Messages

This one doesn't use ML features directly, but it was made a lot easier with ML's help. Shooting IR panoramas can be a bit tricky when you see almost nothing through the viewfinder, but with FPS override and high gain, I could expose very precisely and also align the images without too much estimating.


>bigger image<

EOS 50D | Porst 28mm/2,8 @ f/5,6 | IR720 Filter
7 images stitched with Hugin.

Hey there... wanted to post my first video here. This is just a small snippet to show another awesome possibility with ML - no artistic notions intended.

This was made with a 50D and IR720 filter at 1FPS. There was no postprocessing except a bit of sharpening and contrast. The colors are straight out of the camera, thanks to UV-swap and carefully set white balance.

As I said, this is just the first test... I'll try my luck at a few longer sequences soon.

The "busy" screen is not an error and with a 15 second interval, it can't be the card either. Even the slowest card could handle that.

The only other thing I can think of right now is the long exposure noise reduction... For long exposures (over 1sec, I think), the camera takes a dark frame with the same exposure time to reduce hot pixels. The display shows "busy" while doing that. It can be turned off in the Canon Custom Functions. With my 50D it is under C.Fn.II (Image) and I think it's the same with other cameras, too.

Note that disabling the long exposure noise reduction will increase noise and hot pixels. If you want to avoid that, your only choice is to use a longer interval or shorter exposures.

Do you have, by any chance, activated FPS override? If you lower the frame rate that way, it will directly affect liveview. If you set 1FPS, the screen will only change once every second which can make focussing very hard. Since these very low FPS values are mostly used for timelapse with a tripod, the best way to avoid it is focussing first and activating FPS override after everything is set.

If you don't use FPS override, I have no idea, what could cause such severe lag. The processor should be able to handle all functions of ML without slowing down that much.

I always try the nightlies whenever I have the time for it. Those new features need to be tested and every found bug can help to make ML better and shorten the time until the next full release (though I didn't find anything that wasn't already reported so far)
For regular use, I keep a card with v2.3, though. I can't afford to fiddle with untested features when work needs to get done. Just imagine the camera locking up or doing unexpected stuff when working with models or during an event where every second counts... absolute horror.

Yes, that's how it works. The motion detection needs some getting used to and careful calibration, but once you got that, it works like a charm.
And 8GB is plenty for your needs. More is usually only needed for extended intervalometer use or high quality video. The battery will probably be empty before the card is full.

Thank you very  much, ax.
I've been waiting a long time for this, but never got the hang of making my own picture styles. A little color shift or altered contrast is not that hard, but something extreme like this is beyond my skills.

Yes, you can use any framerate between 0,25 (that's one frame every four seconds) and 30. It's called FPS override in the movie menu. Higher frame rates are not possible.
When viewing the video on your computer, it will always play at 30 FPS, so a video recorded at 24 FPS will play a little faster while 0,25 FPS gives you a true timelapse. To view it at the right speed, you have to load it into a video editing software and change the framerate there.

For you understanding: ML is a piece of software that happens on the card. If you insert a card with ML (and the proper bootflag) on it after all the installation steps from the manual, it will run ML, but if you insert a card without it, the camera works like usual (the ML menu is not available). Please note that starting the camera with a card that has the bootflag set but no autoexec.bin on it can cause problems. It's not something you'd usually do, but something to keep in mind.

Temporarily disabling ML works by either using a card without it or pressing SET while switching the camera on. Both only disable ML until you restart the camera regularly with a card with ML inserted.

Formatting a card in camera with ML is working since several versions ago. Just use the regular "format card" from the Canon menu and it will delete all data except the ML files. To be precise, it will temporarily copy the ML files to camera memory, format the card and then put ML back onto it. All your ML settings will be saved as well.

As said before, ML does not load, if it's not on the card. But if you want to remove all traces that it ever was there (sending the camera in for repairs comes to mind), you can simply download a Canon firmware update (I think, their official site has them for downloading) and follow their instructions for installing it. After re-installing the official firmware, the camera's software is in the same state as you got it from the store.

I think, paxaerm meant the focus confirmation that works with EF-lenses in manual focus. So far, the only way to make it work is an adapter with a chip that makes the camera believe it is an autofocus lens.
Making this possible through software has been suggested many times. I don't exactly remember the reasons, but it is not possible and very unlikely to ever happen. I think it was because the lens communication doesn't use the main processor but some other chip that can't be adressed by ML.

That someone is not right (or at least, it was a misunderstanding). After the firmware upgrade process, any card with ML will work automatically without any further steps - just insert the card and start the camera.

No, that's not possible. The only thing you can do is overexpose by 1 stop and then set the exposure to -1 during RAW-developement. Note that you'll lose some dynamic range this way and won't get any less noise. It's basically the same thing that's done in-camera during video recording with 50ISO

It's not yet possible and I doubt it ever will be. Making a video out of JPG files isn't that hard anyway. Just make sure they're sequentially numbered (irfan view batch rename does this in seconds for a few thousand pics), pull the first picture into virtual dub (the rest are inserted automatically), do any editing you may need and export the video with your preferred compression.

Please state, what exactly you're doing and how exactly it doesn't work and the version you're using. I got it working on my 50D with last year's stable versions, 2.3 and several nightly builds, so it's probably not a bug.

One of the most common errors is having the lens set to MF. It may seem logical, because you are focussing manually, but the camera needs the AF motor to focus, so it needs to be active.

If you mean setting speeds like 1/15 with 25FPS, that's just not possible because the laws of time and space don't work that way That's got nothing to do with ML, because there's just no way to squeeze more than 8 exposures of 1/8sec into a second of movie.

If you meant shorter exposures (like 1/100 with 25FPS), it should work automatically. For full manual exposure you need to activate "exposure override" from the expo sub-menu and then set the exposure from there. Note that the "cinematic look" with proper motion blur is best achieved with shutter speed of half the FPS, like 1/50sec for 25FPS.

FPS override is somewhat limited. It's just not possible to go under 0,2FPS. For star timelapse and similar effects, there's no other way than stressing the shutter some more. Of course you could raise the ISO (also with ML digital gain) until speeds of about 5sec are enough, but I guess you don't want that much noise. For lower FPS you can use silent picture and intervalometer, but that doesn't change anything about the shutter speed and the resolution is even lower than full HD video. It's better to record low-FPS video and speed it up some more in post-processing.

Getting the desired shutter speed in video mode is done with "exposure override" from the expo menu and then setting the shutter speed from there. It gives you full manual exposure, so you need to carefully check the histograms for a proper exposure. At least that's how it's done with my 50D, so I guess it's the same for a 5D.

As for the timers in FPS override... only the devs know how that works. I have no idea and usually just select the setting that's closest to what I need for the current situation.

You can't watch the recorded video in the camera. There's no other way than downloading it to a computer.

Setting manual exposure in video works with "exposure override" from the "expo" sub-menu. Activate it and then set your desired shutter speed from there.

Since I use a 50D, I use it mainly for video. When I bought the camera, I didn't think I'd need video, but now that I have the possibility, I use it quite regularly. My favourite part is FPS override for timelapse. I also like the expanded ISO settings (ML digital gain)
The second most useful feature are the LV overlays. Having zebras and spotmeter in LV is very useful, even for still pictures. Focus Peaking can also be useful with manual lenses. Before ML, I almost never used liveview, but now I prefer it in all situations where I have the time to think about proper exposure.

You need to use display gain/ML digital ISO an addition to the highest possible Canon ISO. It can be set from -2 to +7 EV. With ISO 100 and -2 EV, you can use 25 ISO, though it tends to make the highlights a very ugly pink. ML digital ISO only works in liveview (for video and silent picture). For still pictures, it is pretty useless since it does the same as over- or underexposing raw pictures in post-processing.

Sorry, this wasn't supposed to go against you in a personal way. Actually, I don't disagree at all. If you don't read manuals but find out how to do everything by yourself, that's ok. It's not my style, but everybody has their own methods.
I meant that someone (like the guys at canonrumors.com) who has massive problems with ML should have read the instructions first and checked each setting before complaining about it. Many people have been using ML and had no problems of this kind. If such errors occur, it's probably just the user's error because he was to lazy to learn proper usage of ML.
And even if there was no user error, reporting the bug here and letting the experienced developers solve it instead of complaining in an unrelated forum, can actually help making the firmware better and more stable in the future. 

That's the way it is... ML is just not made for all people, or rather there are some people who aren't made for ML. When you're using a professional gas stove, you're not automatically a better cook, but you might set your kitchen on fire, if you're not careful. A microwave oven may be a better choice for you.

I read the whole documentation and every bit of information I could get before installing ML on my 50D. I only enable certain functions when I really need them and if anything doesn't work (rare, but possible), I know that it's my own error in 99% of cases. I won't complain or file a bug report until I'm 100% sure that it's the camera's of firmware's fault. If everyone did it that way, there'd be very, very few jobs in customer support.
Some people seem to think, reading manuals is for noobs. I think, it's pretty much the opposite: If you're not reading the manual, you'll never use the full potential of your gear. You may still be a good photographer, but always limited by your ignorance.
Canon firmware helps to avoid most errors by limiting the user. The majority of camera users just want to press the buttons and let the camera do the rest and it works well for them. ML takes away many of the limits and therefore some of the idiot-proofing. It's definitely not magic, as the name might imply, but a very powerful tool for people who reached the limits of Canon firmware and want more.
</rant>


/tl&dr: If you're too dumb to use it properly, stay away from ML. Also, RTFM.

Also FPS-override:
+ very fluent motion (well, that's a matter of taste)
- hot pixels and noise after a while, even with low ISO

I often use intervalometer and silent picture together with my 50D. The resolution is very low (still good enough for web video and many projectors), but it allows you to use any interval and is easier on the battery since ML can turn LV off between shots. Several hours of timelapse (taking pictures every minute) are no problem without killing the shutter mechanism. I guess noise is also less of a problem, though I didn't make a direct comparison yet. Since FPS override also affects silent pictures, shutter speeds to 5sec are still possible and that's enough for most situations except night shots far from the city. For my timelapse shooting of urban life, I find it the best solution.

Why not just pin a note to your tripod or camera bag? I'd say this feature is somewhat unnecessary and I'd prefer if the devs put their valuable time into bugfixing and new features that actually enhance the camera functionality.

A simple .txt file is probably the easiest solution. Or a bitmap image... which gives me an idea: What about a cropmark with custom text on it? Just a suggestion for those that need additional on-screen notes.

I use silent picture for timelapses. Sure, the resolution isn't great but since it's goin to be a video anyway, I don't really need 15MPx. With ML automatically turning off the display between shots, the battery will last a very long time.
Using video and FPS override is the easiest way. 0,2FPS  (one picture every 5 seconds) gives a rather nice effect and if you want to speed it up even more, you can still do so afterwards by increasing framerate or dropping frames. Again, the energy saving options can help to keep the camera running for a few hours as long as you've got enough memory. The biggest disadvantage is the sensor heating up and producing a lot of noise and hot pixels after a while. Those can probably be removed in post processing (with a black frame, similar to Canon's long exposure noise reduction), but I don't have any experience with that.

Taking full resolution pictures without using the shutter is just not possible. I think, it's on the list of "things that are probably never done" or even "impossible"

Sounds impossible to me, because you'd need very fast specialized buffers for all that writing an deleting at the same time. I'm pretty sure that current Canon DSLRs don't even have the hardware for this kind of task. Writing and simultaneously deleting that many images at full resolution would also require much faster memory access which is just not possible. I guess, Casio built special hardware to accomplish this.
The best solution for now is simply recording a video in 1080p and highest possible framerate with fastest possible shutter (video shutter speeds like 1/15 or 1/30 rarely make good still images) and pick out the best pictures afterwards. 1920x1080 is actually good enough for newspaper prints, but not much more.