Your client sends you a folder of photos. You open it. Immediately you think: oh no. Soft edges. Pixelated corners. That weird grainy texture that screams "compressed three times over." You know these images are going to look awful on the website, but explaining why to a non-technical client? That's a conversation that takes 20 minutes and ends with "can't you just make them bigger?" The truth is—they can be made bigger. Not by stretching them (that's a disaster), but by using AI upscaling. And if you don't know about this yet, your clients probably do. So let's talk about how to become the hero who actually fixes this problem instead of just complaining about it.
Here's what happens in real life: clients have thousands of images. A mix of phone photos, old scans, screenshots, stuff from their cousin's camera. None of it is perfect. Some of it is actually pretty bad. Your job used to be "we need better photos, reshoot everything." Now? Your job is "let me show you a smarter way."
Phone photos with grain and noise: Shot in poor lighting with auto-settings. By default, phone cameras apply aggressive compression to save space. The moment you need that image larger than the phone screen, it falls apart.
Scanned documents and old prints: Someone literally held their phone over a piece of paper and took a picture. Or they ran it through a flatbed scanner that's been collecting dust since 2015. The resolution is there, but it's muddy. Yellowed. Details are lost.
Screenshots at weird resolutions: A client grabs a screenshot on their laptop, crops it weird, sends it to you. You have 800x400px of something that should be 1600x1200px. It's sharp—but tiny. Stretch it and everything goes soft.
Blurry portraits: Someone was moving. The focus was slightly off. The lighting was dim so the camera boosted ISO and introduced noise. Now you have this image that's almost good, but it's never going to print well or look crisp on a big display.
Low-res stock photos: Your client grabbed something cheap from an old stock site. Back then, 1200x800px was "good enough." It's not. Not anymore.
Here's the thing nobody tells developers—most of these problems can be fixed. Not perfectly. Not magically. But well enough that the image goes from "ugh, no" to "actually, that works." And it takes minutes. Not hours.
This is critical. You don't want to spend time on every image. Some of them should just be reshot. So here's how you assess:
Can you see the original detail? Look closely. Is there texture, fine lines, definition? Even if it's soft, is there information there? If yes—you can probably upscale it. If the image is blurry, the person moved, the focus was wrong—upscaling just makes it sharper blurry. That's not useful.
Is the color accurate? Faded? Weird color cast? That's separate from sharpness. You can fix that after upscaling.
What's the source quality? A 600x400px photo on a phone? You can take it to 1200x800px (2x). A 2000x1500px image from a DSLR? You can go 4x safely. A 400x300px screenshot? Honestly, that one's probably a reshoot situation.
How critical is this image? If it's a hero image on the homepage, above the fold, high resolution—reshoot. If it's a supporting image, interior page, smaller display—upscaling saves money.
The rule of thumb I use: If you can't tell it was upscaled by looking at it, it worked. If it looks plastic, oversharpened, or weirdly artificial—you either picked the wrong tool or the source was too damaged.
Let me walk you through how I actually handle this when a client hands me a folder of 50 photos that "should look better on the website."
Step 1: Get the originals in their best form
Call the client. "Send me the biggest files you have. Original exports if they exist. We need to work with what has the most information." This is crucial. Working from a JPEG that's already been compressed once is way harder than working from a PNG or RAW file.
Step 2: Triage the photos
Spend 10 minutes sorting them. This folder goes "upscale this," that folder goes "reshoot," and this one goes "maybe." You'll find about 70% are salvageable, 20% will look great after upscaling, and 10% should just be reshot.
Step 3: Identify what you're working with
Step 4: Pick your tool for the job
Step 5: Process the batch
If you're using Upscale Media: drag and drop a folder, set to 2x or 4x, hit go. Come back in five minutes.
If you're using Topaz: point it at the folder, select the appropriate AI model for the image type, pick your upscaling level (I usually do 2x for web, 4x for print), and run it.
Either way—you're not manually editing each image. That's the whole point.
Step 6: Optimize for web
Upscaling gives you a bigger file. Now you compress it properly. Use WebP format. Use Tinypng or ImageOptim or Squoosh. You want the sharpness but you don't want a 5MB photo because you upscaled it.
The flow is: Original → Upscaled → Compressed for web. Not: Original → Stretched → Crushed by poor compression.
Here's what nobody talks about: just because you can go 4x doesn't mean you should.
Use 2x when:
Use 4x when:
Don't use beyond 4x for web. Seriously. You're chasing diminishing returns. The file gets huge, the processing takes forever, and honestly? Your website visitors can't see the difference on a screen.
This is the part where you show them the before and after and explain what just happened. Here's how I frame it:
"Your original photos were compressed when they were taken and sent to you. We ran them through an AI upscaler that analyzed the image, predicted what details should be there based on patterns it learned from millions of photos, and reconstructed a sharper version. It's not magic—the original information still limits what's possible—but it's enough to make your images look professional on your website."
Translation: It worked, they look better, you saved them money by not paying for a reshoot.
If they ask "why couldn't we just stretch the original?" you say: "Because stretching just makes blurry bigger. This actually adds detail back." Show them a close-up comparison. They'll get it.
Here's what this means for your business: A client has 30 mediocre photos. Old option: "We need to reshoot. Budget: $2,000, timeline: two weeks." New option: "Let me run these through upscaling. Budget: $0-50, timeline: two hours." They pick the second one. You look like a genius. Their website launches on time.
You're not replacing photographers. You're solving a real problem that 90% of small businesses face. They have the content. It's just not formatted right. You fix it in batch. That's a service worth offering.
When you actually do this for clients—when you show them upscaled images—they often ask: "Can you do this for the other 200 photos in our library?" Suddenly you're not just optimizing their website. You're upgrading their entire digital archive. Batch upscaling becomes a service, not just a one-off fix.
That's where tools like Upscayl or Topaz batch processing become worth their weight. You're not upscaling 30 photos. You're upscaling hundreds. Running it overnight, waking up to perfectly processed images.
If you're recommending AI upscaling to clients who want to do it themselves? Tell them to try upscale image with AI tools like Betterimage first. No software to install. No account needed (well, usually). Just upload, pick 2x or 4x, download. It's the fastest way to see if upscaling even helps their specific images.
Sometimes after they try it, they realize they do need to reshoot. Sometimes they see the difference and want you to upscale their entire library professionally. Either way, you've given them a fast, free way to test the concept.
Your job as a developer isn't just to build websites. It's to solve problems. And blurry images are a real problem. Not a "nice to have" problem. A "client is upset because their products look bad online" problem.
AI upscaling doesn't fix everything. It won't rescue a completely out-of-focus disaster. But it solves 70% of the cases you'll actually encounter. And it turns a two-week reshoot timeline into a two-hour batch processing job.
Learn the tools. Run a few test batches. Show your clients what's possible. They'll thank you. And more importantly? They'll hire you again because you solve problems instead of just identifying them.