MTS to PNG Converter

Convert MTS to PNG Online

Grab a still frame out of AVCHD camcorder footage and save it as a PNG. An .mts file holds MPEG-4 AVC/H.264 video that photo apps will not open, so the only way to get a picture out of it is to decode a frame and write it to an image format. We do that in two passes: ffmpeg seeks to the timestamp you choose and writes the frame as a lossless PNG, then the image stage applies your resolution and quality settings.

One thing to know before you start, because it is the opposite of what most people assume: PNG output here is not lossless by default. The pre-selected quality preset runs the frame through a palette quantiser. If you are pulling frames for editing, compositing or archival, change it — the table below says exactly which setting to pick.

How to Convert MTS to PNG

1. Upload Your MTS File: Drag the .mts clip onto the page or click "Add Files" to browse. Several clips can queue and each returns its own still.
2. Set Frame Selection and the Timestamp: Frame Selection opens on "Specific Frame" with Time (seconds) at 0. Type the moment you actually want — decimals work, so 4.5 is four and a half seconds in. Switch to "Multiple Screenshots" and set a Capture Rate to sample the whole clip instead.
3. Set the Quality Preset and Image Resolution: Pick "Highest" under Quality Preset for a true-colour lossless PNG, or leave the pre-selected "Very High (Recommended)" for a smaller palette-quantised file. Image resolution opens on "Keep original"; Preset Resolutions, Width, Height and Width x Height are there if you need to scale.
4. Convert and Download: Click Convert and save the PNG. Files upload over an encrypted connection, are processed on our servers, and are deleted automatically after a few hours — no sign-up, no watermark.

Which Quality Preset Is Actually Lossless

We traced the encoder path for this page rather than assuming, because the answer is not obvious from the labels. Only one preset skips the quantiser:

pngquant is a lossy compressor by design — it generates a palette using vector quantisation and can cut a PNG by 70% or so, which is exactly what you want for a web thumbnail and exactly what you do not want for a master still. The Colors group gives you direct control over the same mechanism: leave it on Original, or switch to "By Color Reduction + Dither" and choose a palette size from 256 down to 2, with dithering on or off.

Frame Selection: One Still or a Whole Sequence

Leaving Time (seconds) at 0 is the most common mistake on camcorder footage. The opening moment of an AVCHD clip is frequently the lens still hunting for focus, an exposure ramp, or a black lead-in before the sensor settles — so frame 0 is rarely the frame you meant. Scrub the clip in any player, read off the timestamp, and type that number instead.

What the Conversion Fixes, and What It Doesn't

AVCHD is interlaced in most camera modes; the official format specification lists 1080i alongside the progressive modes, and on many camcorders 1080i is the only option. A frame woven from two fields captured a moment apart shows comb teeth on anything that moved between them. We do not deinterlace at any point in this pipeline, so those comb lines land in your PNG exactly as they exist in the stream. No preset removes them, and neither does routing the clip through MP4 first. The workable answers are to choose a timestamp where the subject is nearly still, or to deinterlace the clip in a desktop editor before extracting.

The same honesty applies to shape. Some AVCHD modes record 1440x1080 with non-square pixels that a player stretches to 16:9 on screen. Our resize step normalises the pixel aspect ratio to square, so if you scale one of those frames with a percentage or a height preset the picture comes out narrow. Use Width x Height and give both numbers — 1920 x 1080 for a 1440x1080 source — and the frame ends up the shape you saw on the camera.

Frequently Asked Questions

Is the PNG I get actually lossless?

Only if you set Quality Preset to "Highest". That is the single setting that writes a true-colour PNG straight from the decoded frame. Every other preset, including the pre-selected "Very High (Recommended)", runs pngquant, which reduces the image to an indexed palette — genuinely lossy, even though PNG is a lossless container format. The distinction catches people out because "PNG" and "lossless" are usually treated as synonyms, and here the format is lossless while the encoder pass in front of it is not.

Why is frame 0 black or blurry on my camcorder clip?

Because the first frame of an AVCHD recording is often the camera still settling. Autofocus hunts, the iris ramps, and some models write a short dark lead-in before the exposure stabilises. Frame Selection opens with Time (seconds) at 0, so if you press Convert without touching it you get that unusable opening frame. Enter a timestamp a second or two in and the problem disappears.

Can I get rid of the horizontal comb lines on moving subjects?

Not on this page. There is no deinterlacing step anywhere in the conversion, so an interlaced 1080i frame arrives with its combing intact and no quality preset, resolution setting or intermediate format changes that. Pick a moment with less motion — a pause between gestures rather than mid-pan — and the two fields align closely enough that the teeth mostly vanish. If you need clean edges on genuinely fast motion, deinterlace the clip in a video editor first and pull the still from the progressive result.

Why did my 1440x1080 frame come out looking narrow?

Because that AVCHD mode stores non-square pixels. The stored frame is 1440 wide but is meant to be displayed at 1920 — the player does the stretching. Our resize step writes square pixels and does not carry the original pixel aspect ratio, so scaling by percentage or by a height preset gives you the unstretched, too-narrow picture. Choose "Width x Height" in Image resolution and type both dimensions explicitly (1920 and 1080) to get the proportions the camera intended.

Should I use PNG or JPG for a camcorder still?

PNG at the "Highest" preset when the frame is going into an editor, a composite or an archive, because nothing is discarded and you keep a clean starting point for further work. MTS to JPG when the still is going into a message, a document or a web page, because a JPG of the same frame is a fraction of the size and the difference is invisible at viewing scale. If you want PNG's file structure but small files, the palette presets on this page get you most of the way there.

Does the extracted PNG have a transparent background?

No. PNG supports an alpha channel and pngquant preserves alpha when it is present, but a frame decoded from camcorder video is fully opaque — video formats cannot store transparency in the first place, so there is nothing to carry over. The alpha channel is simply available and unused, which is useful later if you cut the subject out in an image editor.

My file is called .m2ts and this page rejects it — why?

Because the page is keyed to the .mts extension. AVCHD camcorders write .mts on the card, and import software (Sony's included) renames the identical stream to .m2ts when it copies to a computer. The bytes are the same but the extension gate is not, so the .m2ts copy is refused here and an .mts file is refused on the .m2ts page. The video to PNG extractor accepts both spellings and runs the same extraction.

How large will the PNG be, and can I make it smaller?

A true-colour 1920x1080 still at the "Highest" preset typically runs to several megabytes because nothing is thrown away — that is the cost of lossless. Three levers shrink it: drop to the "Very High" or lower presets to engage the palette quantiser, set the Colors group to a smaller palette such as 128 or 64, or scale the frame down under Image resolution. There is also a Compression level control from 1 to 10, pre-selected at 6, which trades encoding time for a few more bytes saved without touching image quality.