Combine Images is a browser-based tool that takes two to twelve local raster files and arranges them into a single PNG using horizontal, vertical, or grid layouts. The output keeps every input's original pixel dimensions, centers shorter images in their cells, and never stretches, crops, or resamples a single source. The whole pipeline — decode, layout, paint, encode, download — runs inside the active tab, so source files never leave the device. Supported inputs are PNG, JPEG, WebP, GIF, AVIF, and BMP, while SVG is deliberately rejected because external references and large render instructions would break the bounded local workflow. Each layout choice places images at their decoded pixel sizes, so no source is distorted by the merger. The exported file uses the browser's built-in PNG encoder, so the result is identical to what your tab paints pixel by pixel. PNG output preserves sharp pixels and decoded transparency, but the selected opaque background is painted first, so transparent areas become that color in the final file.
Inside Canva's Blend Image App
Canva's Blend Image feature lives in the top toolbar, and it only appears when a single image element is selected on the canvas. The controls include an Opacity slider and a list of blend modes — Multiply, Screen, Overlay, Darken, Lighten, Color Burn, and others — that recalculate each pixel of the active layer against every layer underneath. Because the effect runs on the composited page, two stacked photos can become one tinted image, a texture can darken or lighten a portrait, and a gradient can wash over a flat illustration. The result is one Canva page that you can export as PNG, JPG, PDF, or share as a link.
What the Blend Image app does not do is take several local files and place them side by side or in a grid as separate, original-pixel items. Layered blending and tiled arrangement are different problems: blending mixes colors where layers overlap, while arranging keeps every source at its own dimensions and only positions it on a shared canvas. If the job is to line up three product shots into one banner, merge two screenshots into a single comparison image, or build a centered grid of icons, the Blend Image app is the wrong place to look.
Using the Blend Image App in Canva
Working inside Canva takes a few deliberate steps, and the order matters because locked or grouped elements hide the blending controls.
- Add two or more images to your design and arrange them so one sits on top of another where you want the blend to happen.
- Click the top image to select it. Confirm the selection handles appear and the image is unlocked; right-click to unlock if necessary and use the right-click menu to ungroup the layer so it becomes a single editable element.
- Open the image's toolbar at the top of the canvas. Adjust Opacity to soften the layer, then pick a blend mode — Multiply darkens, Screen lightens, Overlay increases contrast — to merge the colors with what sits underneath.
- If two photos meet at a visible seam, cover the join with a gradient element, a soft-edged shape, a frame crop, or a foreground object such as foliage, paper texture, or a light leak.
- Export the page as PNG, JPG, or PDF when the blend looks right. The composited page keeps the effect, while the source images remain separate layers inside Canva until you export.
This is the right workflow for layered, in-canvas compositions. For a row, column, or grid of local files merged into one static PNG outside the editor, a different approach fits better.
When Combining Locally Is the Better Fit
Arranging two to twelve local photos into one PNG is a different problem from layering with blend modes. The result is not a composited image with merged pixels but a single canvas that holds each source at its original size, in a chosen order, with a chosen background. That is exactly what the Combine Images tool does in your browser. It accepts PNG, JPEG, WebP, GIF, AVIF, and BMP, places them in the order the browser sees them, and exports one PNG that contains every input unchanged.
The local approach also matters for sensitive material. Because decoding, layout calculation, drawing, PNG encoding, and download generation all happen in the current tab, the source files are never sent to a server. If a project mixes client logos, product photography, or private screenshots, that privacy boundary is part of the workflow. Canva uploads design state to its servers as part of saving and sharing, which is fine for cloud-native projects but adds friction when the goal is simply to merge a few images and download one file. For more on how the two workflows compare and what each one is designed to do, see how to use the Blend Image app in Canva and beyond.
How to Combine Two to Twelve Photos Into One PNG
The combine workflow is three deliberate steps, and each one feeds the calculation the next step performs.
- Choose between two and twelve supported raster images in the order you want them placed. The browser keeps that order, so the first file goes left-most in horizontal mode or top-most in vertical mode.
- Select a horizontal, vertical, or grid layout, then set the gap and background color. The gap is an integer from zero to one hundred pixels, and the background is a six-digit color that fills unused canvas space and the area between images.
- Choose Combine images, verify the output dimensions shown after layout, and download the generated PNG. The file is built with the browser's PNG encoder, so the saved file matches what your tab painted.
If the second step rejects a file, the cause is almost always that the bytes are corrupt, the codec is unsupported, or the input is SVG. SVG is intentionally excluded; the tool is designed for browser-decodable raster formats only. If you need to convert a vector first, run it through an SVG-to-raster converter, then come back to this step.
Layout Options: Horizontal, Vertical, and Grid
Three deterministic layouts are available, and each one calculates output size from integer decoded dimensions. None of them stretches, crops, or rescales a source.
| Layout | Direction | Output width | Output height | Empty space rule |
|---|---|---|---|---|
| Horizontal | left → right | sum of widths + gaps | tallest image | shorter images centered vertically |
| Vertical | top → bottom | widest image | sum of heights + gaps | narrower images centered horizontally |
| Grid | ceiling(√count) columns | cols × max width + gaps | rows × max height + gaps | empty cells and unused padding show background color |
For a horizontal row, the calculation is concrete: three images with decoded widths of 400, 600, and 500 pixels and a 20 pixel gap produce a canvas of 400 + 20 + 600 + 20 + 500 = 1540 pixels wide, with an output height equal to the tallest image. The two shorter images are centered against that tallest height, so the row looks balanced without resampling.
Grid layouts use the smallest square-like column count, taken as the ceiling of the square root of the image count. Four images land as 2 × 2, six images as 3 × 2, nine images as 3 × 3, and ten images as 4 × 3 with two empty cells. Every cell uses the widest and tallest decoded images as its bounds; each source is centered inside its cell without stretching. The empty cells and the padding around shorter sources show the chosen background color.
File Limits, Transparency, and Animation Behavior
Hard limits exist on three axes: per-file bytes, total bytes, and canvas geometry. Each source file must be no larger than 25 MiB, all selected files together must not exceed 100 MiB, and the combined canvas cannot exceed 8192 pixels on either side or 32 million pixels in area. These numbers are designed to prevent the kind of memory spike that can crash a browser tab, not to advertise a soft ceiling. If the file picker lets you select an item but the layout step returns nothing, the input is likely over one of those limits or fails to decode.
Transparency is treated as part of the canvas paint order. The tool paints the chosen six-digit background first, then draws every bitmap on top in the selected order. PNG preserves sharp pixels and decoded transparency, but because the background is opaque and painted first, transparent areas in the source become that background color in the exported PNG.
Animated GIF, WebP, or AVIF inputs contribute only the single frame the browser decodes on its image bitmap path. Animation is not preserved in the output. The exported file is one static PNG with no metadata, no EXIF, no color profile, no original filename, and no layers. For print production or color-managed compositing, verify the exported PNG in dedicated software before sending it downstream.
Matching the Layout to the Job
Picking a layout is mostly about the canvas shape you can already picture. A horizontal row suits banner-style composites, before-and-after comparisons, and timeline strips where every source needs to read at the same baseline. A vertical column fits mood boards, step-by-step screenshots, and Pinterest-style pins where the reader scrolls downward through the story. A grid suits product catalogs, character sheets, contact sheets, and icon sets where uniform cells keep the eye calm across many items.
Two practical details shape the choice as well. First, the gap is an integer pixel value between zero and one hundred, and it controls visual separation as well as the math; a zero gap means the images touch, while a wide gap creates a poster-like frame around each source. Second, the background color is what every empty cell and the canvas border will show, so picking it deliberately lets you build a consistent brand color or a transparent match for downstream placement. The image stays at original resolution, so nothing in the result is worse than the worst source you fed in.
For a quick check before clicking combine, run the math on paper: output width and height should both fit inside 8192 pixels and total fewer than 32 million pixels, and the sum of all file sizes should sit comfortably below 100 MiB. If those numbers work, the layout step will succeed. If they don't, scale the sources down or compress them first, then return to the combine step.
For a deeper look, see Placeholder Image Generator Alternative: Skip Remote URLs.