Choose one patch you can track across the editor and the browser, such as the red glaze on a mug. Compare the working image with the actual file requested by the browser. A storefront thumbnail may be a separate export with different color information.
Start with a tagged source whose working profile you know. If two browsers show different colors on the same physical display, verify that both loaded the same URL and asset variant. A mobile crop or a CDN conversion can return a different file.
Compare assignment and conversion
Adobe's Photoshop color profile instructions describe the two operations. Assign Profile changes how the editor interprets the existing RGB values. Convert to Profile calculates new values for a destination space with the aim of preserving appearance where the destination can represent the colors. Some saturated colors may fall outside the destination's range. Compare those regions by eye after converting.
Make three copies from your tagged source. Keep the original as your reference outside this exercise.
| Copy | Operation | What the comparison tells you |
|---|---|---|
| Assigned | Assign sRGB to the original RGB values | Shows the effect of changing the interpretation without changing the numbers |
| Converted and tagged | Convert to sRGB and embed its profile in the export | Gives the browser a file with explicit color information |
| Converted and untagged | Export the sRGB conversion without a color tag for this comparison | Shows whether the delivery path's fallback agrees with the tagged copy |
Use the assigned copy to see the effect of reinterpretation. Expect a small difference if the original was already near sRGB. A more saturated product on a clearly tagged source makes the effect easier to see. The CSS Color specification describes sRGB as the fallback for untagged web images unless otherwise specified.
Check the exact file that arrived
Open the three images locally on the same display at the same size. Keep the display mode and lighting fixed. Check one saturated region, one neutral highlight and the boundaries of a dark shadow. Then load the delivered image from your test storefront and download it from the same request that the browser shows. Check that it still carries the intended profile or has been converted to the color space your delivery path expects.
For a standard web catalog, a correctly converted sRGB export is a practical baseline. Keep the working master with its known color profile so you can make another output later. Run the same export on two neighboring products and compare their neutral backgrounds side by side. After the files agree on one display, check the target devices for remaining differences in appearance.
Identify which copy a customer sees
Start by inspecting the actual thumbnail URL in the browser. A product page can load a small image in the grid and a different file in its full-size viewer. Save each from the test store and compare dimensions and color information with the source export. On a second browser, make sure you opened the same photo, crop and size variant. An image looked at in an editor may use different scaling or a monitor profile, so begin with side-by-side comparisons on one display.
Use the same neutral page background while comparing the proposed assets. A warm card background can change how a pale mug seems to look even if the image file is identical. Temporarily use a default stylesheet in a test page and inspect the same region on both sides. If the difference disappears with that control, examine the presentation before changing the image profile. Keep display modes such as shifting warmth or automatic brightness stable during the trial.
For the three-copy experiment, start from a tagged working file with a known profile and choose a red or green patch that has a noticeably saturated example. Save a color-neutral part of the same photograph as a second comparison. Reopen all copies in your editor and then in the browser. Try two settings when a CDN generates a scale variant, checking whether it preserved, stripped, or converted the color information. A missing tag after processing does not tell you by itself whether the underlying numbers were properly converted.
Decide where the change entered the pipeline
| Observation on a single display | Check next |
|---|---|
| The assigned copy differs but the converted copy looks close | The same RGB values were interpreted in a different space. |
| The converted tagged file agrees locally and the CDN version differs | Inspect the stored asset and the transformation used to serve it. |
| Tagged and untagged converted copies agree, but browsers load different URLs | Compare the actual variants before changing profiles. |
| All files agree on this display and vary on other devices | Review display modes and the target environment. |
A converted color can still look different if the destination has a smaller gamut. Inspect highly saturated colors and fine texture after conversion. If adjustment is necessary for the catalog, make it on a separate delivery version and keep the master. Save a proof showing both a product color that matters and some adjacent neutral detail. It helps another editor confirm that the adjustment affected the intended region.
Check JPEG and WebP paths separately if the application generates both. The final image can vary by format or pipeline configuration, and a test on the uploaded JPEG establishes little about a small WebP generated after upload. When the source is a screenshot rather than a photo, use the screenshot encoding comparison to check edges and readability after the same color conversion.
Give the next run a repeatable reference
For a batch, record the source profile, chosen destination, format and intended served image variant. Keep one representative product with a clearly recognizable reference patch and one high-saturation example. Repeat the proof after changing an export preset or CDN rule. If your service has a separate product card crop, check a sample of that actual crop on the same device.
Set expectations for the environment you can control. You can compare source exports and network-served files on a known display. Different displays, brightness and viewing conditions can still affect the buyer's impression. Keep the product identifiers and exact versions in the proof so a customer-support report about a photo can be traced to the appropriate assets.


