I've been away for a while 😀
zilath
Hello guys,
a firmware version with microfocus adjustment will be a great thing.
In the meanwhile you can try focus adjustment with...a scewdriver.
BAD idea. I read about this thing too and apparently it's used for something completely different. I don't recall the exact details but it could completely screw over the entire AF system so you're going to have to send it back to Canon. If I recall correctly it's some sort of mirror adjustment screw which causes the light to reach the AF sensor under a different angle and, if misaligned, might work for one AF point but very likely offsets the measurement on all other AF points. I think it also had the side effect of having the focus points in the viewfinder point to something completely different than what the AF unit is looking at.
Anyway, I think the screw has been (re)moved in the 40D to make this kind of tampering impossible but still... don't try it.
gerard
Why not give what you have and let other people work on it ?
I can't just publish all my work somewhere, since the firmware is still Canon property. Also, my notes are of limited use to anyone but me and anyone trying to continue my work will have to do the majority of research all over again anyway, just because you have to be familiar with the firmware in order to understand it. As for the practical part: the method is largely the same as on any powershot camera, except that there is no ready codebase to work with. Anyone with a decent understanding of the CHDK boot process, task hook and function layout will be able to produce the necessary code in no-time. I'd be happy to provide some 'simple' stuff like memory addresses but even I have barely scratched any surface and frankly, given the number of camera features, heck, even considering only the basic ones I have barely scratched anything. The code is vast and it's almost immediately clear that the Powershot/IXUS and EOS teams are not the same. I have no idea how any of the camera's shooting functions work, how the keyboard works (so far I managed to get two buttons going), what state the camera is in and so on.
Then there's the part where all other developers (with 40D, important!) failed to create anything that hooks into the main OS. If that's too hard, there is really not much I can do but sharing my code won't help in any way. I think it's about 10-20 kB and that's including a whole lot of crap, test-code, comments and so on.
And even if what you've already done is hard to use, if you've successfully published a video I'm sure this could interest a lot of people...
It was only after publishing this video that I learned of the technical restrictions. I don't remember if I already explained them here but basically the limit is 30 seconds for 1024x680 MJPEG and 4GB for 320x214 YUV/MJPEG. Even at 320x214 I haven't yet been able to do this without frame loss. (I also didn't work on it for the last couple of months)
gerard
[..]
I doubt it because it's been some months now since the first "proof of concept" and nothing new has been said... He told us he had to concentrate on his exams and other stuff, but come on ! So much time and nothing, he either has no interest whatsoever with this, or either has nothing to release and all of this would be a fake (not saying he is a liar, I surely hope not)...
Sad as it may sound, I feel like I have indeed lost my interest in this issue. The technical limits are impossible to work around and because I'm not sure whether or not I'm allowed to release it, I didn't feel like seriously working on movie mode anymore. By the way, you're right about that I have nothing to release: everything is hardcoded and works without user interaction, so that's not really easy to use. It also requires a PC to convert the resulting data into a movie file.
Apart from that, I really don't have too much time anymore. The reason I had a lot of time last year was because I failed a couple of courses the year before and as a result of that I suddenly had this 6-month vacation. That's over and I have to spend a lot of time studying again. Besides that I took on several other things so all in all, that leaves very little time to work on this.
I'd be happy to show you that it's real, you can come over here and shoot a movie yourself.
If he really has no time, I don't understand why he doesn't release his stuff indeed... Why wouldn't he ??? Would he want to sell his stuff ? I don't see any advantage of keeping everything secret, do you ? Open source software surely works, and if everybody was keeping every bit of code secret, CHDK would not exist I think...
Selling the code is my goal indeed. I got a lucrative offer from Nikon, they are currently paying my salary. I started this research as soon as the 40D was launched and my work is basically the foundation for the movie mode in the D90. Don't tell Canon, though, I could end up in jail!
I already explained why I'm keeping it 'secret', as you like to call it. I'd be happy to act as a (very limited) consultant for someone who shows he is capable enough to get something going. It's also very easy to misjudge the required amount of time if you have no experience with software development, reverse engineering or anything remotely related to that. As I said, I have barely no knowledge of how the camera works, though I estimate that I spent at least 200 hours doing RE, coding, testing, documenting and so on. At $30 an hour (no idea if that's reasonable for someone with a similar skillset) that works out to $6000. I could have used that time to do non-free work and buy myself a 5D2 with some really nice lenses.
gerard
I just wanted to say that we shouldn't expect too much of this development because it was already a long time since the anouncement...
If anything goes out sometimes, I'll be very very very happy. If not... well too bad but at least I don't want to get my hopes too high.
That's more or less what I said, if I recall correctly. I can never guarantee that I'll be able to work on spare-time projects.