Image Combiner
Join 2–20 static JPEG/PNG images into one strip or stack locally, without uploads, labels or pagination. Refresh clears memory.
2–20 static JPEG/PNG files; 20 MiB each, 100 MiB total; each source and final canvas ≤20 megapixels / 8,192 px per edge. All processing stays local. Decode overhead can exceed raster storage; allocation can fail on smaller devices.
Original files and small thumbnails are kept for reconfiguration. Reset offers one memory-only Undo, with no duplicate native rasters. Changing settings/order invalidates generated output. PNG preserves alpha; JPEG flattens against the explicit export color. Preview is scaled from the actual encoded output, not a separate layout.
How to combine images
- Add images and check their oriented dimensions. Move controls set source order; failed images require retry or removal.
- Choose direction, size policy, alignment, gap, outer padding and background. Keep-original never scales. Match scales proportionally to an explicit height horizontally or width vertically.
- Review final geometry and scales, Generate, inspect the complete preview, download PNG/JPEG or copy the dimension summary.
Worked geometry example
100×80 and 50×100 originals horizontally, gap 10 and padding 5: final 170×110, centered origins (5,15) and (115,5). Vertically: 110×200, origins (5,5) and (30,95). Matching height 100 changes the first to 125×100 and produces 195×110 horizontally.
Rounding, pixels and transparency
Matched main-axis sizes round nearest integer half-up, minimum one pixel. Upscaling creates no new detail. Cross-axis center offset floors half the surplus; an odd extra pixel stays at the end. Main positions are cumulative sizes plus gaps and padding: no crop or overlap. Originals are orientation-normalized once and drawn directly into each new composition; reordering does not reencode source bytes.
Transparent PNG leaves gaps/padding clear and preserves original alpha. Opaque background fills behind source alpha too. JPEG composites the full composition over its explicit flattening background; checkerboard is preview-only. Browser-decoded colors, resampling and JPEG compression may differ from encoded originals. Metadata/profiles are not preserved.
Limits and alternatives
Inputs and final canvas are limited to 20 megapixels/8192 px per edge. Final dimensions are calculated before allocation; oversize settings fail rather than silently shrinking. One source is decoded at a time. Allocation/encoding can still fail on limited devices. No panorama seam detection, blending, freeform layers or ZIP. Use contact sheets for labelled grids and pagination rather than a continuous composition.
What should we improve next?
Help shape the next update to Image combiner. Tell us what would make it better for you.
Prefer email? feedback@tooltulip.com