A screenshot and a camera photo can look similar after you drop them on the same canvas. They are not the same kind of file. Screenshots are usually PNG (or a lossless framebuffer dump later saved as PNG): they are a copy of pixels the operating system already composited — type, icons, round rects, subpixel antialiasing. Camera photos are usually JPEG or HEIC: a lossy encoding of a sensor image, plus a rich EXIF block. Covers that sit cleanly on UI text can grow JPEG ringing on a re-saved photo. Covers that hide a face in a noisy night shot can look unnecessarily harsh on a settings screen. The metadata story splits too: photos carry GPS and device tags; screenshots usually do not carry Camera EXIF, but they can carry OS or app tags and they often contain secrets in the pixels of the chrome itself (names, URLs, notification banners).
HideShot edits both as rasters in this tab. Use redact a screenshot when the still is UI. Use redact an image online when the still is a general photograph. This page is the comparison: why a technique that worked on one transfer is the wrong default on the other.
Guide
Start from the container, not from the subject. If the file is a PNG from a Print Screen, Volume-Up+Power, or Snipping Tool path, you are almost certainly on a screenshot pipeline. If the file is HEIC from an iPhone Camera roll or JPEG from a Pixel or a DSLR, you are on a camera pipeline. Mixed cases exist — a photo of a monitor is a camera file of a screen; a screenshot of a photo in an app is a screenshot of someone else’s JPEG. Treat the file you actually have.
Screenshot: lossless UI, secrets in chrome, little camera EXIF
A screenshot is sampled from a rendered frame. Text is sharp. Edges are 1-pixel aligned. Contrast is often near-binary (dark type on a light surface, or the reverse in dark mode). That is why a black box or a coarse mosaic looks “official” and why a light Gaussian looks smeared and ugly — there is no optical bokeh to hide in. It is also why incomplete coverage is obvious at 100%: a single missed character in a URL bar is readable.
What to cover is often not a face. It is a display name, an email, a message preview, a window title, a URL with a token, a sidebar of other channels, a clock that timestamps a private event, a battery and carrier cluster that is harmless, a notification shade that is not. Some screenshot tools on desktop write a Software string or creation time into PNG metadata. They do not write a GPS IFD from the Camera app. Do not skip a metadata check out of dogma; do not invent Camera GPS that was never there either.
Keep the export as PNG when you can. Re-encoding a screenshot as JPEG to “make it smaller” is how you add block artifacts around every box you drew and sometimes around every glyph you left visible. If you must JPEG a screenshot, use a high quality factor and re-check the cover edges after the encode — the ringing can make a box look incomplete when it was complete in PNG.
Camera photo: lossy samples, optical scene, rich EXIF
A JPEG is already quantized. HEIC is a different lossy family. Both store a scene with noise, demosaicing, and often in-camera sharpening. A face or a plate sits in that texture. A Gaussian blur on that texture is a second low-pass on top of the first. Deconvolution arguments from the blur-versus-box page apply more here than on a PNG screenshot of a settings toggle. High-contrast plates and photographed documents behave more like screenshots (sharp type on a plane) and should be treated with opaque fills, not “photographic” soft blur.
Re-saving a redacted photo as JPEG is a second generation of loss. The edge of a black box or a mosaic becomes a place the codec spends bits on a discontinuity, which shows up as ringing. If the recipient needs a JPEG, export once at high quality and verify that generation — do not bounce JPEG to JPEG through several apps after the cover. PNG from HideShot is a stable flattened raster; converting that PNG to JPEG later is one controlled generation, which is easier to inspect than five mystery re-encodes.
EXIF is the other half of a camera file. Covering pixels does not clear GPS. That split is the metadata page in this batch. Screenshots are not a free pass to skip inspection, but camera files are where people are routinely surprised by a pin.
Techniques that do not transfer cleanly
Soft blur on UI text. On a screenshot, it looks like a mistake and is easier to sharpen than it looks. Prefer a box or a large mosaic for strings. On a distant face in a photo, a heavy blur or mosaic can be the right visual trade.
Tiny mosaics on a photographed whiteboard or a paper statement. The photo already has JPEG blocks. A 4-pixel mosaic on top of that can still leave glyph structure. Use larger cells or a box, as you would for a screenshot of the same text.
Assuming a screenshot of a photo inherited the photo’s GPS. The screenshot file is a new raster. The photo in the Camera roll still has the tags. Sending the screenshot may avoid Camera EXIF; it does not clean the original you still hold, and it may still show the scene — including a readable plate you forgot to cover in the grab.
Using the same margin on both. Screenshots need boxes that include antialiased glyph stems. Photos need boxes that include motion blur and JPEG ringing around a plate. Check each export at 200%.
How to handle either file in HideShot
- Know which you have. PNG from a screen grab versus JPEG/HEIC from Camera. Convert HEIC to a still HideShot can open if needed, then cover that still.
- For UI strings, prefer Black Box or coarse Pixelate. For photographic bystanders, Pixelate or strong Blur can be enough if they fail a zoom test. For identifiers in either kind of file, use a box.
- Download PNG. If you must deliver JPEG, convert once, then zoom the JPEG edges.
- Inspect metadata on camera files as a second step. On screenshots, inspect anyway if a Software or timestamp field would matter, but do not confuse that with Camera GPS.
Dedicated screenshot chrome coverage is redact a screenshot. Dedicated general image cover is redact an image online. Use this comparison when you are switching between the two in the same hour and the last setting that “worked” is about to be applied to the wrong kind of raster.
Mistakes that mix the pipelines
JPEG-compressing a redacted screenshot to attach it, then wondering why the box looks dirty. Keep PNG, or accept one high-quality JPEG and re-verify.
Treating a photo of a laptop screen as a screenshot. It is a camera file: EXIF, glare, and optical focus. Cover the glare that still shows a password. Strip tags.
Covering a face in a photo and leaving the notification screenshot on the lock screen in the same frame unread. That is screenshot-like chrome inside a photograph. Read the pixels.
Skipping zoom on a screenshot because “text is obvious.” Missed characters in a URL are obvious only if you look.