- Upload one image at a time so you can frame each resize deliberately.
- Type the target pixel width and pixel height. For UK marketing teams: 1200×630 for share cards, 1080×1080 for Instagram squares, 1080×1920 for Stories.
- Pick Stretch only if you genuinely want distortion; Crop for fixed-aspect placements; Fit when nothing must be lost from the edges.
- Choose an output format from JPG, PNG or WebP. WebP wins for the open web; JPG is the safest fallback.
- Hit Resizer and download. Compare the new file size displayed next to the link with the original.
Technical handbook from the Gilhe team
The following sections extend the quick steps above with the engineering detail we would give to a colleague. Gilhe Image Converter is built around speed-first, low-friction format conversions; every recommendation below is written against real workloads, not generic marketing copy. If anything conflicts with your in-house policy, your policy wins — but if you are starting from scratch, this is the baseline we ship in production.
Stretch, Crop and Fit in production terms
Stretch maps the source rectangle exactly to your width/height box, discarding aspect ratio. Use it only when distortion is irrelevant (abstract textures) or when you intentionally squash for a stylised effect. Crop keeps aspect ratio and fills the box by trimming overflow; it is the right mode for fixed-aspect slots such as Instagram squares or banner rails where losing edge pixels is acceptable. Fit inscribes the image inside the box without trimming; transparent PNG/WebP canvases may show empty gutters. For marketplace main images that mandate a percentage of frame occupied by the product, Crop is usually correct; Fit is correct when you must not lose any edge detail.
Upscaling limitations
Enlarging an image cannot invent true high-frequency detail. Upscaled output interpolates existing pixels; sharpening in post helps perceived acuity but does not restore information that was never captured. Gilhe allows upscaling because many social templates require a minimum pixel dimension larger than a small source, but you should treat upscaled assets as pragmatic, not archival.
Social and performance dimensions
Modern retina displays mean you should target roughly 2× the CSS pixel width of the eventual slot when exporting raster heroes. A 600 CSS-pixel-wide column benefits from a 1200-pixel source. Our resizer makes that explicit: type the CSS dimension doubled, or type the CSS dimension once and accept softer output on high-DPI phones. The product blog on Gilhe Image Converter documents platform-specific safe harbours.
Encoder interaction with resize
Resizing is applied before lossy recompression. That ordering minimises generation loss. If you resize in one tool and compress in another, you add generations; Gilhe keeps both operations in one pass for supported paths. PNG outputs are lossless but can balloon in size if you resize up dramatically; consider WebP lossless as an intermediate if byte size matters.
Operational guidance
For teams standardising imagery, pick three canonical output widths (mobile, tablet, desktop) and encode presets into your design-system documentation. Gilhe is intentionally narrow so that it drops into such a preset-driven pipeline without retraining staff on seventy checkboxes. For speed, bandwidth efficiency and dependable conversion, treat each resize as a deterministic function: same source, same settings, same bytes out.