The fit mode is the decision
Several images are arranged into a grid, a row or a column, with a cell size, a gap and a background colour. Each image is fitted into its cell one of two ways, and that single choice accounts for most of what makes a collage look deliberate or accidental.
Cover fills the cell completely and crops whatever does not fit. Every tile is the same size with no gaps, which is what a photo grid should look like.
Contain fits the whole image inside its cell with space around it. Nothing is cropped, and images of different proportions leave different amounts of space.
The [msm_tool slug="collage-maker"] offers both, and the right answer depends entirely on what the images are.
Cover for photographs, contain for products
Photographs survive cropping. A landscape has more landscape at the edges; a portrait has more room around the subject. Cover produces a clean grid and loses nothing anyone will miss.
Logos, product shots on white, screenshots and diagrams do not survive cropping — cutting the edge off a logo makes it a broken logo. Those want contain, and the space around them is a feature rather than a flaw.
Mixing the two kinds of image in one collage is where it goes wrong, and the fix is to split them into two collages rather than to compromise on the fit.
Cover crops from the centre
A subject near an edge of its original photograph is the part that goes. That is fine for landscapes and awkward for portraits where the person stands to one side.
If the composition is unusual, crop those images individually first so the important part sits centrally, then build the collage. It is more work and it is the difference between a grid of faces and a grid of foreheads.
Mixed orientations need deciding in advance
A portrait and a landscape image in the same square grid lose very different amounts to a cover fit — the portrait loses its top and bottom, the landscape its sides.
Three options, and it is worth choosing rather than discovering:
- Use contain and accept the space, with a background colour that makes it look intentional.
- Pre-crop everything to the cell's ratio so cover has nothing to remove.
- Use a layout whose cells match the images — a row for landscapes, a column for portraits.
Gap and background work together
A gap of zero produces a seamless mosaic where the images touch. A gap with a contrasting background produces framed tiles.
Both work; the middle ground rarely does. A small gap in a colour close to the images reads as a printing error rather than as a design choice. Either commit to the seam or commit to the frame.
Check the output dimensions
Cell size multiplied by the grid, and the result is a single image. A 4×4 grid of 800-pixel cells is 3200 pixels square, which is a large file — particularly in a lossless format.
Decide the final size you need and work backwards to the cell size, rather than choosing a comfortable cell size and finding out afterwards.
Order matters more than you expect
A collage is read like a page: top-left first, then across, then down. The strongest image should not be in the bottom-right corner, and two similar images should not sit adjacent, because the eye reads them as a pair whether or not you meant it.
Vary the visual weight across the grid and the whole thing holds together better than any individual placement would suggest.