Skip to guide
Practical guide

Choose image dimensions, format and file size together

A practical sequence for preparing a readable photo, signature or document image for an upload form.

Maintained by Ajay Kumar Maurya · Updated 13 October 2026

Write down the destination’s requirements

Record the accepted format, pixel dimensions, maximum file size and any background or cropping requirements. These are separate checks. A small file can still have the wrong shape, and a correctly sized image can still be unreadable.

Use the instructions on the destination portal at the time you submit. A preset called “Exam” or “Passport” is a convenient starting size, not confirmation that it matches a particular application.

Start with a useful crop

Keep an untouched original. Crop out unnecessary blank space without cutting off a face, signature stroke or required document edge. If the source is blurred or badly lit, exporting more pixels cannot recover missing detail.

Illustrated sequence: dimensions, format, file size and legibility
A workflow illustration. Follow the destination’s actual specifications instead of copying example limits.

Match the dimensions before chasing a KB limit

Use Resize Image to set the output width and height. Contain preserves the whole image and can leave a border; Cover/Crop fills the canvas and may cut off edges. Inspect the preview before exporting.

Changing a DPI label alone does not reduce the pixel count of an existing image. For an online form that checks pixels, compare the exported width and height with those requirements.

Choose an accepted format

FormatPractical consideration
JPGOften useful for photos when accepted. It does not preserve transparency and quality reduction can soften fine detail.
PNGSupports transparency and can suit line art, but may be larger for a photo.
WebPCan be efficient, but use it only if the receiving website accepts it.

Reduce the file size gradually

Use the image compressor after choosing dimensions and format. A target size is a request to the tool, not a promise that every source can fit without visible loss. Inspect the reported output size, then open the download independently.

If a signature loses thin strokes, increase quality, use a clearer source, or reduce unnecessary whitespace. Do not repeatedly compress an already damaged output.

Check the actual downloaded file

The signature tool and photo crop tool provide task-specific controls. Their output still needs your review.

When pixels and KB cannot both be satisfied

The signature and photo tools preserve their requested pixel dimensions. An above-limit size badge means that the file still exceeds the chosen KB value. The general image compressor can downscale further while searching for a smaller file; its result card reports the exported dimensions. If both dimensions and bytes are mandatory, inspect both in the download and choose a clearer or less complex source when necessary.

Objective Visual Fidelity Metrics: SSIM and PSNR

Evaluating image compression quality requires moving beyond subjective visual checks to objective, mathematical fidelity benchmarks recognized by computer vision engineers:

Quality Metric Mathematical Meaning Target Range for Official Uploads
Structural Similarity (SSIM) Measures structural correlation, luminance, and contrast preservation 0.92 to 0.98 (Virtually indistinguishable from original)
Peak Signal-to-Noise Ratio (PSNR) Logarithmic ratio between peak signal power and corrupting noise power 35 dB to 45 dB (High fidelity across facial features)
Chroma Subsampling 4:2:0 reduction of chrominance resolution relative to luminance Standard for JPEG; discards color bandwidth while keeping text sharp

Print Resolution vs Screen Density: The DPI Equation

A common misconception among portal applicants is confusing Dots Per Inch (DPI) with file weight. DPI represents the physical printing density: Physical Size (Inches) = Pixel Count / DPI. A 350 × 350 pixel image printed at 350 DPI occupies exactly 1.0 × 1.0 inch of paper. Increasing the DPI setting in software without changing pixel dimensions does not increase the file size; it simply adjusts the physical print scaling metadata embedded in the file header.