3G2 to BMP Converter

3G2 to BMP — Pulling Frames Out of a CDMA Phone Clip

A .3g2 file is the 3GPP2 container that CDMA2000 handsets recorded to, built on the same ISO base media file format as MP4 and specified by 3GPP2 in C.S0050. A .bmp is a Windows device-independent bitmap: a header, an optional palette and a run of pixels, with no modern compression anywhere in it. This page takes a frame out of the clip and writes it as a BMP.

Two things make this conversion less obvious than it looks. First, you choose which frame — a single still at a timestamp, or a whole run of stills at a capture rate. Second, the "Quality Preset" list on a BMP target does not mean what it means on a JPEG target: each rung selects a genuinely different BMP writer, producing a different bit depth, a different header version and a different compression scheme. The section below spells out what each one writes. This route accepts .3gp uploads as well as .3g2.

How to Convert 3G2 to BMP

1. Upload Your 3G2 File: Drag it onto the page or click "+ Add Files". .3gp is accepted here too, and several files can queue together.
2. Choose the Frame with Frame Selection: Specific Frame is the pre-selected mode and takes a Time (seconds) value, which starts at 0 — the very first frame. Multiple Screenshots instead walks the whole clip at a Capture Rate, which offers everything from a frame every 0.1 seconds up to one every 10 seconds and lands on "1 second per frame".
3. Pick a Quality Preset — Deliberately: The list on a BMP target contains High, Medium and Low only, and each one is a different writer rather than a different quality number. Pick one explicitly rather than assuming a default is in force; the table below says what each produces.
4. Set Image resolution and Convert: Image resolution opens on Keep original. You can scale by percentage, take a preset height, or set an exact Width x Height. Click Convert. Files are uploaded over an encrypted connection, processed on our servers and deleted automatically after a few hours — no sign-up, no watermark, never shared or made public.

What Each Quality Preset Actually Writes

We ran a 320x240 frame through all three rungs and read the resulting BMP headers back byte by byte. They are three different files in three different formats:

A few things follow from that table that are easy to get wrong:

• None of the three writes a plain 24-bit uncompressed BMP. If a tool insists on 24-bit BI_RGB, no rung here produces it; open the High output in an editor and re-save.
• "High" is not automatically the most faithful. It stores colour directly, but only 5 bits per channel, so smooth gradients band. A dithered 256-colour Medium file can measure closer to the source on gradient-heavy frames, while High wins clearly on frames with flat blocks of colour and text. Which one looks better genuinely depends on the frame.
• "Low" is the least portable. The 124-byte V5 header is a Windows 98-era structure; simple and embedded BMP readers that only parse the classic 40-byte header will reject it, even though it is a smaller file. If compatibility with an old tool is the reason you wanted a BMP at all, use High.
• Size is not a reliable ranking. On dithered detail, run-length coding expands rather than compresses, which is why Low can come out larger than Medium on some frames.

Choosing the Frame

The frame itself is extracted losslessly before anything else happens: the converter pulls it out of the clip as a PNG first, then hands that PNG to the BMP writer. So the only quality decisions that matter are the ones in the table above and whatever rescaling you apply.

Common Errors and How to Fix Them

• "The image is tiny" — that is the source, not the conversion. Handsets of the CDMA era recorded QCIF (176x144) and QVGA (320x240); "Keep original" gives you exactly those pixels. Upscaling under Image resolution enlarges the file without adding any detail.
• "The BMP is enormous" — the High preset stores two bytes for every pixel with no compression at all, so the size is fixed by the frame dimensions. A 320x240 frame costs 153,654 bytes and a 1920x1080 frame would cost 4,147,254. Use Medium, reduce the resolution, or grab the frame as 3G2 to JPG if a compressed still would do.
• "My old software will not open the Low output" — it is written with a 124-byte BITMAPV5HEADER rather than the classic 40-byte header. Use High or Medium, both of which write the older structure.
• "The colours look grainy or speckled" — that is Floyd-Steinberg dithering on the Medium and Low presets, which scatters the quantisation error across neighbouring pixels to disguise banding. It is normal and it is how those rungs reach 256 and 86 colours. High does not dither because it does not use a palette.
• "There are comb lines across the picture" — the source was interlaced and this converter does not deinterlace, so the combing is captured in the frame. Grab from a progressive source, or deinterlace in an editor.
• "I wanted a whole folder of frames and got one image" — Frame Selection was left on Specific Frame. Switch to Multiple Screenshots and set a Capture Rate.

Frequently Asked Questions

Which preset should I choose if I just want the frame to look right?

High, in most cases. It is the only rung that does not reduce the picture to a palette, so text stays crisp and there is no dithering speckle. Its weakness is smooth gradients — 5 bits per channel is 32 levels, which bands visibly on skies and soft lighting. If your frame is mostly gradient and the banding bothers you, try Medium: the dithered 256-colour output measured closer to the original than High did on a smooth test gradient, and it is roughly five times smaller.

Why is the BMP so much bigger than my whole video clip?

Because BMP does not really compress and video does, aggressively. A 3G2 stores each frame as a difference from its neighbours at a few hundred kilobits per second; a High-preset BMP stores two full bytes for every pixel with no compression whatsoever. The arithmetic is exact and easy to check: width x height x 2, plus 54 bytes of header. That is the price of a format designed in 1990 for direct copying into video memory.

Is the frame extracted losslessly, or is it re-encoded first?

The extraction itself is lossless — the frame is decoded out of the clip and written as a PNG before the BMP writer ever sees it, so nothing is thrown away at that step. The losses that do occur are the ones you choose: 5-bit-per-channel colour on High, or palette reduction on Medium and Low. What you cannot undo is the compression the phone applied when it recorded, which is baked into the frame you are extracting.

Can I get every frame of the clip as a separate BMP?

Not every frame, but as many as you like. Switch Frame Selection to Multiple Screenshots and pick a Capture Rate — the fastest option is a frame every 0.1 seconds, which is 10 stills per second of footage. Bear in mind what that means for storage on an uncompressed format: ten seconds of QVGA at High is a hundred images of about 154 KB each, so around 15 MB for ten seconds. Use Medium if you are sampling densely.

Does the output keep the clip's timestamp or any camera metadata?

No, and there are two reasons. BMP has no metadata container to begin with — no EXIF block, no timestamp field, no colour profile in the classic 40-byte header — and the Medium preset strips metadata explicitly as part of its palette conversion. Beyond that, a frame grab is a picture pulled out of a video stream rather than a photograph, so there is no camera EXIF attached to it in the first place. No image target on the site can invent one; if you need to know when a clip was shot, read that from the original .3g2 before converting.

Why would I choose BMP over PNG for a frame grab?

Almost only for compatibility with something old or something simple. PNG is losslessly compressed, keeps full 24-bit colour, supports transparency and is a fraction of the size, so 3G2 to PNG is the better answer for any general purpose. BMP earns its place when a piece of legacy Windows software, an embedded device, a game-engine importer or a hardware tool reads nothing else — those readers usually want the plainest possible file, which is what the High preset comes closest to.

What is the difference between the two resolution preset lists?

There is one list here, Preset Resolutions, and it sets the height while the width follows the source's aspect ratio. That is why it also carries small entries like 256P, 128P and 16P, which are aimed at icon-sized output. If you need an exact pixel size, use Width x Height and type both numbers; if you just want a proportional reduction, Resolution Percentage is simpler.

My file is .3gp rather than .3g2 — will this work?

Yes. Both extensions are accepted on this page, because 3GP and 3G2 are the GSM and CDMA variants of the same ISO base media container and carry the same video codecs. The frame extraction is identical either way. The differences between the two formats are in the speech codecs and a couple of AAC profiles, none of which matter when you are pulling out a picture.

How long do you keep my file?

Your clip is uploaded over an encrypted connection, the frame is extracted and converted on our servers, and both the upload and the BMP are deleted automatically after a few hours. Nothing is shared or made public and no account is required. Phone clips from this era are small, so the whole job is usually finished as fast as the upload completes.