Sourced Low-Res Product Photos? How to Fix Blurry Images for Dropshipping

The first batch of supplier photos I received for my dropshipping store was 47 images. They averaged about 400 pixels wide. Some of them looked like they had been screenshotted from a 2007 flip phone. One was a photo of a physical photo print, complete with glare from the overhead light. I remember staring at it and thinking there was no way I could put this on a product page and expect anyone to buy.

Every dropshipper and marketplace seller deals with this. Suppliers send you the smallest, most compressed versions of their product images. They do it because it is faster to send and they assume you will figure it out. After fixing over 300 supplier photos for various stores, here is what actually works and what is a waste of time.

Why supplier photos are always terrible

It is not malice. Most suppliers, especially overseas manufacturers and wholesalers, are not photographers. They take product photos with whatever phone they have, in whatever lighting is available, and they send them through WeChat or WhatsApp which compresses everything to a fraction of the original size. A photo that started at 3000 pixels wide on their phone arrives on your computer at 800 pixels, full of compression artifacts, looking like it was rescued from an old hard drive.

The messaging app compression is the biggest culprit. WhatsApp compresses images by default. WeChat does the same. Even email sometimes downsizes attachments depending on the provider. If you are getting photos that are suspiciously small and blocky, the compression happened after the photo was taken, not during. The supplier probably has a better version. You just need to ask for it the right way.

Step one: get the best possible source file

Before you do any AI processing, ask the supplier for the original photo file. Not a screenshot. Not something forwarded through chat. The original file from their phone or camera. Use the phrase "original file, not compressed" and ask them to send it through Google Drive or Dropbox or as a direct email attachment at original size.

About half the time, the supplier has a better version and will send it if you ask specifically. A 3000 pixel original that gets compressed to 800 pixels is still 3000 pixels on their device. You just need them to move the file without the compression step. This alone saved me from having to fix about 30% of the photos I was receiving.

When the supplier does not have a better version, or when you are working with what you have because asking is not an option, that is when AI processing becomes necessary.

What AI upscaling does to compressed product photos

I tested this systematically. I took 50 supplier photos that were too small or too compressed and ran them through AI upscaling at 2x. The results broke down into three groups.

Group one, about 25 photos. These were images that were just too small but otherwise clean. The original was maybe 600 to 800 pixels wide with no visible compression artifacts. AI upscaling to 2x produced 1200 to 1600 pixel versions that looked genuinely good. Sharp enough for a product page. The zoom function on these worked correctly. These were the easiest 25 wins of my life.

Group two, about 18 photos. These had visible compression artifacts. Blocky areas in smooth gradients. Jagged edges where there should have been smooth curves. Ringing artifacts around text or logos. The AI upscaler improved the overall sharpness but the compression artifacts were still there, just at a higher resolution. These photos needed an extra step. I ran a light noise reduction pass before upscaling, which smoothed the compression artifacts enough that the upscaler could work with clean data.

Group three, about 7 photos. These were the disasters. Photos of photos. Images so blurry you could not tell what the product was even at the original size. The AI upscaler tried its best but the results looked like an AI trying to hallucinate product details from noise. These photos could not be saved. For these, I had to either find the product on the manufacturer's website and download their version, or order a sample and photograph it myself.

The noise reduction before upscaling trick

This was the biggest insight from processing all these supplier photos. If the image has visible compression artifacts, upscaling first makes them worse. The AI sees the blocky compression patterns as image detail and tries to enhance them. You end up with a higher resolution version of the artifacts.

The better order is to reduce noise and compression artifacts first, then upscale. Most image editors have a basic noise reduction filter. You do not need anything fancy. Just enough to smooth out the worst of the JPEG blocking without blurring actual product details. Then feed the cleaned image into the AI upscaler.

On ClarifyPix, you would run a light background removal or restoration pass first to clean up artifacts, then upscale. Each step costs 1 to 4 credits depending on the operation. For a photo that needs both cleanup and upscaling, total cost is around 6 credits. That is about what it costs to process a single image from terrible to usable.

When to give up and re-shoot

There is a point where AI processing costs more in time and credits than just ordering a sample and taking your own photo. If a product photo needs more than two processing passes to become usable, or if the original is below 300 pixels wide, you are better off getting a new source image.

Ordering a sample costs money and takes time. But a single clean product photo that you own, at full resolution, with proper lighting and a white background, is worth more than twenty AI-restored supplier photos. For your best selling products, invest in original photography. Use AI restoration for the long tail of products that do not justify the cost of samples and photo shoots.

I learned this the expensive way. I spent about 200 credits restoring terrible supplier photos for a product line that sold maybe three units total. The photos were technically better after AI processing, but the products were just not popular. Now I only invest restoration effort in products that already show sales potential. Everything else gets one pass of basic cleanup and that is it.

The simple batch workflow for supplier photos

After processing hundreds of these photos, my workflow is down to three steps. It handles about 80% of supplier images and takes roughly 10 seconds per photo.

Step one. Open the photo and check the resolution. If it is above 1200 pixels on the long side and looks clean, use it as is. Do not process what does not need processing.

Step two. If it is below 1000 pixels or has visible compression artifacts, run a cleanup pass first if there are artifacts, then an AI upscale to 2x. On ClarifyPix, the upscale is 2 credits per image and takes about five seconds.

Step three. After processing, open the result at 100% zoom and check one thing. Can you clearly see the product details? Thread texture on fabric. Button details. Surface finish on metal. If yes, the photo is good enough for your store. If no, and the original was terrible, this might be one of the unsalvageable ones. Order a sample or find a better source.

Supplier photos will never be as good as ones you shoot yourself. But with a quick AI cleanup and upscale, most of them go from embarrassing to perfectly usable. And for a dropshipping store with hundreds of SKUs, perfectly usable is exactly what you need.