JPG Quality Explained
Last reviewed:
Overview
This guide explains how a JPG quality setting directs a lossy encoder's tradeoff between retaining visual detail and reducing stored information. It is not a universal percentage of original quality, and the same number can behave differently across encoders, applications, source images, and supporting options.
The most reliable judgment comes from the actual output. A useful review considers intended display size, important image regions, measured file size, and the availability of the original. The setting is an input to that decision, not a final quality certificate.
What a quality control changes
JPEG encoders decide how much precision to keep for different parts of the image representation. A higher quality choice commonly retains more detail and stores more information, while a lower choice may remove more detail and expose visible artifacts.
The scale belongs to the encoder implementation. Quality values from two programs should not be assumed equivalent, and supporting decisions such as color sampling can also affect appearance. Compare real files rather than transferring a number blindly between tools.
Why quality is not a percentage
A setting labeled with a number does not report the percentage of original pixels retained or predict a fixed similarity score. It selects an encoder policy whose effect depends on image content and implementation.
Source quality sets the ceiling
Encoding cannot recover detail missing from the source. A small, blurred, or previously compressed image remains limited even if the next save uses a generous setting. Upscaling or changing extensions does not recreate authentic texture.
Repeated lossy saves can compound changes because each encoder works from pixels reconstructed from an earlier approximation. Maintain the strongest available source and generate new delivery copies from that source whenever practical.
How to review JPG output
Inspect hair, foliage, fabric, gradients, text, and high-contrast outlines. Look for block patterns, ringing, banding, color softness, or loss of fine texture. Check at the actual display size as well as a closer view for critical assets.
Balance visual findings with the destination's compatibility and transfer needs. A larger file is not automatically better, and a smaller file is not successful if artifacts distract from its purpose. Preserve the original before accepting any delivery copy.
For a repeatable review, compare candidates with synchronized zoom and identical color-management conditions. Record the selected destination and intended use so another editor understands why that output was accepted without treating its quality number as a universal standard. Include representative dark and bright regions in that decision.
Format capability and current encoder policy
JPG format capability
As a file format, JPEG uses lossy compression to keep photographic files compact and broadly compatible. Lossy; repeated encoding can add artifacts. It is best suited to photographs, email attachments, and images that must open almost anywhere. These capabilities describe the format itself, not a promise about a particular encoder.
Current ForgeConvert JPG output policy
Encoded at quality 82 with 4:2:0 chroma subsampling. Normal output metadata is stripped.
For JPG Quality Explained, the current workflow does not permanently store uploaded or converted files, accepts up to 20 files of 8 MB each, limits decoded images to 40 megapixels, and allows 15 seconds for processing. These operating limits come from the active converter configuration.
Convert an image
See also
Frequently asked questions
Does the maximum JPG quality setting mean lossless?
No. Standard JPEG remains a lossy format, and a maximum control value is an encoder setting rather than a lossless guarantee.
Why do equal quality numbers produce different files?
Applications can use different encoders and supporting options, so their numerical scales and resulting tradeoffs are not directly interchangeable.
Can a high setting restore an old low-quality JPG?
No. A later encoding can preserve the decoded input more gently, but it cannot recreate information removed by earlier compression.