Dangerous hacks – what’s the truth?

Sorry if I’m bringing this up again. But the whole ‘reverse engineering’ business has taken on an incredible momentum. Even Michael Guthmann, who was product manager at Olympus Europe and later at OMDS (see ‘Olympus Link’), warns on his blog against using, installing or whatever such hacks.

I’ve said before that I have a bit of an insight into this hacker scene. Every now and then I get updates on what the AI is churning out, and if you know where to look, you can find it on GitHub too.

Before the AI modifies the firmware, dozens of checks are carried out to see whether any jump addresses are being altered that might prevent the subsequent installation of earlier versions. So it’s no more likely that the hacked firmware will ‘brick’ the camera than it is that official firmware will – which does happen, after all. And since, so far, only existing functions have actually been unlocked, it’s also unlikely that anything will crop up at runtime – because, naturally, these functions have all already been tested by the manufacturer during the programming stage. They simply weren’t allowed to be unlocked for marketing reasons.

So, provided you don’t fall for some malicious hacker’s trick, the modified firmware is unlikely to be a problem.

So should you install every hack?

No.

Why not?

Things can always go wrong. Even when flashing via SD card. Every firmware modification carries a risk. Flashing via SD card is 99 per cent reliable if you use a good card and follow a clean procedure. But if you do it a hundred times, there’s still a chance you’ll brick the camera. As for the ‘This is not an authorised update path’ message – yeah, just ignore it. That’s just an excuse. Most cameras being flashed at the moment are out of warranty anyway, so what’s the big deal? Yeah – but if OMDS does release updates again, the modified firmware won’t work with them! Waaah. The likelihood of updates for older cameras is exactly zero. Zero. Yeah… back in the day, that used to scare gullible customers. Nowadays, nobody believes in Father Christmas anymore.

The manufacturer claims that the features can’t be unlocked because it would damage the hardware. On the one hand, that’s an argument that’s utterly ridiculous, because other cameras do offer those features. Or because a slight colour shift in the PEN-F’s colour mode certainly won’t damage any hardware. But on the other hand, it could actually cause problems. One example is the E-M1. The E-M1’s sensor was capable of filming in 4K. There was even a preproduction firmware in which this feature was enabled. The problem, however, was that whilst it did work, it only lasted a few minutes. After that, the sensor would overheat. So it’s only a matter of time before someone finds this routine in the E-M1 and enables it, and then someone else tries it out and fries their sensor. What a laugh.

With the E-PL5, the first version had an overheating problem – the camera would switch off after half an hour. In the first production version, this was fixed via a firmware update. If someone were to tinker with it now and disable the supposed ‘brake’ – they’d end up with a slightly faster camera – but in return, a veritable hand warmer.

Hasselblad has just put a stop to a hacker who was trying to hack the X2DII – to unlock video and so on. The ‘HRAP’ project. Hasselblad first sent the hacker a letter from their solicitors, then they held talks via ‘mediation’, after which all the source code disappeared from the internet; and in the meantime, even a post on Reddit reporting on the incident has been deleted. Why are the Swedes so sensitive about this? The X2DII is still on sale. If someone films with it and ends up frying the sensor in the process, you can imagine they’d be inundated with cameras needing repair. The camera isn’t cheap – and neither is the sensor.

And that’s the problem. There are functions that don’t damage the hardware. And there are functions that can. The AI that analyses the source data knows nothing about the camera’s thermal management. It knows nothing about the aperture mechanisms in the lenses. It knows nothing about linear, ultrasonic or stepper motors in the AF drives. And yes, I’ve seen such AI analyses. They’re brilliant – but sometimes just as daft as only AI analyses can be. You shouldn’t imagine it as simply telling the AI, ‘Have a look and see what you can get out of this thing’. You need to have a rough idea of what’s going on inside the camera and what might be in there before the AI can find it. It’s just a dumb machine, after all. And no, the AI has no clue about photography.

So what now? Install it or give it a miss?

I haven’t yet seen a single feature that would have persuaded me to take the risk – or even put in the effort – of doing the firmware upgrade. If I were to get a zoom frame for FHD on the OM-1, I might be tempted, but everything else I’ve seen so far isn’t worth it to me. Even if, as expected, it becomes possible to run LiveGND on the OM-1 – it would be an embarrassment for the manufacturer – I simply don’t need the feature.

One example of a feature I’d really like to have, but which isn’t included in the KI hacks, is de-fisheyeing the FT fisheye. Workspace can do it, but the camera can’t. Why? Does the camera read the lens parameters from the mFT fisheye, whilst the FT fisheye simply doesn’t provide these parameters yet? The AI can probably provide the answer – but it can’t do anything about it.

This ‘nothing-is-impoooossible’ approach of the AI hacks is fascinating, but it does have its limits. What I find really exciting is whether there might be some great features hidden away in the firmware versions that nobody knows about yet. But this is precisely where it gets tricky. Because the AI can detect that there’s a function that hasn’t been activated yet. This function does something to the image. However, it can’t determine what it does, what it looks like, or where it can be used. What if you find a feature that embeds an invisible digital watermark into all images, which the manufacturer can then track via the internet? Such features exist to identify unreliable individuals during the prototype phase who post images of pre-production models. If the feature is labelled ‘Secure’ in the source code, perhaps someone might get the idea to activate it.

So: have a look at who’s hacking, see if what comes out of it offers any added value for personal photography, and then don’t go and install the very first alpha version.

Firmware hacks aren’t toys, and they’re not something to show off about. “I’ve got Alpha 3, mate! Totally cool, eh!”

Take photos. That’s the point.

Cover image: An old Star Trek pinball machine from 1979. These things now go for just under 5,000 euros. Why a pinball machine as the cover image? I think that’s self-explanatory.

2 Replies to “Dangerous hacks – what’s the truth?”

  1. Danke für den interessanten Artikel.

    Dass man mit einem FW-Hack auch Schäden an einem Objektiv provozieren könnte, haben wohl die wenigsten auf dem Radar. Ich stelle mir aber die Frage: Könnte das sogar mit einer hausgemachten FW passieren? Im konkreten Fall: Besteht ein erhöhtes Risiko, dass native Zuiko FT-Optiken mit Schrittmotoren durch die “schnelleren” AF-Algorithmen in den aktuellen Top-FT-Bodies (OM-1, OM-3) Schaden nehmen können? Garde wenn der AF-Motor in extrem kurzen Zeitintervallen wesentlich mehr (kleine) Schrittbewegung in beide Richtungen durchführen soll, als dies bei deren Konstruktion vorgesehen war? Oder gibt es in den Optiken einen Prozessor, der bereits mit den FT-Boliden (E-3, E-5) am Anschlag war und deren Langsamkeit dafür sorgen, dass ein Teil der Befehle erst gar nicht ausgeführt werden?

    1. Nein. Gerade die aktuellen Bodys behandeln die FTs schonender. Warum? Die alten Kontast-AF.-Bodys (PEN und OM-D 5 und OM-D10) stressen die alten FTs durch die vielen Mikroschritte. Ab OM-1 sprechen die Kameras mit den FTs wieder “phasisch”. Das bedeutet, sie geben einen ziemlich genauen Befehl “Fahre auf 17,30 Meter” und justieren nur ab da nach, falls notwendig. Das schont die Motoren. Die E-M1II/III/X hatten noch keinen Phasen-AF- drin, der so exakt war, dass das Objektiv mit einem Befehl ziemlich genau ans Ziel kam. (Deshalb werden schnellere Prozessoren und Sensoren – und Firmwarehacks – die FTs nicht mehr wesentlich beschleunigen. Die arbeiten bereits an der Grenze, was die Motoren hergeben.)

Leave a Reply

Your email address will not be published. Required fields are marked *

The maximum upload file size: 2 MB. You can upload: image. Drop file here