There are two different questions behind “without losing quality”: whether the merger resizes the pictures, and whether the output format discards information. PNG can avoid lossy encoding of the final canvas while the layout still reduces its pixel dimensions. Changing a JPEG to PNG also cannot restore detail already missing from the JPEG.
Start with a measurable size check
Take two original files measuring 1600 × 1200 pixels each. In the homepage editor, choose Horizontal, Original sizing, Free aspect ratio, zero spacing, and zero margin. With no lower output limit, the expected result is 3200 × 1200 pixels.
Inspect the dimensions in the saved file's properties. If you get a smaller result, check the sizing selector and output limits. Reduce intentionally matches the smaller source dimension. Magnify enlarges smaller sources. Crop to fill can cut off edges, especially in a grid. These are composition decisions, independent of PNG or JPEG compression.
Understand the 8000-pixel boundary
The homepage exporter caps the longest side at 8000 pixels and uses a smaller limit if selected in the size controls. Consider two 6000 × 3000 files joined horizontally without gaps. Their unscaled composition is 12000 × 3000.
To fit the cap, all dimensions scale by 8000 / 12000, or 2/3. The downloaded image becomes 8000 × 2000, and each photo occupies 4000 × 2000. A PNG of that result is still smaller in pixel detail than the two sources. The cap is a boundary, not a promise that every device can successfully process a canvas of that size.
The template editor has another boundary: templates and free layouts use an 800-pixel longest side and export PNG. Stitch mode instead joins source-sized images. Use the appropriate editor and verify the resulting dimensions rather than assuming every mode preserves full resolution.
Decide between PNG, JPEG, and WebP
| Content and purpose | Starting format | What to inspect |
|---|---|---|
| Screenshot with small letters | PNG | Letter edges at 100% |
| Logo or transparent background | PNG | Transparent regions and contours |
| Photograph for a smaller attachment | JPEG | Fine textures and high-contrast edges |
| Web image where supported | WebP | Browser compatibility and final size |
The homepage offers all three formats; the template editor offers PNG. Browser canvas export supports a quality setting for lossy formats such as JPEG and WebP. That number controls the encoder, so a setting of 100% does not mean the original file's bytes or metadata are preserved. See MDN's canvas export documentation for the distinction.
For JPEG, transparent areas become the selected background color. Use PNG when transparency is part of the intended result. Start from the original files again when trying another export, rather than reopening a previously compressed result repeatedly.
Inspect the saved image, not a screenshot of the preview
The page scales its preview to your screen. A screenshot of it records that displayed size and can be much smaller than the generated file. Download directly and open the result at 100% zoom. Compare a few specific features: a small letter, a thin line, a face, and a textured area.
If the saved file is sharp but its shared version is blurry, the destination may have resized or recompressed it. Keep a master copy and review the version a recipient actually receives. Enlarging the blurry version later does not reverse the loss.
What this workflow does not preserve
A merger produces a new raster image; it is not an archive containing the untouched source files. This guide does not promise byte-for-byte identity, retained camera metadata, retained animation, or identical color appearance across all devices. GIF inputs yield a still image. For a record that needs the originals, share or archive those source files separately alongside the merged preview.
Find where a blurry result changed
Compare source, downloaded file, and shared copy separately with the blurry image diagnostic workflow. Changing compression settings is useful only after checking whether pixel dimensions were reduced.