Initializing... drag & drop files here
Supports: MXF
MXF is the SMPTE container professional cameras and playout systems write, and it is deliberately large: broadcast essence is coded for editing headroom, not for delivery. HEVC (H.265) is the opposite — a delivery codec that stores the same picture in far fewer bits. This page re-encodes the first into the second, and is precise about the one thing most converters gloss over: what a file named .hevc actually contains.
| Property | Value |
|---|---|
| Full name | Material Exchange Format |
| Standard | SMPTE ST 377-1 (current edition ST 377-1:2019) |
| First released | 22 September 2004 |
| Operational patterns | OP1a (SMPTE 378M, self-contained), OP-Atom (SMPTE 390M, one essence per file) |
| Common video essence | MPEG-2 MP@H14/HL long GOP at 25–35 Mbit/s (XDCAM HD), MPEG-2 422P@HL at 50 Mbit/s (XDCAM HD422), AVC-Intra, DNxHD/DNxHR |
| Common audio essence | Uncompressed PCM, mapped per SMPTE 382M (AES/EBU and Broadcast Wave) |
| Audio layout | Frequently several discrete tracks, not one stereo mix |
| Used by | Sony XDCAM, Panasonic DVCPRO P2, Avid systems, broadcast playout and archive |
| Consumer playback | Essentially none — MXF is an interchange and mastering format |
Calling a file .hevc normally implies a raw H.265 elementary stream with no container and no audio. That is not what comes out of here, and it matters for playback.
| Property | Value |
|---|---|
| Container written | Matroska (our converter maps the HEVC target to ffmpeg's matroska muxer) |
| Video codec | H.265 / HEVC — the Video Codec list on this page contains only these two entries |
| Audio | Preserved and re-encoded; AAC is the matched default, with AC3, DTS (DCA), EAC3, FLAC, MP2, MP3, Opus and Vorbis also offered |
| File extension | .hevc, which no player associates with Matroska |
| Plays as-is in | VLC, mpv and other content-sniffing players |
| Needs a rename in | QuickTime, Windows Photos and anything that dispatches on file extension — rename to .mkv |
| Standard | H.265 approved by ITU-T on 13 April 2013, published June 2013; ISO/IEC edition November 2013 |
| Efficiency vs H.264 | Roughly 25–50% smaller at matched quality |
.mxf onto the page or use "Add Files". Several clips can be queued, and they all convert with the settings you set once.A Matroska container holding an H.265 video track and an audio track — not a raw elementary stream. Our converter routes the HEVC target through ffmpeg's Matroska muxer, so the bytes are .mkv bytes under an .hevc name. VLC and mpv sniff the content and play it directly; QuickTime and Windows Photos dispatch on the extension and will refuse it. Renaming the download to .mkv fixes that everywhere, and if you would rather skip the rename, MXF to MKV writes the same container with the expected name.
One of them, chosen automatically. There is no track picker on this page and our command builder emits no -map for MXF input, so ffmpeg's own automatic stream selection applies: per the ffmpeg documentation it picks "the stream with the most channels" for audio, and "in the case where several streams of the same type rate equally, the stream with the lowest index is chosen." We confirmed that on a three-stream MXF carrying mono, mono and stereo essence — the stereo pair was the one that survived. If your mix is spread across discrete mono tracks and you need a specific pairing, bounce a stereo mixdown from your NLE before uploading.
The preset maps to a codec-specific CRF, then gets a small resolution adjustment (up to ±2, based on how the output size compares with 1080p). At 1080p the mapping is Highest → 15, Very High → 17, High → 20, Medium → 26, Low → 30, Very Low → 32, Lowest → 36. A 2160p output shifts those up by 2, because artefacts hide better at higher pixel density. If you want an exact number instead, switch the mode to Constant Quality and type the CRF you want.
No. The server always decodes and re-encodes — there is no stream-copy path, so the H.265 output is a new encode regardless of what the MXF was carrying. That is fine for review, delivery and archival browsing, and wrong for finishing: keep the MXF as the master if you still have grading, conform or VFX work to do, because every re-encode is another generation.
Not automatically. There is no deinterlacing anywhere in this pipeline, so interlaced essence keeps its combing baked into the H.265 output — deinterlace in your NLE first if the destination is progressive. There is also no HDR tone-mapping: an HLG or PQ master gets a plain colour conversion and can look washed out or dull on an SDR screen. Grade to Rec. 709 upstream when the target audience is SDR.
Not on its own. The pipeline normalises the sample aspect ratio to square pixels without stretching the frame to compensate, so non-square-pixel sources come out geometrically wrong. The workaround is explicit: pick Width x Height under Video resolution and enter the display dimensions you want (for example 1920 × 1080 for a 1440 × 1080 anamorphic HD master). Preset Resolutions will not do it for you, because that control sets height only and lets width follow the stored pixel grid.
Because the source was never trying to be small. XDCAM HD422 runs at 50 Mbit/s and XDCAM HD long GOP at 25–35 Mbit/s, and AVC-Intra and DNxHD sit higher still, all of it chosen for edit-friendliness. HEVC is a delivery codec with roughly 25–50% better compression than H.264 at matched quality, so a delivery-grade CRF lands one or two orders of magnitude below a broadcast bitrate. The output is smaller, not better — H.265 cannot restore detail the camera codec never recorded.
Target HEVC when size is the constraint and you know the playback device is modern; Wikipedia's tally puts HEVC playback at roughly 92% of desktop and mobile browsers as of late 2025, which is high but not universal, and patent licensing kept it off some platforms for years. Target H.264 with MXF to MP4 when the file has to open on anything — a client's old laptop, a station's review machine, a phone — because H.264 support is the closest thing to a floor in video. HEVC encoding is also slower, which is noticeable on a multi-gigabyte camera master.
Your MXF uploads over an encrypted connection, converts on our servers and is deleted automatically after a few hours, along with the output. There is no sign-up, no watermark, and files are never shared or made public. The practical constraint on a multi-gigabyte camera master is upload size and time on your connection rather than anything after that — if the clip is long and you only need part of it, set Trim → Time Range before converting, or use trim MXF to cut inside the original container first.