A photo delivery is a chain of four file operations that always run in the same order: resize to the size the client will actually use, compress the set, add any watermark the proofs need, then strip the metadata. Do them in that order and you never re-encode twice or scale your own mark.
The order matters because each step changes what the next one sees. Resizing after compressing throws away the savings. Watermarking before resizing scales the mark with the frame. And stripping metadata has to happen on the file you actually send, because some editors write EXIF back on export. None of these steps uploads the photo, which is the point for client work.
1. Resize to the size the client will use
A full resolution export is rarely what a client needs. A web gallery, a proofing page and an Instagram post all want something smaller, and shipping a 6000 pixel file wastes their storage and your upload time. Image Resizer opens JPG, PNG, WebP and AVIF and saves JPG, PNG or WebP, with the ratio locked by default so the frame keeps its proportions.
Set the long edge directly, or use the percentage shortcuts to halve a set quickly. The tool redraws the photo on a canvas, which means EXIF data such as camera model, capture date and GPS coordinates is dropped as a side effect. That is usually welcome in a delivery, but know that it happens here rather than counting on a later step.
Two limits. Downscaling resamples the pixels, so very fine detail can soften at aggressive ratios, and saving as JPG or WebP re-encodes what is left. Enlarging does not add detail, so a small file cannot be rescued by typing a bigger number.
Image Resizer
Set a delivery size with the ratio locked, then export as JPG, PNG or WebP. Nothing is uploaded.
2. Compress the set in one pass
Image Compressor accepts multiple files, so the whole gallery can be tuned once and exported together. Keep the quality slider between 75 and 90 for photographs. AVIF and WebP produce the smallest files, while PNG with OxiPNG recompresses losslessly and suits graphics, logos and screenshots inside the delivery.
There is no fixed cap on how many images you add in one run; the real limit is browser memory, so a very large gallery is better handled in a few passes than one. Two honest caveats. There is no target size field, so you lower the slider until the live preview shows a number you can accept. And a small screenshot or logo can come out larger, because it was already stored efficiently and a lossy re-encode adds overhead.
Image Compressor
Batch compress a gallery with a quality slider and a live before and after preview.
3. Watermark the proofs, not the finals
Image Watermark places text at one of five corner and centre positions, or tiles it diagonally across the frame, with size, colour, opacity and rotation, and the preview updates live. Style it once and export the whole batch with the same mark.
Two things to know before you rely on it. The tool draws text, not an image, so a logo PNG cannot be used as the mark; if the brand mark matters, it has to be added in your editor before the file reaches this chain. And the watermark is drawn into the pixels at full resolution, which means it cannot be removed afterwards. Use it on the proofs you send for approval and export the clean files separately for final delivery.
Image Watermark
Text watermark in five positions or a tiled layout, with batch export at full resolution.
4. Strip metadata before sending
Remove EXIF cuts metadata out of the JPEG or PNG byte stream instead of re-encoding it, so the compressed pixel data, quality tables and colour data stay byte for byte identical. That is a different operation from opening the file in an editor and saving it again, which would re-encode the pixels and lose a little quality for no reason.
The default mode strips EXIF, XMP, IPTC and comments while keeping the ICC colour profile, so colours look the same on the client's screen. When a photo carries GPS coordinates the tool lists latitude, longitude and altitude before you commit, and it re-reads the output afterwards to confirm that no metadata segments remain.
Three limits. Only JPEG and PNG are supported, so WebP files have to be converted first and HEIC cannot be opened at all. The second removal option also drops the colour profile, which changes how the image looks in a colour managed viewer, so leave it off unless you specifically want a bare file. And stripping metadata does not watermark anything, so it is not a substitute for the previous step.
Remove EXIF
Strip GPS, camera and capture metadata from JPEG and PNG without touching the pixels.
When the client wants a PDF or a bigger file
Three side paths come up often enough to plan for. A proof sheet, a model release or a signed delivery note usually has to arrive as one PDF rather than twenty attachments. Image to PDF turns the images into pages, and you can set A4 or Letter, orientation, margin and whether each image fits inside the page or fills it. Pages keep the order you drag the cards into, and no fixed image count is imposed.
An old frame that a client wants printed goes through Image Upscaler. It runs on your own graphics chip, offers 2x, 4x, 6x or 8x, with 2x and 4x in one pass and 6x and 8x chaining two passes, and downloads about one megabyte of model on first use. It rebuilds edge contrast and cleans compression artifacts, but it cannot recover detail that was never captured, so a large print will look clean rather than newly detailed.
Format changes are handled by Image Converter, which opens JPG, PNG, WebP, AVIF and GIF. GIF is an input only; the output choices are the still formats. Transparent areas survive conversion to PNG, WebP and AVIF, and turn white in JPEG because the format has no transparency.
What this chain cannot do
- No RAW or HEIC decoding. Camera formats and phone HEIC files cannot be opened. Develop RAW in your editor and export JPG or PNG before the delivery chain starts.
- No colour management conversion. The tools can keep an ICC profile or strip it, but they do not convert between colour spaces. Soft proofing belongs in your editor.
- No image logo watermark. The watermark is text only. A brand mark has to be placed before this chain.
- No batch renaming. Files keep the names they arrive with, so numbering and client naming conventions are yours to set outside the browser.
- No real detail from upscaling. The model produces plausible pixels, not recorded ones, which is enough for a small print and not enough to claim a recovered frame.