The Blur Image tool processes one JPG, PNG, or WebP file at a time inside the current browser tab, so "bulk multiple images" with this tool means running the same local workflow on each file in sequence rather than uploading a queue. Decoding, drawing, blurring, and PNG creation are all done in the browser, which keeps every photo on the device that opened it. The export is always a PNG rendered on a separate offscreen canvas at the source image's natural pixel width and height, so the file you download is not a down-scaled copy of what was on screen. Rectangle mode takes X and Y coordinates plus width and height in original-image pixels, which makes a sensitive area repeatable across several photos with the same dimensions. The on-page canvas may be smaller than the original so a large photo remains practical to inspect; the download, however, is built from a separate full-resolution canvas. Each new generation replaces the previous browser download object, and changing files or leaving the page releases the temporary object URL.
What the Tool Does With a Single File
The Blur Image tool accepts a JPEG, PNG, or WebP file from the local device, lets the reader choose whether the blur should cover the full image or a rectangle, sets a radius, and exports a PNG. Everything happens in the same browser tab. There is no upload step, no remote processing queue, no account storage, and no server-side copy of the original or the result. That property matters when the photos in the folder are drafts, screenshots, or anything that should not leave the computer merely to receive a visual effect. The supported inputs are documented at 25 MB per file and 40 megapixels of decoded pixels, so a small collection of mid-resolution phone photos will almost always pass, while a single very high resolution panorama may bump into the megapixel ceiling even when its compressed size looks modest.
For most readers handling more than one file, the practical loop is simple: open the first photo, choose the area and radius, generate the PNG, save it, then move on to the next file. Rectangle coordinates are entered in the original image's pixels, so the same coordinate string can be reused on a second photo only if both photos share the same width and height. For a mixed-size batch, treat each file as its own job and either measure each rectangle from scratch or anchor it relative to a corner the reader can see clearly.
Run the Same Workflow Across Multiple Files
- Pick the first JPG, PNG, or WebP file and confirm the displayed pixel dimensions in the preview before doing anything else.
- Choose Full image, or switch to rectangle mode and enter the X, Y, width, and height in original-image pixels.
- Set the blur radius. Start low and increase only if the visible softening is weaker than intended.
- Choose Blur image, read the reported output dimensions next to the preview, and download the PNG.
- Open the downloaded PNG at 100 percent zoom and check the selection boundary, the output dimensions, and any text, faces, or numbers that motivated the edit.
- Repeat from the first step with the next file, updating coordinates when the source dimensions change.
Each repetition overwrites the previous browser download object, so saving the file before starting the next iteration is the only safe way to keep every output. Changing files or leaving the page releases the temporary object URL, which means a refresh without a download effectively throws the result away. A simple folder convention, keeping originals in one folder untouched and writing PNGs to a separate output folder with matching filenames, turns the sequence into something close to a bulk operation without the tool having to invent a batch queue.
Full-Image Softening vs Rectangle Blur
| Aspect | Full-image blur | Rectangle blur |
|---|---|---|
| Best for | Backgrounds, draft graphics, evenly softened photos | Addresses, faces, account numbers, license plates, message bubbles |
| Coordinates needed | None | X, Y, width, height in original-image pixels |
| Edge behavior | Covers the entire frame | Clamped to the image boundary; pixels outside the rectangle stay as drawn |
| Kernel bleed | Applies to every pixel | Colors near the rectangle boundary can blend into the selected area |
| Repeatability across files | Trivial, the same radius works on every image | Needs identical source dimensions to reuse coordinates verbatim |
For an evenly softened background, full-image mode is usually the simpler path. Rectangle mode is the right choice when only a small part of the image needs softening and the rest of the frame should stay sharp. Readers who want the same edit applied to a folder of screenshots that share identical dimensions can copy coordinates from the first successful run into every following job. Readers who want a guide that goes deeper into both modes can compare notes in How to Add Blur to an Image (Full or Partial).
Inside the Blur Radius Control
The radius slider is the only knob between "barely softened" and "sharply blurred." Browser canvas filters treat the value as a kernel size, so doubling the radius visibly softens much more than doubling feels intuitive. A small radius removes fine noise and slightly hides paper grain, while a much larger radius turns small text into unreadable color blocks and pushes faces into abstract shapes. A workable habit is to start with a small radius, generate the PNG, open it at 100 percent zoom, and only increase the radius if the area the reader wanted hidden is still readable. Because the canvas blur samples neighboring pixels, shapes and colors close to the rectangle boundary can blend into the selected area; that is normal blur behavior and may show up as a soft halo along the edge. A larger rectangle or a slightly higher radius gives the filter more surrounding context, which can make the transition cleaner.
File Limits and What Triggers a Rejection
| Limit | Value |
|---|---|
| Supported input formats | JPEG, PNG, WebP |
| Maximum file size | 25 MB per file |
| Decoded-image safety budget | 40 megapixels |
| Output format | PNG |
| Output dimensions | Source image's natural pixel width and height |
| Processing location | Current browser tab, no upload |
Files outside the size or megapixel limits are rejected with a specific message before the blur is generated, so the reader does not have to wait for a long decode to fail. The 40-megapixel ceiling exists because compressed file size does not reveal how much memory a browser needs for a canvas. A 24-megapixel photo is well inside the budget; a 50-megapixel photo at any compressed size will be rejected. Readers working with a folder that mixes formats can run the tool without converting first, because JPEG, PNG, and WebP are all accepted. For an overview of how each format behaves under the blur effect, How to Use Blur Effect on JPG, PNG, and WebP Files lays out the differences side by side.
Preview Canvas vs Downloaded PNG
| Stage | What it shows | Why it is different |
|---|---|---|
| On-page preview canvas | A possibly reduced version of the source | Reduced so a large photo remains practical to inspect in the browser |
| Result text before download | The source image's natural pixel width and height | Reported so the reader knows what the exported PNG will actually be |
| Downloaded PNG file | A separate offscreen canvas at the source's natural dimensions | Built from a separate full-resolution canvas, not the reduced preview |
This separation prevents the common failure where a convenient 900-pixel preview silently becomes a low-resolution export. The blur is applied on the offscreen canvas at full resolution, so a rectangle drawn at X=1200, Y=400 on a 4000-pixel-wide photo will land on the same physical pixel in the downloaded file. The reduction is purely a display choice for the on-page canvas, and the result text next to the preview button exists specifically to expose the actual export dimensions before the reader commits to a download.
Blur Is Not Secure Redaction
A blurred face, address, account number, license plate, document field, or private message may remain recognizable, recoverable, or inferable from context. The tool provides a visual blur effect for design and casual privacy needs; it does not certify anonymization, de-identification, or irreversible deletion. For material that genuinely must be disclosed safely, remove or cover the source pixels with a suitable redaction workflow and verify the exported file, its metadata, hidden layers, and surrounding context. A reliable check is to take the downloaded PNG, open it at 100 percent zoom, and ask whether a person who already knows the hidden detail could still pick it out, recover it from the blurred region, or infer it from the rest of the frame. If the answer to any of those questions is yes, blur alone is not the right tool.
Putting It Together for a Folder of Images
The most reliable workflow is to keep the originals in one folder untouched, copy the files that actually need softening into a working folder, and process them one at a time through the same browser tab. Each PNG is saved to a separate output folder with the same filename so the pair stays obvious. Coordinates are reused only when source dimensions match exactly, and otherwise re-measured from the displayed preview. The exported PNG is opened at 100 percent zoom before the original is deleted, which catches off-by-one rectangle coordinates, kernel bleed near the rectangle edge, and any case where the blur radius was too small to do its job. Because canvas filtering and PNG export are both browser-native operations, the whole loop stays local, so the reader never has to decide whether a third-party server has a copy of the photos that were just softened.
For readers who want a deeper reference on the underlying canvas mechanism, the MDN documentation for the canvas filter property describes the blur effect that the tool applies through the standard browser pipeline.