A competitor took product images from our website and put their own watermark on them—if we only have compressed files left, can we still prove who took the photos?

Whether you can prove ownership when a competitor copies product images from your website and adds their own watermark depends on whether the original photo files, design source files, and authorization records were stored separately at the moment of delivery; the watermark on the website only raises the cost of copying and is not proof of ownership by itself. Based on common project delivery experience in 2026, keeping records beforehand is much easier than arguing afterward. The key question is usually not how large the watermark is, but whether the evidence chain broke at the step where 'only compressed images are left.'

What a watermark can and cannot do

The direct effect of a watermark is to add one more step and some psychological cost to low-threshold copying such as right-click save, and to leave an identifiable mark after an image is reposted. It does not stop screenshots, cropping, recoloring, or the watermark-removal tools and AI retouching common in 2026. Treating a watermark as technical anti-theft will set the wrong expectation; treating it as a way to raise copying cost and leave a trace is closer to reality.

For a website, the impact of copied images is not only about copyright: when a competitor uses your product images in low-price comparisons, customers who see the same image elsewhere may doubt your qualifications. A watermark can help you explain after discovery that the image has long been on your own website, but it cannot replace original files, first-publication time records, and authorization documents. Without original files and authorization records, a watermark itself can hardly prove ownership on its own.

• Can stop some cases: casual saving by ordinary visitors, and direct image grabbing by some scraping tools.
• Cannot stop: screenshots, cropping out the watermark area, redrawing and recoloring, and processing with watermark-removal tools.
• Real value: reducing willingness to copy, leaving clues, and supporting explanations together with retained records.

If only compressed images are left, filling the gap later is too late

What is usually displayed on a website page is a compressed presentation file: dimensions are reduced, information is processed, and there are not enough details to compare point by point with the original photo. After a competitor copies the image, adds their own watermark, and recolors it again, both sides actually hold 'a product image.' At that point, when you go back to look for evidence, the only things you can often rely on are first-publication screenshots and original files—and these are exactly the two things most easily missed at the delivery stage.

There are three common ways this gets missed: the photographer delivers only finished JPGs and keeps the originals in the studio; the design file is flattened and only a PNG is exported; after launch, the original image sits on a colleague's computer and disappears when the person changes. At delivery time, each of these is a one-sentence matter; by the time you need to prove ownership, they cannot be recovered.

What to retain at delivery: do three things in order

1. Get the source files first: original photo files, editable design source files, and uncompressed images before export—at least one of these three should be complete and of sufficient clarity; if the other party only provides finished images, state this in the delivery note.
2. Then build an original-image pool: store compressed images used on the site, watermarked images, and original images in separate directories, with dates and product lines in directory names; put authorization records (contract terms, photography commission forms, stock image authorization screenshots) in the same directory, not scattered across chat logs.
3. Finally set the watermark level: for main product images and list thumbnails, use a light corner mark covering about 8%–15% of image width at 10%–30% opacity (experience range); for case site photos, a light central watermark can be used; full-screen overlaid text should only be used for external sample images and low-resolution previews, and website detail large images generally should not be covered.

Do not reverse the order. If you decide the watermark style first and fill in materials later, it is easy for the whole site's images to follow a designer's preference rather than their purpose.

On the delivery floor: how one image revision round happens

One project had these constraints: the client had saved a batch of product images from WeChat, the resolution was limited, they asked for obvious watermarks across the whole site, and launch was pressed within one week. Following common delivery practice, the images were first grouped by purpose: main images and list images got light corner marks, detail large images kept an unwatermarked version for the time being, and the client was asked for original files to archive separately. After launch, the client felt the detail page 'could not see the watermark' and asked for uniform bolding, which led to one revision round; low-resolution images became blurrier after enlargement and text overlay, and in the end the materials had to be replaced anyway. The cost was one extra revision round—for dozens of images across a site, revisions commonly take from half a working day to several working days (experience range), and redoing materials requires another shoot or searching for original files.

Most such rework can be avoided before work starts: first confirm image purpose and how visitors will view them, then confirm whether material resolution is sufficient, and only then decide the watermark style. A watermark is not an isolated design decision; it is tied to material quality and page purpose.

Applicability and boundaries

Watermarks are suitable for raising copying cost and leaving clues, not for technical anti-theft; detail large images that need zooming in to see details usually should not be fully covered with watermarks. Once the boundary is clear, compare against the following.

• Suitable for: product images have indeed been copied by competitors, images are core assets, sample images need to be sent externally, case images involve partners, or website images are being batch-scraped by collection sites.
• Not necessary: images come from public stock libraries with clear authorization, visitors are mainly existing customers, the page mainly communicates through specification tables, or the original material resolution itself is low.
• Use with caution: detail pages need zooming in to see materials and interfaces, mobile small images account for a high proportion, or the page layout is already crowded, so heavy watermarks would further squeeze readable areas.

Comparison of three approaches: full overlay, corner mark, and no watermark plus retained originals

• Full-image text overlay: revision cost is medium to high; for dozens of images it commonly takes from half a working day to two or three working days (experience range); the anti-copying impression is obvious, but it has a large impact on detail viewing, and is more suitable for external sample images and low-resolution previews.
• Corner mark: revision time per image is on the order of a few minutes; for dozens of images across a site it commonly takes within half a day (experience range); it has little impact on conversion and limited anti-copying effect, but it supports traceability, making it suitable for main product images and list images.
• No watermark plus retained originals: revision cost is low and page appearance is good; proof mainly relies on original files, first-publication time, and authorization records; after copying is found, use platform complaints. Suitable for high-definition material images and detail images that need zooming in.

In actual delivery, combinations are more common than choosing one of three: main images and case images get corner marks, detail large images keep a clean version, and all images go into the same original-image pool.

After finding copied use: what to do first and what to do later

1. Fix the evidence: take complete screenshots of the other party's page, note the URL, access time, and image location, and archive the webpage if possible.
2. Check retained records: find the first image published on your website, original files or source files, publication time, and authorization scope, and confirm who owned the image at the time.
3. File platform complaints: submit through each platform's official infringement complaint entry, fill in materials truthfully, and do not falsify.
4. Assess communication: if the copying scale is small and the other party is willing to take it down, communicate in writing first; consider formal letters for batch copying.
5. Revisit the watermark level afterward: after handling, evaluate whether to adjust watermarks; do not cover the whole site's images just because of one copying incident.

Do not reverse the order: changing your own website images before collecting evidence means you have moved the very things that could be compared.

Frequently asked questions

If we add watermarks to website product images, will Baidu Image Search still index them?

Indexing mainly depends on whether images can be accessed normally and whether there are alt text and surrounding descriptions; overly heavy watermarks may reduce click willingness, but they are generally not the direct reason for not indexing, so the two do not have to be opposed.

If a competitor crops out the watermark and uses the image again, is adding a watermark still meaningful?

Yes. Cropping changes the composition and detail proportions, while the original file plus first-publication time can still support the source; the role of a watermark is to raise copying cost and leave clues, not to stop all copying.

If the original files are still with the photography studio, can we just ask them to send a copy?

It is advisable to confirm the delivery scope and permitted uses in writing, and after receiving the original files, archive them yourself and record the date. Relying only on a chat message saying 'I'll send it later' can easily become unclear if personnel change or the studio changes hands.

If product images are copied, is it faster to just send a lawyer's letter?

Usually, fix the evidence first and then use platform complaints; letters depend on the scale of copying and completeness of evidence. If you only have compressed images, it is more practical to first supplement first-publication records and authorization materials.

If website product images are core assets, do two things first: divide images by purpose into three levels—light mark, light watermark, and full-screen overlay—and archive original files and authorization records separately; when copying is found, collect evidence first and handle it afterward. A watermark is not anti-theft technology, detail large images and high-definition material images do not need full coverage, and when budget and schedule are tight, prioritize main images and case images. Complaint entries are subject to each platform's official instructions.