Slashdot Mirror
Video Released, Crowdfunding Underway For Axiom Open Source Cinema Camera
New submitter atagunov writes "Video
clips
have been released as crowdfunding
starts for the world first open source cinematic videocam. "I am a filmmaker myself ... I would like to have powerful
tools that I know to have full control over and that I can tune and tweak," says
Sebastian Pichelhofer of Apertus. He is working on
the Axiom Beta,
the 2nd generation Apertus videocam, fully
open
sourced under
GPL and OHL.
It's not cheap compared to consumer-grade cameras, but being not-cheap hasn't stopped people from snapping up Joel Rubenstein's Digital Bolex.
And they also released the "Hollywood loves open source?" teaser of an upcoming documentary. https://vimeo.com/106452874
Why WOULD it be cheap compared to consumer grade cameras? Why are you even comparing it to consumer grade cameras? You should be comparing it to what it's intended to compete with.
The digital camera market for video is very competitive and so there are some great feature-rich cameras available for shooting cinematic video. Most notable is the Blackmagic Design Pocket Cinema (was on sale for $500 recently), the Panasonic GH2 ($400 used), GH3 ($600 used), and GH4 ($1700 new).
The camera market is undoubtedly saturated, but personally I found the project most appealing not because of the camera alone( Don't forget to note the 'beta' part), but because of the implications it has for a community. I like the idea of an open source camera community, one that is pushing the boundaries of what we think our roles should be as filmmakers and artists in a consumer market. I want to have a say in the how the tools I use are designed and built.
Those cameras aren't comparable.
Black Magic has a smaller sensor. The GH2 is limited to 23Mbps h264 (without resorting to hacking). Both have visible rolling shutter effect. Maximum resolution and frame-rate for both is 1920x1080@30fps. That's not to say these cameras are bad.
Yes, the GH4 has 4k recording + much higher bitrates and also costs a fair bit more.
For the Axiom, think along the lines of Red cameras but with open development + the option of adding features yourself.
because of the implications it has for a community
Exactly. But I also have another secret hope. I hope that once Axiom cameras take off the ground 3rd party manufacturers will step in. Anybody is free to make cameras/modules/parts according to the open specs, so we should see competition. A competition on a level field with no patents used for unfair advantage. Competition will curb prices bringing them in balance with manufacturing costs.
Will that not be a world worth living in? :)
And does it not justify supporting the project regardless of the initial camera prices?
My camera-nerd friend loves the new Sony a7s, as it's optimized for video. Considerably less expensive than this.
If I were to hack an image sensor, I'd work on phone cameras. At this price range, I expect the thing to be awesome out of the box, so "open source" doesn't really add anything.
So that just means the project is not for me. I hope it's successful regardless, but I suspect it's much cooler for the inventor than it will be for the users.
Not that this wasn't entirely predictable.
As I understand it, there are plans to add support for electronic lenses at some point (after the initial Beta version).
the world first open source cinematic videocam.
World first? Videocam?
systemd is Roko's Basilisk.
Well, the first thing people hit when shooting movies on DSLR-s/prosumer mirrorless cams are absence of the ability to shoot in RAW and limitations in choice of codecs. Often people are saying that if there is now RAW then at least the codecs should have been provided with higher bit rates. GH2 in particular was hacked to up the bit rates. Why bit rates? The classic example is high resolution video of a tree with wind moving the leaves around. You can only capture that without visible artifacts if you're using a codec with a high enough bit rate or better still if you're using RAW. Axiom's however are designed with highest possible storage bit-rates from the start. We shall probably need to wait a little longer for a proper storage backend (ssd raid) but I do not doubt that it will come soon after the release of Axiom Beta - it's just the most obvious next thing to do.
Not being a professional cinematographer or even a videographer I won't be able to continue the list of things which can be improved on in the present day commercial cameras. However I'm confident that the utmost flexibility of an open source camera will be put to a good use by the industry.
Hi,
I'm not on the Apertus team however I've seen many a discussion on the IRC channel and so I can answer that. Yes indeed powered lens mounts are on the list of features. It's just that they expected after the initial release of Axiom Beta, the one which crowdfunding is being collected.
To this I would however like to add that old manual lenses are arguably rather well suited for shooting movies. I think it is an established fact of life that nearly all cinematic production is done with manual focus (google "focus puller", "first assistant camera"). Now the modern electronic lenses even if they have a manual focusing ring are not well suited for the task. The problems are:
Old manual lenses certainly fix the first two issues, not sure about the last one. So I can conclude that while having a powered mount is very much desirable on Axiom cameras (and so it will come just a bit later) it is also true that the old lenses are in fact more suitable to the task of shooting movies and so the decision to deliver a fully manual Nikon-F mount first is justified
Yes I do have a nice powered lens for GH3. I also have some modern lenses my Nikon. However because of the above limitations I have already decided that should I dare to invest into actually buying an Axiom (presently I'm just a crowdfunding campaign backer) then I'll have to splash out for some old Nikkor-s as well. BTW I've been thinking of buying old glass on occasion anyway. I find it rather charming and the heavy metallic lens barrels inspire confidence in me
I tried to read the linked page but somehow I still do not understand how that camera be " Fully Open Sourced"
Can someone enlighten me a bit on that, please?
Thanks !
Muchas Gracias, Señor Edward Snowden !
Sony a7s is a decent contender. Full frame, amazing low-light capability.
Sure Axiom will win on some features too: 4:4:4 vs 4:2:2 video, RAW output, aptitude for change.
Let wait and see which users (movies, ads, etc) choose which one and for what reason.
P.S. Not letting go of GH3 despite my interest in Axiom, just a totally different purpose of a camera :)
Just because a lens has electronic focus doesn't mean that it doesn't have mechanical manual focus. At least on the Canon side of things, focus-by-wire lenses are rare. Most of the focus-by-wire lenses are old, discontinued models like the 50mm f/1.0. The only current focus-by-wire lenses I'm aware of are their STM lenses (mostly low end) and the 85mm f/1.2L II. The rest of their L line is mechanical, including the 50mm f/1.2 L (popular for movie work), the 135 L II, their various zooms, etc.
The big advantage that fully manual lenses have over autofocus lenses when it comes to manual focusing is that most manual lenses have a longer throw. This makes it easier to get a more precise focus when focusing manually. They don't do that on autofocus lenses because it would make focusing slower.
With that said, I think the industry's obsession with manual focus is badly misplaced. When you're dealing with 4K video, you want the focus to be right, not just close, and autofocus is a lot more precise than any human can possibly be, even with static subjects, with the best long-throw lenses, and with a separate person doing nothing but handling the focusing. The only thing holding back autofocus for video use was the slowness of contrast-based autofocus (and its tendency to seek). With the advent of on-die phase-detect autofocus capabilities, that limitation is rapidly disappearing. Add a bit of eye tracking into the mix, and I think you'll find that within the next ten years, nobody in their right minds will still be focusing manually, particularly when they're shooting 4K.
AFAIK, that's fairly rare even in fully mechanical lenses unless they've been modified. Perhaps dedicated cinema lenses are different in that regard. I'm not sure. But even some of my old screw-mount lenses from back in the black-and-white TV days had mechanical stops, so I'm guessing stopless lenses aren't exactly common.
The problem with old lenses is that they're designed for a world where cameras had relatively poor spatial resolution, and for much less reflective sensor material (film). I enjoy playing with old lenses on a 6D, and they create an interesting artistic feel, but they don't even approach the level of flare resistance, sharpness, etc. that you'd want for a digital 4K cinema camera. So if you're limiting yourself to mostly old lenses, you might as well limit yourself to 720p as well, because you'll be lucky to out-resolve that with most lenses designed more than about a decade or so back.
And if you have the money for modern, full-manual cinema lenses, chances are you aren't in the market for anything less than a highly polished, turnkey camera system.
So I really think that they need to at least lay the groundwork (in hardware) by making the plastic plates in front of the sensor removable and by including USB and DC connectors near the back side of that plate so that the system will be readily extensible in the future. That small change shouldn't require a huge amount of effort, and it will future-proof the design in a way that nothing else will.
Or, if USB isn't feasible, a high-speed serial port capable of at least 230 kbps would probably be good enough.
Just my $0.02.
Check out my sci-fi/humor trilogy at PatriotsBooks.
Critic makes the project stronger and better, as it shows us where work needs to be done, as long as stuff is pointed out. but for doing stuff, the project needs help, your help, right now the best thing one can do is spread the word as far as possible and donate some 100€/$ to the project. regarding comparing features or hardware specs... mhhh i think it is very hard and not well thought to compare proprietary with open source products, as no one knows yet, with what freaky stuff the hackers will come up with, when they can just simply do what ever they want. but i promise you, if the crowdfunding becomes a success and we will have two years development, hacking and testing time, you will see a gamma version, which will blow your mind! we got the creativity of all people we can reach with us, we just need to establish the first product wave. i like to remember on arduino and the rasPI, if they would have never succeeded in releasing their first product, we wouldnt have this sea of options and possibilities yet. so please support and spread the word about this project, for the good of all :) there will definitely be something good
coming back to you from this project 99% sure!
At least on the Canon side of things, focus-by-wire lenses are rare
Oh good to know top grade Canon lenses have a mechanical link even if the throw is short
I do own a Canon EOS 3, a film camera with 45 focusing points and eye control to choose from them (though I haven't used it for a long time now..). It might be that it's just the implementation on EOS 3 that wasn't right but this solution didn't work for me. I tried using eye tracking to choose the focusing point a few times and then switched it off toggling focus points manually.
I think you'll find that within the next ten years, nobody in their right minds will still be focusing manually, particularly when they're shooting 4K
It will indeed be interesting to see if that happens. If I was to bet on it I would say that it's not going to happen in cinematography. Indeed moving focus through the scene is one of the tool that a cinematographer uses to achieve the desired artistic effect. It is hard to imagine that a computer algorithm would be able to predict how fast or slowly we want to bring objects in/out of focus and how much smoothness we want in these transitions.
Further if wikipedia can be trusted the best film scanners go as high as 8K resolution. And if the precision of manual focus pulling was enough to shoot all of the movies we've watched and enjoyed so much on the big screens it should surely remain sufficient to shoot future films which are to be projected to the same screens. Don't think even IMAX films were shot with autofocus which means that manual focus was sufficient even for that vastly superior resolution.
And why wouldn't manual pulling be precise enough in a cinema setting? The scene is planned in advance, everybody knows exactly where each actor and object is going to be, marks are made (or stops set) before shooting begins for real. I heard even tape measuring is still common. Can autofocus beat the precision of a measuring tape? And failing that with digital you always have the option of zooming in to individual pixel levels to ensure your pencil mark on the follow focus device is correct. Lastly focus depth even though sometimes shallow isn't nil in most circumstances so small focusing errors might not have an adverse negative effect on the result.
So if you're limiting yourself to mostly old lenses, you might as well limit yourself to 720p as well, because you'll be lucky to out-resolve that with most lenses designed more than about a decade or so back
Chances are you've got the experience here while I certainly don't. However my impression so far was that what you're saying is true but not to such an extreme extent. People do use old lenses including those which are much older than a decade on modern still cameras and the results they're getting certainly don't look like they were shot with resolution of about 1/3 mega pixel provided by 720p video. I think the truth must be somewhere in between and the old glass must still be a valuable tool. After all if that glass was indeed that bad why wouldn't the prices not be nil today? Some of these lenses still command amounts of money which one on a budget would think twice before spending.
So I really think that they need to at least lay the groundwork (in hardware) ... so that the system will be readily extensible in the future
Here I do agree unconditionally. Groundwork needs to be there. I'm not on the project team but I understand this is precisely what is happening. If you look at the current plans of Axiom Beta design you will notice that the lens mount is attached with 4 screws to the camera body. This is done precisely so that a different mount can be easily fitted
Hhhhj
My camera-nerd friend loves the new Sony a7s, as it's optimized for video. Considerably less expensive than this.
Sony A7s looks an awesome camera, for sure. But unless someone unlocks more features, you're limited to what Sony wants you to have. This might not be a problem for most.
For example, can the A7s do timelapse video recording? I know Sony does a $9.99 app for the NEX cameras, but how capable is it? Can you do speed ramping effects etc. Stuff like this will hopefully be easy to do with the Axiom. But not right away with the Beta model.
If I were to hack an image sensor, I'd work on phone cameras. At this price range, I expect the thing to be awesome out of the box, so "open source" doesn't really add anything.
So that just means the project is not for me. I hope it's successful regardless, but I suspect it's much cooler for the inventor than it will be for the users.
The Axiom Beta is as it's name suggests. There are plans for a Gamma model which will be much more of a mature product.
Phone camera sensors wouldn't be suitable for anything other than experimentation - small sensor, poor low light, no shallow depth of field, probably rolling shutter issues, etc.
Quote test
That's an interesting question, and you're right that for that particular effect, you're probably better off doing it manually—preferably with a long-throw manual lens and a reasonably long stick attached. But that's likely to be an occasional thing, with either static focusing or traditional subject-following focusing used for probably 99% of your shots; if you're using focus to move from one subject to another for 99% of your shots, the viewers are likely to get nauseated rather quickly. :-D
Depends on how narrow the depth of field is. At large f-stops (e.g. the Zeiss 50 f/0.7 lens that Kubrick used), it can be done by hand, sure, but if you blow that up to where you can see pixels at 4K resolution, you're almost certainly going to notice the softness compared with what modern electronics could achieve, particularly if the subject is close to the lens and he or she decides to move a fraction of an inch. Mind you, that's a rather extreme case. :-) At more sane stops, it's not quite that bad. It's still a lot of work, though—work that's largely unnecessary with a decent, modern, subject-tracking AF mechanism (even without eye tracking to set the starting point). It's not that focus pulling can't be good enough, so much as that the extra work to make it good enough is significant, and it makes little sense to bother with that when a simple circuit can do at least as good a job (if not better) without all that effort. :-)
That's true. With that said, the higher the resolution, the more visible that small effect becomes. At some point, you start to swear because the soft focus limits your effective resolution, and all those extra pixels are just taking up more space on disk without any real benefit. I'm not quite sure where that magic point is for manually focused movies—you'd have to ask somebody who regularly does film scanning and media ingestion for their take on it. Obviously it would depend on the f-stop, the distance to the subject, the film format, the focal length, and the skills of the person doing it. :-)
I'm probably being a bit on the cynical side; I'm sure there are some older lenses that are usable at 4K. The point I was trying to make was that the newer lenses are breathtakingly better at high resolutions—maybe not at 4K, but long before you get to 8K. And their handling of bad lighting conditions (lens flare, for example) is just amazing compared with the older lenses. Unless, o
Check out my sci-fi/humor trilogy at PatriotsBooks.
Let me correct myself further. 720p is just shy of a million full-color pixels. On a Foveon sensor, depending on how you count megapixels, that might be the same number. On a Bayer-filtered sensor, it's more in the neighborhood of 3 MP, because each color channel has about a third the spatial resolution of the sensor as a whole.
Check out my sci-fi/humor trilogy at PatriotsBooks.
it looks like someone played Star Control II
all of the above cameras would be ok if they had global shutter.