Messages

Ok, I use a cheap 5-in-1 universal IR remote that can be found on ebay and elsewhere.
Face detection auto focus with that one and my 600D works fine.
(It's nice to have the swivel screen of the 600D for self portraits).

Use an IR remote if your camera supports it.
They often have a 2 second timer button which is convenient. Use live view with face detection if your camera supports it.

me too!

Yes there are surely many ways to combine the images, both manually and automatically.
I haven't tried to use photoshops photomerge function for it.

For now, I personally like doing it manually, it gives the control I want.

Good to hear that you are aware of the issue and that you have already tried several ways to fix it.
(I hadn't seen the other thread so I didn't know that you were already on it).

I'm not a developer and haven't really got a clue, but yes, it sounds like the separate wav function is taking up too much processing power, and that is what is causing the dropouts (which in turn leads to the out of sync issue). So yes, perhaps a more efficient buffer handling could make it work better, I guess the circular buffer technique could be worth trying if that's not too much of a hassle.
Have you tried 22kHz or even 11 just to see if that makes a difference?
Or could the dropouts be caused by disturbance of other processes, I mean, like some other process suddenly gets higher priority and steals the cycles. You don't have to answer, just thinking out loud here.

I did some more testing now -
Filming a clock, I came to the conclusion that ML does indeed achieve the correct frame rate. The video is running at the right speed.

It is the audio that is somehow running too fast. About one second per minute.
(Well, either it's running too fast or there are microskips that causes it to end up being too short)

(this thread should perhaps be moved to bug reports)

There are no skips or soundglitches in the recorded material as far as I can tell. (Don't know if there's a way to check that more scientifically).
And no skips during playback of the re-encoded video.

I was hesitating where to place this, since I don't know if is a bug in magic lantern or if the problem lies elsewhere.

I have been trying to record 720p at 30fps with sound record enabled (WAV).
Using Avidemux I change the video to 30fps and add the audio track. However, after less than 30 seconds the tracks are already noticeably out of sync.
I can only get them to sync decently by changing the framerate to somewhere about 30.6 fps.

Anyone else encountered this problem?
Does magic lantern not really record at exactly 29.97 fps as it states, or is it something else that causes this?

That's right. A combination of one shot wide open with one at the lens' sharpest aperture.
A lens capable of really short DOF will give the most worthwhile result.

Thank you.
I agree, the exposure values aren't totally intuitive when you're used to just choosing the aperture value. But as you say, 1 exposure step is 1 aperture step - I usually use an increment of either 3 or 4 EV.

I use the method to get an image with the subject at maximum sharpness and yet with maximum amount of background blur.

The mushroom in the first picture isn't very sharp because the lens isn't very sharp at f1.8, even when the focus point is spot on.

As you probably know, almost no lens is optimally sharp at its widest aperture. Not only are they not optimally sharp, defects such as chromatic aberration are also more pronounced.
Neither are they optimally sharp at very small apertures such as f22 or more, due to diffraction. (They do of course give the farthest DOF, but that is not really important in this case).
Most lenses are sharpest somewhere around f5.6 - f8.0 (as a rule of thumb, this may be debated forever).

Taking the shot at f2.8 as you suggest would surely be a nice compromise, and useful for most situations, however it would neither give most background blur nor most sharpness.

Therefore I find the best way to achieve this is to merge a wide open shot with one at the lens' sharpest aperture.

You can see the effect more clearly here:
http://www.magiclantern.fm/forum/index.php?topic=3113.0

Edit: By the way I didn't mean to sound so lecturing, I just tried to explain it as clearly as possible to anyone reading this who may wonder the same thing as you.

Thanks, good to hear!

Here you can see the two original pictures, as well as the layer mask:


The method I use for combining the pictures is fairly quick and straightforward, and gives a lot of control:
I open the two pictures in photoshop, copy and paste the sharp one as a layer on top of the blurry one. (If you were shooting on a tripod the pictures will align perfectly, otherwise you may have to move the layer around a bit until they match).
Create a layer mask. Paint it black (zero opacity).
Using the ordinary airbrush tool I paint in the parts that I want sharp.

A special thanks to you again a1ex for implementing this feature!

Trying out the new aperture/DOF bracketing feature that can be found in the latest Magic Lantern nightly.

I use it to shoot two consecutive pictures at different aperture - in this case f1.8 and f7.1 - and later combine them in order to get a picture with shallow DOF and yet with the subject at optimum sharpness.



(view at 100% to be able to appreciate the effect)

forum thread: http://www.magiclantern.fm/forum/index.php?topic=3045.0

Ah, yes, 0++
That makes it work exactly like I want it too. Thanks for pointing that out!

This is great - I'm very impressed and thankful that you implemented this feature so quickly.
I have tried it out now and it works very well.
Thanks a1ex, good job!
Thanks also to nanomad for the nightly.

Now, this may seem like nitpicking but it might be preferable in this case if you could choose the sequence " - 0 " . For example, when shooting a portrait it would be better if the shot with the shortest shutter time would be taken first, to minimize the risk of the model moving between the shots. A minor detail for sure, but worth considering.

engardeknave: That's right, however:

b4rt: That's a correct observation. You have to use your artistic sense. - sometimes it's better to leave the edges from the blurry image to have a smooth transition, and sometimes it works very well to mask a few pixels outside of the edge. In that case, you will get a part of the sharp image where it 'shouldn't be', but oftentimes that is not visible.
I find it quite easy to get good results quickly.
The merging is of course easiest with pictures such as portraits where you have a single sharp object against a blurred background.

yes, exposure should be maintained.
here's an example (view at 100%):


as you can see - at f1.8 the background blur is nice, but even the subject isn't very sharp.
at f5.6 sharpness is good, but background is very busy.
combined, you get the best of both worlds.

discocalculi: nice too see your experimentation. helicon focus and zerene stacker is something I might look into. however, for this purpose (stacking only two images), I find doing it manually works well and gives complete control of the result.

discocalculi: I usually just stack them manually in photoshop.
I align the sharp image as a layer on top of the one with the short DOF, and then using layer mask I paint out the sharp areas.

a1ex: wow, that was fast! I'm impressed, and appreciate your quick reply.
however - I'm afraid I haven't got a clue how to compile from source. I think I'm gonna have to wait and hope that it ends up in the next release - that would be so great.

I would find aperture bracketing extremely useful.
I often shoot two consecutive pictures in Av mode, one with f1.8 and the next with f5.6. I later combine them, putting the sharp part of the f5.6 picture on the f1.8 one.
That way you can get images with optimum sharpness and yet with a nice soft background.
(The two pictures are of course both correctly exposed, unlike HDR bracketing)

The problem with doing it manually - that is changing the aperture using the scroll wheel, is of course both that it takes time between the shots and also that you inevitable move the camera a bit, making it more difficult to align the pictures.