Messages

You post IMHO exemplifies a key misunderstanding and gap between the guys who are coding ML and the guys who are trying to use it.

Speaking as a guy who uses ML and doesn't code it, your posts seem to exemplify a key misunderstanding between someone who thinks he is buying a product from a commercial entity who needs his money to continue operating and someone who realizes that other people have tinkered with something in their spare time and who, out of kindness and a desire to collaborate with others, are now allowing him to download the results of their tinkering.  If you don't evangelize or download anything from ML, the project will still do just fine.

But make no mistake, sheer power without end-user friendliness is doomed from the start. So many programmers believe that just implementing amazing code in all that is needed. This is the critical difference that made Steve Jobs so successful, he understood that the end-user experience is as inherent to a product's success as the product itself.
Don't get me wrong, I don't believe in the need to appease the absolute lowest common denominator, but in the case of ML the situation is very much at the opposite end of that spectrum.

I'll help you with a more apt comparison.
Many years ago, a group of people developed an OS kernel known as Mach.  It was a microkernel which attempted to move big chunks of the kernel into user space (among other things, but that'll be a lot of reading - if you're interested, wikipedia has a nice write-up http://en.wikipedia.org/wiki/Mach_(kernel)#Mach_concepts).  It was designed as a drop-in replacement for the traditional Unix kernel.
When Steve Jobs left Apple for NeXT, they needed an operating system and after looking around for a bit, they settled on using a Mach kernel with a userland heavily based on BSD, a former academic OS derived from AT&T unix which was developed largely, as the name would imply, at Berkeley, with descendants still available today (FreeBSD, NetBSD, OpenBSD, etc).  At NeXT, they developed a number of graphical abstractions on top of Unix.
When Apple acquired NeXT to bring Steve Jobs back, he brought NeXTStep with him and it was retooled to become one of the early versions of OSX.

In this example, Magic Lantern would probably be most similar to the BSD userland or Mach microkernel.  They're powerful tools in the right hands and thanks to their inclusion in OSX, they are running on most desktop computers in the world.

BSD was not built as a commercial product.  It was built by hackers who wanted something to make their own lives better and who were kind enough to share it with others to make their lives better too.  Thanks to the work of those wonderful hippie nerds in the 70's, you may enjoy a Mac today which isn't running a derivative of OS9.

But the reality of the situation is that ML exists on borrowed time and in about two years' time nobody will want to shoot video on DSLR's because there will be absolutely no reason to do so (and that's me being optimistic). There are other massive disruptive forces that are causing a rapid shift in how people shoot video. I live and breathe this industry on a daily basis in a town that fancies itself the World capital of film production. I am constantly interacting with cinematographers and editors, and believe me, when I show them a command-line solution they laugh in my face at the absurdity of it. So yes this is only my opinion, but an opinion shared by just about every single professional that I've interacted with for the better part of last year.

Two years, huh?  Looking at the progress that has been made by Canon with their cameras in the last 2 years, I can TOTALLY see why you'd extrapolate that in 2 more years, they'd be obsoleting the current crop.  They just released... what, like the 5th consumer camera in a row using almost exactly the same body as the last 4 and pretty much the same sensor as the 7D had when it was released in 2009?  Their least expensive raw-capable video camera is in the 5 digits price range now and 2 years ago, it was in the... 5 digits range.  Every camera being released by a major vendor now is still using some form of h.264/avchd/mp4 video, the same as 2 years ago.  Canon's already mentioned how many new camera bodies they're releasing in 2014.  It's not a large number.  So...  whose cameras, exactly, do you think are going to unseat the major vendors?  Black Magic are small and feature-poor still.  Digital Bolex finally shipped something, but their longer-term prospects seem iffy at best.

I appreciate your time and passion, I really do; and I sincerely hope that my posts aren't coming across as too snarky or negative...because in truth if we could hang and discuss this over a couple of drinks I'm sure we'd come to like one another! 

Sure.  I'm sure the developers would love to sit down for drinks with some whiny Internet user who rants at length about everything he thinks they're doing wrong and how difficult he finds their software.  Who wouldn't want that?

How do you guys manage to format the card while still preserve the ML?
I used to do on 5d mark ll. but on 5d mark lll , seem like there are no such option.

Thank you in advance.

DD2020

There's a handy workaround on the 5DIII.  Put ML on an SD card.  Then you don't have to worry about formatting your CF.  Also, all of your preferences will be sticky between swapping cards.

On 5D3 & 5D3 you have Uncompressed HDMI Output Support - however you will need a raw recorder like the Sony AXS-R5 - $5350 or the Convergent Design Odyssey7q (10 bit Canon raw)  for $2295.

Or the Black Magic Hyperdeck Shuttle 2 at less than $400.
http://www.amazon.com/Blackmagic-Design-HyperDeck-Shuttle-Recorder/dp/B00701AN5Y/ref=sr_1_1?ie=UTF8&qid=1379862428&sr=8-1&keywords=hyperdeck+shuttle+2

Though the signal coming from HDMI is still 8-bit.  Raw, in this case, doesn't buy you much.

Even though you can't play back smoothly in camera, if you have a decent GPU in your laptop, you can use Davinci Resolve to play back the files in nearly realtime after shooting.

I'm pretty sure that 1.2.1 does not allow for 10-bit recording via HDMI.  It allows 8-bit 4:2:2.  Your recorder may be recording with 10 bits, but 2 of them aren't being used.

Excuse me! Is this DIY project useful for this?

http://www.davidhunt.ie/?p=2641

It's basically a raspberry pi which communicates with the camera via USB.. and can download the material on realtime to a hard drive.. I don't know if the USB speeds are that high.. but I think it's an interesting project.. the author managed to do lots of stuff with it.. maybe even could open a wide range of possibilites for magic lantern! (i'm just wondering lol)..

The theoretical maximum speed for USB 2 is 480 megabits/second (60MB/s) and even with the best controllers, that throughput is rarely seen in real life.  Not likely to be useful at all in this case.

Hi all,
I would love to hear some input from people who tried both cameras - I love the 5D3 and the beautiful RAW it delivers. Never tried the BMCC. Is the image quality from the 5D3 RAW better/equal to the BMCC RAW ? Dynamic range? Low light performance? Any other important differences between the two that will make me buy/not buy the BMCC? I am only doing wildlife filmmaking with telephoto lenses from 70mm - 800mm.
Thanks,
Ofer

I haven't used both, but the fact that they just dropped the price on the BMCC by $1,000 may make it an attractive option for you.

What is "fast enough?" 

500GB is probably plenty for a field dump/backup solution, though at this point I'd be surprised if someone weren't making one with 1T of capacity or more (but yeah, definitely look for one with good reviews). 
Even at 4GB / minute for raw, 500GB is around 120 minutes of footage.  How much do you plan to shoot in a day?

Without putting actuations and mechanical wear on the shutter? That's the reason I'm trying to do it in video mode, cause those actuations are adding up fast.

Silent shutter doesn't fire the mechanical shutter (unless I am mistaken).  Look under the photo menu in ML.
Just read the manual - guess it won't do full res.  Nevermind (though if you can tolerate less than 1fps, the high-res mode may be worth investigating).

http://magiclantern.wikia.com/wiki/Unified/UserGuide#silent-pictures

Shooting in 3,5K for a 1080p timelapse, the final framings and pans will be chosen or adjusted in post anyway.
Thanks for the tip, if this works I'm blown away.

If I could only get past that 1/30 exposure limit now, I could use this for night sky timelapses too, but for now that shutter-wear will have to be covered by my semi-retired older EOS body.

Edit: I see, in 3.5K the aspect snaps to 2.72 (Kinopanorama). And since 1320px is max vertical the only way to get a taller aspect is cropping the width of the picture. Is there any way to get more than 1320 lines? Hmm need wider glass. 

If your goal is to have the highest resolution possible at 1fps, why not just use a combination of the intervalometer and silent shutter mode?  Even with a relatively slow/moderate card, you should be able to pull off 5760 × 3840 (and your wide angle glass will stay wide angle). 

Right now I have a full frame with a 40mm 2.8 STM (my widest lens) for video.

I am looking to buy a WIDER lens for video and photography, but am stuck between:
17-40mm F4
35mm F1.4

Which one should I choose? I know that the wider the lens, the greater the depth of field which is good for videos. But the 35mm is just so good for photography...

You should rent each one for a week and try it to see if it suits your style of shooting.  Nobody can tell you which lens you'll like more.

I do the usual, RAW2DNG, Camera Raw, move to AE as image sequence, and export.

AE seems to do it frame by frame, and since I am recording at 1280x720 at 60FPS, it takes hella long for only a 30 second clip... I'm pretty sure its not my computer (i7 2600k, 560TI GPU). Anyone have advice for optimizing/minimizing time for the post-processing process to getting the videos ready for timeline?

I plan to have many "clips" and moving them to AE individually to render would take quite a while.

Have you told AE to use multiple threads to render more than one frame at once?  It's still slow, but that will help, at least.

What's the benefit of using Resolve over Lightroom?

Resolve has good GPU support and is able to play back raw footage in real-time even on my older core2quad system since I have a decent graphics card in it.  It also exports a lot faster in my experience than AE or LR. 
Though it's also a lot harder to use than ACR or the LR adjustments.  After poking at it for a few hours, I'm still at caveman level where I was pretty good at LR in about the same amount of time.  I'll probably end up watching some tutorials!

I strongly suggest reading the forum and finding the several posts with similar ideas and their reception before creating another one with the same idea.

I was wondering. Raw video at 14bit takes up and extreme amount of disk space.
Technicaly, Log can get about 95% of the information and about 10% disk space.
So, would it make sense to capture instead of a stream of .png 14bit images from the camera, something like a .jpeg 12bit log (cinestyle for example) stream of images?
That should have some pretty nice results with a lot less disk space.


To what point would that be possible and not only reduce disk space usage but also allow slower cards and possibly faster framerates.


Has anyone thought of that or tried to develop it?
Do you guys think it would make sense?

I know nothing about programming and the insides of ML but if someone would be interested in trying/developing this with me I think it could go somewhere.

Thanks

Well, I don,t know what you actually mean by that but if people already have a card and manage to record 24fps on their 5d mark 3, it,s actually not that hard to set the camera to 25fps and do a little testing. A little help

What I mean is that you seem to be yelling at someone else for not having time to test it.  If you're that upset/anxious, you should buy one and test it.

Still no 25fps 1920x1080 tryout? Come onHow hard can it be??

Sounds like a volunteer to buy one and test since it's so easy!

Hi evrey body

We knows how setting video mode by DSLR camera
For exsmpel like freams with shutter:

If freams 24 - shutter 48
If freams 30 - shutter 60 , to other

Also aperture we knows when we can open aperture 4, 6, 12 ....

But my question how setting magic lantern RAW video by 5d mark 3
Because i am looking vrey complicated

How setting shutter , iso, aperture and what bottem nassery

If you have videos for this pls answer me

Thanks

With the hack enabled, you set shutter speed, aperture, and ISO the same way as you always did.
On the Mark III, assuming you're in "M" mode, you set the aperture with the back wheel, the shutter speed with the front wheel, and the ISO by clicking the "ISO" button atop the camera and using the front wheel to select the desired one.

I think you're posting too much and not reading enough.  Read all of the threads on camera/card performance and understand buffering, etc.  Then consider posting again for any follow-up questions.  Questions like this one demonstrate a profound lack of understanding which can only come from lack of having read the rest of the threads/topics.

I was one of the very first to have a 5D2 and suffered from the line skipping.
But on the 5D mark III it should work - not line skipping or pixel binning AFAIK.

He already explained it to you. 
In full-frame mode, the maximum resolution is 1920x1280.  In windowed (5x/crop) mode, it's 3584x1320, but it's unlikely that recording the maximum resolution will ever be possible because the card writer in the camera is limited to 160-170MB/s and that size requires a bit more than that (190 MB/s).

Also, if it were somehow possible to dump the entire resolution of the sensor in the 5DIII, it'd be 5760x3240.  That comes to 261,273,600 bits per frame (32,659,200 bytes).  At 24 frames per second, that's 783,820,800 bytes per second.  For your theory that ML is writing every single pixel on the sensor out to the card, ML would be writing about 750MB/s.

Read the rest of the posts on this forum and you will understand better.  Right now, you're making suggestions with no understanding of what you're talking about.  Even worse, you're making suggestions that make no sense.

Crop mode recording is always 1:1 on the sensor regardless of the zoom unless I'm mistaken.  You'd only be changing the preview size.

So Redrocks and EatTooMuchJam, I should reinstall the latest build?

I have no opinion on what you should or shouldn't install. 

With that said, development is moving fast.  It's usually worth staying up-to-date.

I'm a little stomp but does that mean its an exact RAW video conversion from the sensor measurement, I think it's because Canon non-compromise for low-level access to the sensor that Magic Lantern cannot push the 1880 up.

Does lowering it to 1872 RAW measurements takes from conversions like 1080p?

Forgive my ignorance but I believe asking make you a little least ignorant.

When a1ex says 1872 is a mod16 resolution, it means that 1872 is evenly-divisible by 16 (1872 / 16 = 117).  In terms of upscaling 1872 x 1053 vs 1880 x 1057.5 to 1920x1080, you could say you see a difference, but you'd probably be lying or fooling yourself.  It's dropping a total of 0.85% of the resolution for much better performance.

sorry for throwing an idea which I don't know how to make it for real.

any tools which can help me to make a full resolution frame becoming interlace frame?

and I will need to combine the interface frames back to some full resolution frame. if you want some text illustration or showing the concept with some simple 5X5 matrix, I can do make it.

Combining frames sounds like a complete mess if anything in the frame is moving unless you some pretty sophisticated motion estimation software to go with it (similar to what compressed codecs do with motion estimation).

You should be able fairly easy to demonstrate this concept in action without motion estimation.  Take an existing dng sequence and use lightroom to export every even-numbered frame as 1080 and every odd-numbered frame as 720.  Then resize the 720 frames back up to 1080.  Then import into your video editor of choice.  It's not complicated or challenging/difficult.  It's just very time consuming and (IMO) not likely to be worth the effort.
I'm not even sure which cameras this would help - the Mark II can already keep up with 972p at 16x9 (looks downright decent when blown up to 1080) and even the 50D can do 896p.  I have to assume that both of those look better expanded to 1080p than a weird mishmash of 720p and 1080p...  and most of the SD cameras can only handle like 540p so at best, having half of the frames at 720p is just going to buy you another fractional second of recording at 1080p.

Oh, and additionally, unless I'm mistaken, it's not easily possible to alternate 1080p and 720p in camera while recording unless you want the 720 frame to be nothing but the center of the 1080 frame.

'Dongle nightmare' Here's my equivalent...



I wonder if the trolls realise how much of my own money was necessary to make ONE BOOK about anamorphic lenses?

While I don't object to you charging whatever you want to people who will buy your guides, it's sort of silly to complain about all the money you spent on anamorphics to write your book about anamorphic shooting since the adapters most likely haven't depreciated a lot since the time you bought them.  You could flip them and recoup a lot of your costs.  If you decided to keep them because you like shooting anamorphic, then more power to you, but that's hardly a cause for complaint.

Has anyone found a workaround for this yet? It's quite annoying to have to go back and readjust my settings each and every time i flip off the camera :/

Are you running a recent build?  I haven't updated in several days, but 4 days ago there were a series of commits by a1ex which seemed to be intended to save module settings.