AVCHD to EPS Converter

Convert AVCHD to EPS: What This Tutorial Covers

AVCHD is the HD camcorder format Sony and Panasonic announced in May 2006 — H.264/AVC video with Dolby AC-3 or linear PCM audio, carried in an MPEG-2 transport stream. EPS is Encapsulated PostScript, the file a print shop, a sign maker or an InDesign template asks for when it wants a graphic it can place on a page. This converter sits between them: it decodes one frame of your camcorder footage and embeds it in a PostScript wrapper.

Two things decide whether that is a good idea, and both are covered below. The picture inside the EPS is a raster, not vector artwork, and no converter can turn photographic pixels into paths. And a great deal of AVCHD footage is interlaced — 1080i is the default on most consumer camcorders — which shows up in a print-bound still in a way it never does on a television.

How to Convert AVCHD to EPS

1. Upload Your AVCHD File: Drag the .avchd file onto the page or click "+ Add Files" to browse. This page's uploader is scoped to the .avchd spelling; camcorder cards that write .MTS or .M2TS have their own pages. Several clips can be queued and share one set of settings.
2. Choose the Frame Under Frame Selection: Specific Frame is preselected with Time (seconds) on 0, which captures the opening frame — type 47.5 for the frame forty-seven and a half seconds in. Multiple Screenshots samples the whole recording instead, at a Capture Rate running from 0.1 seconds up to 10 seconds per frame.
3. Leave Image resolution on Keep original, or Scale It Down: Those are the only two choices on an EPS target — Keep original holds the camcorder's native frame size, and Resolution Percentage shrinks the embedded raster. There is no quality slider, no file-size target and no DPI setting here, for reasons the next section explains.
4. Convert and Download: Click "Convert" and save the .eps. Files upload over an encrypted connection, are processed on our servers, and are deleted automatically after a few hours. No sign-up, no watermark.

Walk-through: The Size and Scale of the EPS You Get

The raster inside an EPS produced here is stored uncompressed as ASCII-hex PostScript — Level 1, seven-bit clean, no image compression of any kind. That has two consequences worth knowing before you convert a stack of camcorder clips.

The first is file size. Every pixel becomes three bytes of colour, and every byte becomes two hex characters, so an EPS lands at a little over twice the raw pixel count. Measured on the same pipeline this page uses:

The second is placement. We write the EPS at one pixel to one PostScript point, so the bounding box is the pixel count expressed in points, and a point is 1/72 inch. Your layout program therefore opens a 1080p grab as a 26.7-inch-wide object at an effective 72 PPI — enormous and soft. That is not a defect and there is no DPI control to correct it, because effective resolution here is decided by the placement scale:

• Place at 100% — 72 PPI. Screen proof only.
• Place at 48% — 150 PPI. Newsprint, large-format work seen from a distance.
• Place at 24% — 300 PPI, the usual commercial-print target.

So a 1080p AVCHD frame will hold 300 PPI across about 6.4 inches, and no further. Lowering Resolution Percentage makes that worse, not better — use it only when the uncompressed file is too large to send, never to improve the print.

The Interlacing Problem, Which Is Specific to Camcorder Footage

Consumer AVCHD camcorders overwhelmingly record 1080i or 576i/480i: each video frame is two fields captured 1/50 or 1/60 second apart and woven together. A television or a video player separates them again. A still image cannot — the frame we decode contains both fields, so anything that moved between them appears as horizontal comb teeth along its edges.

This conversion applies no deinterlacing, so those teeth go straight into the EPS. On a screen at thumbnail size you may not notice; at 300 PPI on a printed page they are unmistakable. Three ways around it:

• Grab a still moment. A locked-off shot, or a pause between movements, has almost nothing different between the two fields and combs barely at all.
• Check before you commit. Sample with Multiple Screenshots, open the results at 100%, and pick the cleanest frame rather than the one at the timecode you first wrote down.
• Deinterlace elsewhere first. If the moment you need is a moving one, run the clip through a video tool that deinterlaces, export a progressive file, and grab the frame from that.

Progressive AVCHD — 24p, and the 1080/50p and 1080/60p modes added by the 2011 AVCHD Progressive revision — has no such problem, because each frame was captured whole.

How EPS Compares to the Other Print-Bound Targets

If nobody has specifically asked for .eps, AVCHD to TIFF hands a printer the same pixels in a far smaller file, and AVCHD to PNG is the right answer for anything that is not a print layout.

Frequently Asked Questions

Will the EPS be a vector file I can scale up?

No. A video frame is a grid of pixels, and what we produce is a PostScript file with those pixels embedded as a raster image — which is exactly what EPS was designed to allow. It places and prints correctly at its natural size, and enlarging it softens the picture just as enlarging any photo would. Real vectorisation means auto-tracing, which turns photographic content into flat blobs of colour; it is a design decision, not a file conversion.

Why did my 1080p grab import at 26 inches wide?

Because the bounding box is written at one pixel per PostScript point, and a point is 1/72 inch — so a 1920 x 1080 frame declares itself as 1920 x 1080 points, or 26.7 x 15 inches. Nothing is wrong with the file. Scale the placed graphic down in your layout to reach the resolution you need: 24% gives 300 PPI, 48% gives 150 PPI.

There is no DPI setting on this page — how do I get a 300 DPI file?

There is no separate DPI instruction in this output to set. Effective resolution equals 72 divided by the placement scale, which is why 24% works out to 300 PPI. If you need a specific printed width at 300 PPI, multiply that width in inches by 300 and check the source frame already has that many pixels: a 1080p camcorder gives you 1920, which covers 6.4 inches and no more.

Why does my still have horizontal comb lines through moving objects?

Because the footage is interlaced, which most consumer AVCHD is. The frame holds two fields captured a fiftieth or sixtieth of a second apart, and anything that moved between them lands on alternating scan lines. This conversion does no deinterlacing, so the pattern is carried into the still. Choose a moment with little movement, or deinterlace the clip in a video editor first and grab the frame from the progressive export.

Why is the EPS so much bigger than the AVCHD clip it came from?

Because the two files are doing opposite things. AVCHD uses H.264 at up to 24 Mbit/s — 28 Mbit/s in the 2011 Progressive revision — and predicts each frame from its neighbours, so a whole second of video costs a few megabytes. The EPS stores one frame with no compression at all, as hex text, which comes out at just over twice the raw pixel count. A 1080p frame is 12.6 MB whatever the picture shows.

Which frame do I get if I change nothing?

The frame at zero seconds, because Frame Selection is preselected on Specific Frame with Time (seconds) set to 0. On camcorder footage that opening frame is often the least useful one in the clip — a lens cap, an exposure still settling, or a hand on the record button. Type a timecode from the middle of the shot instead.

Can the EPS have a transparent background?

No. The PostScript imaging model this output is built on has no alpha channel, and a decoded video frame is fully opaque to begin with — AVCHD cannot store transparency. If you need a cut-out you will have to mask the image in an editor afterwards, working from a format that carries alpha such as AVCHD to PNG.

Does the EPS keep any of the camcorder's metadata?

No. Metadata is stripped during the conversion, so recording date, camera model and any embedded profile do not survive into the EPS. For handing artwork to a print supplier that is usually welcome. If the capture details need to travel with the image, use a raster format that stores them and keep the original clip as the record.

Why can I only scale by percentage, with no preset sizes?

Because the EPS target exposes a deliberately small control surface: Keep original and Resolution Percentage are the whole of it. There is no quality preset because the raster is uncompressed and there is nothing to trade, and no preset resolution list because the meaningful sizing decision happens at placement time in your layout rather than at conversion time here.

Should I be using EPS at all?

Only when someone has specifically asked for it. PDF took over most of the print-delivery role EPS used to hold, modern layout applications place PDF, PNG and TIFF at least as happily, and Microsoft disabled EPS insertion in Office on security grounds because PostScript is a programming language. Choose EPS when a printer, a journal template or a legacy LaTeX pipeline names it; otherwise take the same frame as AVCHD to TIFF for print or AVCHD to PNG for everything else.