AVCHD to MJPEG Converter

AVCHD to MJPEG: What This Tutorial Covers

AVCHD is the HD camcorder format Sony and Panasonic introduced jointly in 2006: H.264 video and Dolby AC-3 or linear PCM audio inside an MPEG-2 transport stream, written to card as .MTS under a BDMV/STREAM folder tree and usually renamed .M2TS after import. MJPEG is the opposite idea — every frame stored as an independent JPEG, with no prediction between frames at all.

A bare .mjpeg file is not a container. It is a run of JPEGs laid end to end, which has three consequences people meet as surprises: there is no audio track, there is no frame timing, and the file is very large. None of those is a setting you can change, because the format has nowhere to store any of them. What you get in exchange is frame independence — every frame decodes on its own, which is why editors, inspection tools and machine-vision pipelines still want it. This tutorial covers the two controls that exist, which of them actually moves the picture, and where the format runs out.

How to Convert AVCHD to MJPEG

1. Upload Your AVCHD File: Drag the .avchd onto the page or click "+ Add Files". Several clips can queue and run with the same settings.
2. Ignore the Video Codec Dropdown: It contains MJPEG and nothing else, because that is the only thing a raw MJPEG stream can hold. There is no decision here.
3. Switch File Compression to Constant Quality: This is the control that works. The slider runs 2 to 31 and opens at 5; lower means finer JPEG quantisation and a bigger file. The alternative mode, Quality Preset, is discussed below and is not what you want here.
4. Rescale or Trim, Then Convert: Video resolution opens on Keep original; on a bulky format, halving both dimensions is by far the most effective size lever. Trim opens on "Unchanged" and takes a Time Range. Click Convert. Files are uploaded over an encrypted connection, converted on our servers and deleted automatically after a few hours — no sign-up, no watermark, never shared or made public.

Walk-through: Which Control Actually Moves the Picture

Only two File Compression modes render on this page: Quality Preset and Constant Quality. The bitrate modes, Specific file size and Constraint Quality are all suppressed for an MJPEG target, so there is less to get wrong than on other video pages — but the two that remain do not do equal work.

MJPEG has no bitrate model. It compresses each frame with a JPEG quantiser, and the encoder sets one on every job whether or not you touched the slider. When a quantiser is in force, a bitrate target handed to the encoder alongside it is mostly ignored. We tested that directly on a 10-second 1280x720 clip: encoding at quantiser 5 with no rate cap and encoding at quantiser 5 with a 4 Mbps target and a 6 Mbps ceiling produced md5-identical files, 12,320,220 bytes both times. Only when the target was pushed absurdly low — 500 kbps with a 750 kbps ceiling — did the output change at all, dropping to 6,459,284 bytes while the rate controller logged buffer underflows all the way through, having missed its target by a factor of ten.

So the practical rule is: Constant Quality is the control; Quality Preset behaves as a loose ceiling that only bites at its bottom rungs, and then imprecisely. Set the quantiser and leave the preset alone.

The slider's span is also narrower than its numbers imply. The value you set is remapped onto the encoder's own 2-to-31 quantiser scale rather than passed through, so the top of the slider stops well short of the quantiser's maximum. Measured on the same 10-second 720p clip:

For scale, the same source encoded as H.264 in an MP4 — with its audio intact — came to 2.4 MB. Even the most compressed MJPEG the slider can reach is more than double that, which is the trade the format asks you to accept.

AVCHD Source and .mjpeg Output, Side by Side

The interlacing row deserves attention on this route in particular. A great deal of AVCHD footage is 1080i, and this converter does not deinterlace — there is no yadif or bwdif anywhere in the pipeline. On an ordinary video target the combing is at least hidden by motion; in MJPEG every frame becomes a still image, so the interlace comb is baked into a JPEG you will very likely be looking at frame by frame. If frame-accurate inspection is the reason you are here, deinterlace in an editor before converting.

Another consequence of the missing timing: we encoded a 30 fps source and read the result back, and the probe reported 25 fps with no duration at all. That is the reader guessing, not the file telling it. Anything where playback speed matters needs a container.

Common Errors and How to Fix Them

• "The .mjpeg has no sound" — expected, and not fixable here. The Audio Codec group draws a full list of codecs because the options panel is shared across every video target, but a raw MJPEG stream has nowhere to write an audio track and the AC-3 soundtrack is dropped regardless of what is selected. To keep the sound alongside JPEG frames, use AVCHD to MOV and set the Video Codec there to MJPEG — a real container holds both.
• "It plays at the wrong speed" — also expected. There are no timestamps in the file, so players assume a rate. Use a container if timing matters.
• "The file is enormous" — every frame stores the whole scene again, even when nothing moves, so a locked-off shot costs as much per frame as an action sequence. Reach for Video resolution and Trim before the quality slider; halving both dimensions is worth far more than sliding the quantiser.
• "There are comb lines through every frame" — interlaced AVCHD, and no deinterlacing is applied. Deinterlace in an editor first.
• "My phone, TV or browser will not open it" — those expect a container with headers and an index. VLC, ffmpeg-based tools and professional editors read raw MJPEG; consumer software does not.
• "My camera files are .MTS or .M2TS, not .avchd" — the uploader matches on the extension. Rename a copy to .avchd, or use the MTS or M2TS route directly.

Frequently Asked Questions

Why does the Audio Codec dropdown offer choices if the output is silent?

Because the options panel is a single shared layout drawn for every video target, and the audio group appears before the target's own limits take effect. On a container target — MOV, MKV, MP4 — those choices are real and the AC-3 from your camcorder is genuinely re-encoded into whatever you pick. On a raw .mjpeg stream there is no track to write into, so the selection has no effect on the file you download. Treat the group as leftover interface on this page.

Should I use Quality Preset or Constant Quality?

Constant Quality. MJPEG compresses with a quantiser rather than to a bitrate, and the quantiser is set on every job regardless. We measured a quantiser-5 encode with a 4 Mbps target coming out byte-for-byte identical to the same encode with no target at all, which is what "the bitrate is ignored" looks like in practice. Quality Preset only changes anything at its very lowest rungs, where the rate cap becomes tight enough to force the encoder to degrade — and even then it misses the target badly. Set the slider and be done.

Why does the slider say "CRF" when MJPEG has no CRF?

The label is shared across every codec the site supports and it is wrong here. MJPEG uses a JPEG quantiser scale from 2 to 31 where lower is better, and the slider maps onto it — but not one-to-one. We measured the mapping: slider 2 lands on quantiser 3, slider 5 on 4, slider 10 on 7, slider 20 on 13, and slider 31 on 19. The far end of the slider therefore stops short of the encoder's real maximum compression, which is why the total span is about 62% rather than the much larger range the numbers suggest.

Will the MJPEG be bigger than my AVCHD file?

Almost certainly, often by several times. AVCHD carries H.264 with interframe prediction, so a static part of a scene is stored once and referenced afterwards; MJPEG stores the entire frame again every single time. Camcorder footage tops out around 24 Mbps, and 1080p MJPEG at a sensible quantiser routinely exceeds that. The only reliable ways to bring it down are lowering the resolution and trimming to the section you need.

Why would I choose MJPEG over H.264 at all?

For frame independence, not efficiency. No frame depends on its neighbours, so an editor can cut on any frame without disturbing the ones around it, scrubbing is instant and exact, and a corrupted frame damages only itself instead of propagating down a prediction chain. That is why the format survives in editing intermediates, IP cameras, scientific capture and machine vision. For sharing, streaming or ordinary playback, H.264 wins on every axis and AVCHD to MP4 is the right conversion.

Can I get the frames out as individual JPEG files instead?

Not from this page — the output is one continuous stream, not a folder. If separate stills are what you actually want, AVCHD to JPG is the right route: it offers a Frame Selection group where "Multiple Screenshots" walks the timeline at a capture rate you choose and returns each frame as its own image. That is a much better fit for contact sheets, thumbnails and per-frame inspection than a raw stream you would have to demux yourself.

Does the conversion keep the original resolution?

Yes, unless you change it. Video resolution opens on "Keep original", so a 1920x1080 camcorder clip stays 1920x1080. Given how bulky MJPEG is, reducing it is usually the most effective thing you can do — but note that the Preset Resolutions list sets the height only and lets the width follow the source aspect. Use Width x Height when an exact frame size matters, and bear in mind that any rescale flattens the source's pixel aspect ratio to square, which matters for the 1440x1080 anamorphic mode some camcorders record.

Is a single .avchd file the same as an AVCHD disc structure?

No, and it is worth being clear about. Real AVCHD is a folder tree — clips in BDMV/STREAM, playlists and clip information alongside them — and camcorder software often expects that whole structure. What you upload here is a single stream file that was pulled out of one, and what you download is a flat MJPEG stream with even less structure. Nothing on this page reconstructs a disc layout in either direction.

What happens to my files after converting?

Your AVCHD clip is uploaded over an encrypted connection, decoded and re-encoded on our servers, and both it and the MJPEG output are deleted automatically after a few hours. Nothing is shared or made public and no account is required. Because MJPEG output is very large, the download over your own connection is usually the slow part of this job — trim to the section you need before converting rather than after.