Original OM-5: pre-alpha for testers

Pre-alpha. This has never run on any camera. The Mark II build it is made from has not run on a Mark II either, and three earlier Mark II builds left the camera in a boot loop. Nobody knows whether an original OM-5 can be brought back from the card if this fails. Read all of this page, and Safety, before you start.

Hold off (15 November 2026). A fourth Mark II build (alpha5) boot-looped. The likely cause is the length of a compressed block in the update file (note). Pre-alpha 1 was made before that check existed and its block lengths have not been checked. Do not flash it; wait for a rebuilt pre-alpha.

This page is for owners of the original OM-5 (2022, firmware 1.2) who are comfortable with a terminal, accept the risk to their camera, and want to help: by trying the first steps below and reporting what happens, good or bad, or by improving the builder. If you only want to use Caduceus, wait for a tested release.

What it is

The original OM-5 runs an older program that has no Color Profiles to restore. So this build does not patch it. It takes the whole OM-5 Mark II program with Caduceus and fits it into the original OM-5's update file, with the original OM-5's file layout and its own controller firmware. If it works, the camera should behave like an OM-5 Mark II running Caduceus: the Mark II's menus, looks, HSL, tone controls and Live ND64.

Whether it can work depends on hardware the two cameras may not share: sensor and image processor, stabiliser, charging and USB, and above all the separate controller chip. The original OM-5's controller firmware is kept, and it is a different build from the Mark II's.

What is known

Builder checks Both OM firmware files are checked against OM's published checksums. Every Caduceus code range, all 557 OM-5 resource records and all 116 pictures are preserved; the pictures are recompressed losslessly to fit, pixel-checked. The OM-5's own firmware is written alongside, byte for byte, to go back.
Emulator, OM-5 model Runs OM's own OM-5 firmware, but cannot run Mark II code at all: no result.
Emulator, Mark II model (same chip family) Starts, all 13 look checks pass, 23 screens identical to the Mark II build. It does not use the original OM-5's resource layout or controller, so this says little about an OM-5.
Comparing the two OM programs About 86 % of functions match. Differences found: renumbered identifiers, one moved variable in on-chip memory, a different controller firmware. Details.
Real original OM-5 Nothing. You would be the first.

What can go wrong

No responsibility is accepted for any damage or loss.

1. Make the file

You need Linux, macOS, or Windows with WSL, plus Python 3.10 or newer, git and a C compiler. About 30 minutes, most of it building the assembler once.

  1. The project, and Pillow:

bash git clone https://github.com/xulfr/caduceus-om5.git

bash cd caduceus-om5 git switch --detach 02c3cb2a64687e3aa74b0639550c82a95cfd66d7

This revision pins the Mark II alpha2 donor used by the published pre-alpha 1 checksum.

bash python3 -m pip install pillow

  1. jpegtran from libjpeg-turbo. Debian/Ubuntu/WSL: sudo apt install libjpeg-turbo-progs. macOS (Homebrew): brew install jpeg-turbo. Check that jpegtran -version works.

  2. The assembler: GNU binutils 2.42 for mn10300-elf, built once into ~/mn10300:

bash curl -LO https://ftp.gnu.org/gnu/binutils/binutils-2.42.tar.xz && tar xf binutils-2.42.tar.xz

bash mkdir binutils-build && cd binutils-build && ../binutils-2.42/configure --target=mn10300-elf --prefix="$HOME/mn10300" --disable-nls --disable-werror --disable-gdb --disable-sim --disable-gprof --disable-gprofng && make -j4 MAKEINFO=true && make install MAKEINFO=true && cd ..

bash export PATH="$HOME/mn10300/bin:$PATH"

  1. OM's two firmware files, from OM's own server (about 180 MB; each is checked):

bash python3 tools/fetch_stock.py E144 E148

  1. Build. The output folder must not exist yet:

bash python3 patches/ports/om-5-i-from-ii/build.py --stock OLY_E_144_1200_0000_0000.BIN --donor-stock OLY_E_148_1100_0000_0000.BIN --out generated/om5-prealpha1 --write-untested-package

  1. Check the result. generated/om5-prealpha1/build.json must show

"package_sha256": "70788992f032dad53bfa4e145b2147e376d3f25a62e2b1ce2929973fec26d1b4"

If it shows anything else, stop and report it (a different jpegtran version is the likely cause). Do not flash a file with another checksum.

You get:

generated/om5-prealpha1/install/DCOLYMP/E1449999.BIN the pre-alpha
generated/om5-prealpha1/restore-stock/DCOLYMP/E1449999.BIN OM's own firmware 1.2, to go back. Keep it.

These files contain OM's firmware: keep them to yourself, do not share or upload them.

2. Test, in this order

Use a fully charged battery and the card in slot 1. The update method is the same as for the Mark II: camera off, hold OK, switch on, let go when the orange light is steady, do not touch anything until it blinks (Install, step 3). Delete DCOLYMP from the card after each update.

Step Do Tells us
0 Put the restore DCOLYMP folder on a card and update with it. This is OM's own firmware 1.2, so nothing changes. That the card update works on your camera, before anything risky. If it does not, stop here.
1 Prepare a second card with the restore DCOLYMP folder and keep it at hand. Then update with the install file. Switch on, touch nothing for a minute. Whether it starts at all: the biggest unknown.
2 If it started: note Menu › Setup › Firmware, open the menus and the super control panel, take a photo, play it back. Whether the basics work.
3 Half-press once, then Picture Mode: look for Color Profile 1–3; try the colour wheel and the trash button there. Whether Caduceus works.
4 With LookLab's LOOKS.BIN in OM5MOD: looks, as in Using it. Switch off and on several times. Looks, saving, restart.
5 Check what you can of stabilisation, charging over USB, battery level, autofocus, and video. The hardware unknowns.

At the first thing that is wrong, stop, write it down, and go back (below). A photo of the screen helps.

Going back

Update from the card with restore-stock/DCOLYMP/E1449999.BIN, the same way. If the camera does not start or does not take the update, try again from fully off, with a charged battery. If it still will not take it, the camera needs OM service.

Report

Please open an issue whatever happened, including "it did not start" and "I stopped at step 0". Give the checksum you built, how far you got, and what you saw. A failure report is as useful as a success: this page changes only with reports from real cameras.