Drop your PNG files into the emerald zone below and get one document back. Every sample value survives: the image is decoded, compressed again with the same deflate algorithm PNG already uses, and written into the document as a FlateDecode stream. Transparency travels with it as a proper soft mask rather than being flattened onto a white box. Nothing is uploaded. This image to pdf converter runs on your own processor, inside this tab.
Your files are read by JavaScript running on your own machine. There is no upload, no queue, and no copy on our server to delete later.
Drag a card to reorder it, or use the arrows under each card. The order here is the order in the output. Each card is a preview of the finished page, so what you see is what the PDF will contain.
Esc close ← → move R rotate
Starting
This is your processor doing the work, so the time it takes depends on your machine and not on how many other people are on the site.
The engine above is live and ready to convert PNG to PDF right now. The notes below cover what happens inside it: why PNG has a genuinely lossless route into a document, what becomes of your alpha channel, the one PNG variant that loses both its quality and its transparency, why a large screenshot never shrinks on the way in, and where the batch limits sit.
Two minutes. Nothing to install, and no account for a first run. This online PDF editor runs inside the page you are reading, on your own processor, and your files are never transmitted because there is no upload step in the code to transmit them through.
Click the drop zone above, or drag your PNGs straight onto the page. Every file header is read as it lands. The card reports the pixel dimensions and the resolution the file claims, and it names any file that cannot take the lossless route. A card with no rewrite notice on it is a file going through with every sample intact.
The order of the cards on screen is the order of the pages in your document. Drag a card to move it. The sorter reads the digits in a filename as a number, so Screenshot 2 lands ahead of Screenshot 10 instead of somewhere after it.
Pick a standard sheet such as A4 and each image is scaled to fit inside it. Pick Fit to image and the page is built to the physical size the pHYs chunk claims, or to the resolution you set by hand when that chunk is empty. Then choose your margin.
Press Convert to PDF. The work happens on your own hardware. If any source carried transparency, open the finished document and view it against a white background and then a dark one. That comparison is the only reliable way to see whether the alpha channel survived or was quietly flattened.
Most tools sold as a free pdf converter are vague about what lossless means. They decode your image to a grid of pixels, hand that grid to whichever encoder their library defaults to, and write out whatever comes back. Sometimes that default is a JPEG encoder. When it is, the damage is permanent and nothing on the screen tells you it happened. PNG is one of only two formats with a route into a PDF that costs no image data at all.
PNG stores its pixels as a deflate-compressed stream inside IDAT chunks. A PDF image object stores its pixels as a deflate-compressed stream under the filter name FlateDecode. Those two use the same compression algorithm, which is what makes the route possible, although the transfer is a repack rather than a straight byte copy: PNG runs a prediction filter across every row before compressing it, and a PDF image stream does not read those filters. So the engine decompresses the image, undoes the row filters, and deflates the raw samples again.
That is the route a standard PNG takes here. Nothing is resampled, nothing is quantized, and no encoder makes a judgment call about what to discard. Every sample value in the finished document matches the sample value in your file, which is a stronger guarantee than a smaller byte count would ever be. The header is read first for dimensions, bit depth, color type and the interlace flag, at a cost of well under a millisecond per file.
Interlaced PNG is the exception, and it costs more than most tools admit. Interlacing stores the image as seven interleaved passes of increasing detail, an arrangement called Adam7. It solved a 1996 problem: showing a rough version of a picture over a dial-up modem before the whole file had arrived. Export tools with a stray Interlaced or Progressive checkbox still turn them out.
That seven-pass layout is not something the PDF library can read, so an interlaced file cannot take the lossless route. It is decoded and re-encoded once as a JPEG at quality 0.92. Two things follow from that: the image picks up one generation of compression loss, and any transparency it carried is flattened onto white, because JPEG has no alpha channel to hold it. The card warns you when it happens. The fix takes ten seconds: re-export the file from your editor with interlacing switched off and it goes straight back onto the lossless route.
Read the file card before you press convert. A card with no rewrite notice means that file went in with every sample intact. A card naming a reason means the file was decoded and written again, and the reason tells you what it cost. One file in twenty, usually. We would rather show you that one than print a blanket lossless claim over the whole batch and hope nobody checks.
A PDF has no alpha channel in the way a PNG does. It handles per-pixel opacity through a soft mask: a second greyscale image attached to the first, where every value sets the opacity of the matching pixel. Same result, different structural language. This is the point where a competent PDF engine separates itself from a script that calls a library and hopes.
When a transparent PNG goes through, its alpha channel is separated out and written as a soft mask on the PDF image object. A logo with a feathered drop shadow keeps the feathering. A screenshot with rounded transparent corners keeps the corners. Drop that graphic onto a colored background in a page layout and it composites the way it should, with no halo and no white rectangle sitting behind it.
The lazy alternative, which most online converters take, is to fill every transparent pixel with white before encoding and throw the alpha data away. On a white page nobody notices. On a colored background, inside a printed brochure, or on an architectural proof, that white box is obvious and permanent.
Two cases lose it here. An interlaced PNG is re-encoded as a JPEG, and JPEG has no mechanism for carrying opacity, so its transparent regions are filled in. The same applies to a WebP or a HEIC that takes the JPEG route. We flatten onto solid white rather than black, because a logo that turns into a black rectangle on a printed contract is the kind of failure a client finds before you do.
| PNG Color Type | What It Stores | Route Into the PDF |
|---|---|---|
| Truecolor | 8 bits each of red, green and blue | Lossless repack. Every sample identical. |
| Truecolor with alpha | RGB plus a dedicated opacity channel | Lossless repack, with the alpha separated out and written as a soft mask. |
| Indexed | A palette of up to 256 colors | Lossless repack, palette written as an indexed color space. |
| Greyscale | A single luminance channel | Lossless repack as a single-channel image. |
| Interlaced (Adam7) | Seven-pass layout for progressive rendering | Re-encoded once as JPEG at quality 0.92. Transparency is flattened onto white. |
A PNG measures itself in pixels. A PDF measures the page in points, at exactly 72 points to the inch. Converting between the two needs a resolution figure, and when the file supplies none the engine has to assume a value, which is the reason people find their document came out at a size nobody asked for.
PNG keeps its physical resolution in an optional block called the pHYs chunk, recorded in pixels per meter. Optional is the operative word. Screenshot utilities and design tools rarely bother writing it, so a mockup exported from Figma or a capture from your desktop usually arrives with that chunk missing entirely. The file genuinely does not know how large it is meant to be.
The engine reads pHYs when it is there. When it is not, it falls back to 96 dpi, the old desktop screen standard, and the card says as much rather than presenting a guess as a measurement.
If you scanned a legal document at 300 dpi and the file lost its resolution tag somewhere in an email round trip, choose Fit to image and set Assumed resolution to 300. The page comes out at the size of the original paper.
Check the scale before a long print run. Open page one at exactly 100 percent zoom and hold a ruler against it. Thirty seconds. Finding the same dpi error after four hundred pages have already gone to a commercial printer costs you the entire reprint.
People try to make PDF smaller by running ten 500 KB screenshots through a converter, then wonder why five megabytes came back. The arithmetic is working correctly. Because the image data takes the lossless route, the finished PDF is the sum of your original files plus a few kilobytes of document structure. A converter that returns a dramatically smaller file has recompressed your images, and what it removed is not coming back.
PNG files are large for a specific reason. Lossless compression can only exploit real mathematical redundancy, so how well it works depends entirely on what is in the picture. A screenshot of a code editor compresses beautifully, because it is mostly flat repeated color. A screenshot of a photograph, a smooth gradient background, or an interface with blurred glass panels compresses badly, because deflate finds almost no exact repetition to work with.
Running the finished PDF through a separate compression tool rarely achieves anything. The image data inside is already deflate-compressed, and compressing compressed data hands back a file the same size it started. The source graphics are the only real lever here.
A handful of transparent interface elements or a quick stack of receipts is trivial, and the free engine does that all day without complaint. Hundreds of high resolution graphics is a different problem. Every decoded image sits in your RAM uncompressed while the document is assembled, which is the price of processing locally, and a browser tab that runs out of memory does not fail politely.
Anonymous visitors get one operation every 12 hours. Free accounts get two. Both tiers cap a batch at 20 files or 100 MB, whichever limit you reach first. CartHacker Premium removes the operation counter and raises that ceiling to 100 files and 500 MB, without moving a single byte off your own machine.
The standard build for a PDF modifier online is an upload form that puts your files on a remote server, a backend script that processes them, and a temporary download link at the end. What protects you at that point is a privacy policy. A policy is a statement of intent, and it lasts exactly as long as the backup retention window, the staff who wrote it, and whoever acquires the company next.
This tool refuses uploads because it has nowhere to send them. The JavaScript on this page uses the browser File API, works on the samples using your CPU, and hands the document back through a Blob URL that exists only inside your tab. No server takes part, so there is no remote copy to leak, no graphic to scrape for training data, and no forgotten storage bucket holding your contracts.
You do not have to take that on trust. Load this page, disconnect the machine from the network, and drop a batch into the queue. It converts. That is a demonstration rather than a policy, and it takes under a minute to run.
Watch the network tab while it runs. Open developer tools, clear the network log, and convert a heavy batch. You will see one small request recording that an operation happened, and nothing outbound anywhere near the megabyte size of your images.
No. PNG and PDF both compress image data with the deflate algorithm, so the image is decompressed, its per-row prediction filters are undone, and the raw samples are deflated again into the document. Every sample value in the PDF matches the value in your original file. The one exception is an interlaced PNG, which cannot take that route and is re-encoded once as a JPEG instead.
Yes. The alpha channel is separated out and written as a PDF soft mask, which is how the format handles per-pixel opacity. A logo with a feathered shadow keeps its shadow, and dropping that graphic onto a colored background in a larger layout composites correctly. We do not flatten transparency onto a white box the way most basic converters do. The exception is an interlaced PNG: it is re-encoded as a JPEG, and its transparency is lost to a white background.
No. This pdf tools engine runs inside your browser using your own processor, and the code contains no upload endpoint at all. Your images are never transmitted anywhere. Disconnect your Wi-Fi once the page has loaded and run a conversion if you want to watch that for yourself.
Drop the whole stack into the zone and leave Merge all images into one PDF switched on. One image goes on each page, following the order of the cards on screen. Drag them to change PNG to PDF in whatever sequence you need, and the pagination of the finished document follows that arrangement exactly, so what you see on screen is what lands on paper.
Because the image data was copied across rather than recompressed. The finished PDF is the sum of your images plus a few kilobytes of document structure. If you need a smaller file, resize the source graphics before they go into the queue, or export flat artwork such as diagrams and terminal captures as an indexed 8-bit PNG first.
Yes. Baseline JPEGs are embedded byte for byte, which is even less work than the PNG route. BMP and GIF are rewritten as lossless PNGs. WebP goes to PNG when it is lossless or carries transparency and to JPEG otherwise, while AVIF always goes to JPEG. Drop them in together and the engine works out the routing per file.
The file has no resolution header. PNG stores resolution in an optional pHYs chunk that screenshot utilities almost never write, so the engine falls back to an assumed 96 dpi. Select Fit to image and set Assumed resolution to 300 to get a 1:1 physical reproduction.
No. The repack carries the pixel samples and the rendering instructions and nothing else, so text chunks, software creation strings, embedded color profiles and any EXIF block are dropped on the way through. Converting a PNG here is an effective way to remove the tracking metadata a design tool wrote into it.
No. There are no watermarks, trial banners or attribution links in the output, and the document is ready for commercial use exactly as it comes. The free tier limits how many operations you get and how large a batch can be. It does not touch the quality of what comes out.
Anonymous visitors get one operation every 12 hours and free accounts get two, with both capped at 20 files or 100 MB per batch to stop browsers running out of memory. CartHacker Premium removes the operation counter and raises the batch ceiling to 100 files and 500 MB. Processing still happens entirely on your own machine at either tier.
All of these run the same way: your file is read by JavaScript on your own machine, the work happens on your processor, and the result is handed back to you as a download that never existed anywhere else.
Find the apps that slow your store down, by name and in milliseconds, then answer shoppers' doubts with trust badges under Add to cart. Both are free, with no card and no access to your store admin.
Paste any page and get its three Core Web Vitals, the apps costing time, like Klaviyo or Judge.me, and the Shopify or WooCommerce fix.
Show the promises that settle doubts, from free shipping to a money-back guarantee, right under the Add to cart button. One line in your theme.
Change the link after printing and count the scans. 5 tracked codes on a free account, static codes unlimited.
Anonymous visitors get one operation per 12 hours, per network address. That is not a sales tactic, it is the only lever we have: this tool runs on your processor, not ours, so the run itself earns us nothing. An account tells us who you are, and the ceiling goes up the moment you have one. Your allowance resets in 12 h.
No credit card, no trial period, no sales call. Join with Google, GitHub, Facebook or email.
Free accounts get two operations per 12 hours, and yours reset in 12 h. Waiting costs you nothing and is a completely reasonable choice. If PDF work is a weekly job rather than a one-off, the counter is the wrong tool for you and Premium takes it off the page.
Cancel anytime. Everything you have already produced stays yours either way.