3FR to MPEG-2 Converter

3FR to MPEG-2: Set the Frame Size First

Read this before you convert, because the default settings do not work on this route. A .3fr is Hasselblad's 3F RAW — a medium-format capture whose long edge is thousands of pixels beyond anything MPEG-2 was designed for. Our pipeline normalises every image to a maximum long edge of 4096 pixels before encoding, and the MPEG-2 encoder specifically refuses any width or height that is an exact multiple of 4096. Leave "Video resolution" on "Keep original" with a Hasselblad file and those two rules collide: the encoder reports "Width or Height are not allowed to be multiples of 4096" and the job fails. Choose an explicit frame size in step 3 and the conversion runs normally.

How to Convert 3FR to MPEG-2

1. Upload Your 3FR File: Drag and drop the .3fr onto the page or click "+ Add Files". "Merge strategy" is preselected on "Merge images", so several captures become one clip; switch to "Video per image" for a separate file per capture.
2. Set the Image Duration: The dropdown runs from a single frame at 60 fps up to 10 seconds per image and starts at 5 seconds. Stills encode at one frame per second, so five seconds is a five-frame clip.
3. Choose an Explicit Video Resolution: This is the step that matters. Pick "Preset Resolutions" and a height such as 1080p or 480p, or type exact numbers into "Width x Height" — 1920 x 1080 for HD, 720 x 480 for NTSC DVD, 720 x 576 for PAL. Anything except "Keep original" avoids the 4096-pixel collision described above.
4. Convert and Download: Click "Convert" and take the .mpeg2 file. The video codec is fixed at MPEG-2 on this target and a photograph has no soundtrack, so there is no audio group and no audio in the output. No sign-up, no watermark.

Why "Keep original" Fails on a Hasselblad File

The normalisation step scales the long edge down to exactly 4096 and rounds both dimensions to even numbers. Every Hasselblad digital back is well past that, so every one of them lands on a 4096-pixel width — the one number the encoder will not take:

We reproduced this directly with the same encoder the conversion uses: a 4096 x 3072 frame produces a zero-byte file and the "multiples of 4096" error, while the identical frame at 3840 x 2880 or 1920 x 1080 encodes without complaint. It is a hard encoder rule, not a quality setting, and picking any explicit resolution steps around it completely.

All Three Compression Modes Land in the Same Place

MPEG-2 has no constant-rate-factor mode, and on an image source there is no incoming bitrate to scale, so the three options under "File Compression" all resolve to the same thing: a bitrate derived from the output frame size, which the encoder treats as its floor.

Those two file sizes are measured from real encodes of a detailed 1080p still under exactly those settings. They are far below what "7.5 Mbps for five seconds" would imply, and that is expected: only the first frame carries a full picture, and the four that follow it are nearly empty because nothing moves. The practical reading is that frame size is your only real control on this route — "Quality Preset", "Constant Quality" and "Constraint Quality" all end up at the same target, so set the resolution and leave the compression mode alone.

What MPEG-2 Costs a Medium-Format Capture

• The raw is developed and fixed. The 3FR is demosaiced with the white balance the camera recorded and written as 8-bit sRGB before encoding. Phocus, Lightroom and the rest can still reinterpret the original; the video cannot. Grade first if the look matters.
• The pixel loss is enormous. A 100-megapixel capture reduced to a 1080p video frame keeps roughly 2% of its pixels. That is arithmetic, not a setting — if resolution is the point, 3FR to TIFF keeps the full frame as a print master and 3FR to JPG gives you a viewable full-resolution picture.
• The clip is silent, always. There is no audio group on an image-to-video route and no audio track in the output.
• Nothing of the metadata travels. Camera body, lens, exposure and capture time live in the raw's EXIF; an MPEG-2 program stream has nowhere to keep them.

If your player refuses the finished file, the extension is the usual culprit rather than the encode — the output is a standard MPEG program stream carrying the .mpeg2 name. Rename it to .mpg, or use 3FR to MPEG, which writes the same stream under the .mpeg name and additionally offers MPEG-1 in the codec list.

Frequently Asked Questions

My conversion failed with an error about multiples of 4096. What do I do?

Set an explicit frame size. The image normalisation step caps the long edge at exactly 4096 pixels, and the MPEG-2 encoder rejects any dimension that is a multiple of 4096, so a Hasselblad capture at "Keep original" hits the one forbidden number every time. Choose "Preset Resolutions" and a height like 1080p, or enter a Width x Height such as 1920 x 1080 or 720 x 480, and the same file converts without trouble.

Why is MPEG-2 the only codec offered here?

Because the output extension is .mpeg2, and the whole reason to write one is that something downstream expects MPEG-2 elementary streams. The codec list on this target collapses to that single entry. If you want a choice, 3FR to MPEG adds MPEG-1, and 3FR to MP4 gives you H.264 in a container that plays on modern devices.

Which compression mode gives the best-looking frame?

They give the same frame. On a still-image source none of the three modes has anything to work from — MPEG-2 supports no CRF, and there is no incoming bitrate for a preset to scale — so all of them fall through to the bitrate the encoder derives from your chosen frame size. The way to improve the picture is to raise the resolution, not to change the compression mode.

Can I put a whole Hasselblad shoot into one clip?

Yes. Upload the captures together and leave "Merge strategy" on "Merge images"; each one is held for the Image Duration and they are concatenated in the order added. Thirty captures at four seconds each make a two-minute clip. Choose "Video per image" instead to get one .mpeg2 per capture. There are no transitions between images.

Is this the right way to get stills onto a DVD?

It produces the video stream a DVD authoring application needs, but not the disc. Enter 720 x 480 for NTSC or 720 x 576 for PAL under "Width x Height" — do not use a height preset, because that lets the width follow your 4:3 or 3:2 capture and it will not land on 720. The authoring application then builds the VIDEO_TS structure with its IFO and VOB files and burns the image.

How long does the clip run, and how many frames does it hold?

Image Duration multiplied by the number of images, at one frame per second. A single capture at the preselected five seconds is a five-frame, five-second clip. Sub-second durations exist in the dropdown for timelapse work — 1/24 second per image turns 240 captures into ten seconds — but for a single Hasselblad still the multi-second values are the sensible ones.

Will the colours match what Phocus shows me?

Not exactly. The converter develops the raw with the white balance recorded at capture and writes 8-bit sRGB, while Phocus applies Hasselblad's own colour rendering on top of the same sensor data. Expect a faithful, neutral development rather than a match for a specific look. When the rendering matters, export a TIFF or JPEG from Phocus and convert that file instead.

Why would anyone choose MPEG-2 over MP4 for a photograph?

Only for compatibility with equipment that predates modern codecs: DVD authoring software, DVB and ATSC ingest chains, older digital signage players, and broadcast workflows that specify MPEG-2 program streams. For every other purpose — sending the clip to someone, putting it on a website, playing it on a phone — 3FR to MP4 is smaller at the same quality and far better supported.

How are my files handled, and how long do you keep them?

Your capture is uploaded over an encrypted connection, developed and encoded on our servers, and deleted automatically a few hours after conversion — no sign-up, no watermark, never shared or made public. Hasselblad publishes an average of around 206 MB for a 100-megapixel 3FR, so the upload dominates the job: the encode of a five-second held frame takes moments by comparison.