Upscaling can improve perceived detail but cannot recreate missing source information perfectly. Best results come from clean inputs with minimal noise and compression artifacts. For low-quality sources, run light denoise and compare 2x versus 4x before batch processing.
Use 2x for most web and documentation tasks, then resize down if needed. Use 4x only when you truly need larger dimensions, because larger outputs increase processing time and file size significantly.
For logos, screenshots, and UI captures, review edges at 100% zoom after processing. These assets often benefit from the simpler setting path here, but they still need a quick visual check before you export a full batch for production use.
4x stays available only when the requested result fits the configured server output budget for that image.
Auto is a good default. Force on for visibly noisy camera images; force off when you need to preserve grain texture.
No. If the original high-resolution source exists, that will almost always outperform any upscale result.
Use Upscale when a small image needs a sharper delivery version: enlarge product photos, improve low-resolution assets, prepare marketplace images, or create a higher-resolution source before conversion, cleanup, compression, or background removal.