perfectly tailored visual asset in real time that automatically adapts to the user's context and eliminates backend friction. This level of agility is the new standard for the modern web. It's about ensuring every user, on every device, receives the best possible digital experience."

Cutting mobile load times in half with dynamic optimization

For Dreamstime, speed isn't just about user satisfaction; it directly impacts search engine rankings. Core Web Vitals, particularly Largest Contentful Paint (LCP), measure how quickly the primary image loads on screen. After migrating to Fastly, Dreamstime saw dramatic improvements. "Our average mobile LCP dropped from around 4 seconds to just 2 seconds," Ionescu said. "This 50% reduction makes our mobile users perceive the site as loading nearly twice as fast, which is vital for lowering bounce rates."

The transformation came from Fastly's ability to generate WebP formats on the fly from source files. WebP images are 25–40% smaller than standard JPGs, leading to faster load times and lower bandwidth costs. "WebP is the preferred image format for Google SEO, but its general adoption is quite recent," Ionescu noted. "With Fastly's ability to generate it on the fly, we now support WebP for all our images." This capability gave Dreamstime a competitive edge overnight, without requiring massive backend processing of their entire library.

Protecting content with real-time watermarking

Dreamstime collaborates with numerous partners to distribute its +320 million files globally. These partners serve different markets with unique cultural needs, but all require content protection to prevent fraudulent downloads or AI scraping. Before Fastly, watermarking was rigid and limited. "Our previous solution lacked real-time resizing and watermarking, forcing us to manually regenerate all images whenever changes were needed," Enache said. "This made us hesitant to make adjustments, and our partners weren’t able to access our content as a white label service."

Fastly changed this completely. "We can now instantly watermark our content with partner logos, fully integrating it into their sites at no extra cost, since it is still delivered from our servers," Enache explained. Dynamic cropping, which was nearly impossible before, now works seamlessly. "Since the watermark was added during image generation, it was impossible to crop a watermarked image dynamically without also cropping part of the watermark. With Fastly, we no longer face these issues because of its real-time capabilities."

Simplifying operations and cutting storage costs

Fastly's on-the-fly image optimization fundamentally changed Dreamstime's storage strategy. "Instead of needing to pre-generate and store countless variations of each image in our massive library, we now only need to store the high-resolution original," Enache said. "Fastly dynamically creates the specific size and format needed at the moment of request." This generate-on-demand model simplified their origin architecture, reduced their storage footprint, and cut the operational costs of managing billions of static files.

The operational savings extended beyond storage. Tasks like applying partner-specific watermarks or generating new image sizes were once resource-intensive engineering projects. "By automating these functions, Fastly has freed our technical teams from managing static image assets," Enache noted. The team can now run rapid A/B tests and deploy design changes instantly. "We can configure any new crop or overlay at the edge and deploy it immediately," Enache said. "This ability to test rapidly without complex backend changes allows us to continuously optimize our platform for faster conversions in a way that was simply not possible before."

Key takeaway

Dreamstime's migration to Fastly was remarkably smooth, with minimal disruption to operations. "The platform's ease of use and straightforward integration process had an immediate positive impact on our operations team," Enache said. With the infrastructure now in place, Dreamstime is prepared for continued growth, particularly as AI-generated content becomes increasingly significant. "As AI-generated content becomes more popular for our contributors, it adds another layer of content to our database. We expect that Fastly's scalability will help us maintain our top performance metrics," Ionescu noted. For Dreamstime, Fastly isn't just about handling today's massive library - it's about staying ready for whatever comes next.