Suggesting: The thin camera (A new camera)

The place for questions about shooting with Blackmagic Cameras.
  • Author
  • Message
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Suggesting: The thin camera (A new camera)

PostWed Oct 18, 2017 3:01 pm

The future is probably a thin small mobile like camera rather than POV cameras. Even Red is going Mobile but honestly, most mobiles are far to unreliable with not the best performing mobile sensors, and some you can't even put a card into, let alone 1 terabyte in future.

Now, a phone is a phone, even if a great camera, it is prone to getting damaged in use. How would you feel if you only had one lens on your main camera and it got scratched (maybe after the phone rang and you lost balance and dropped it). So, a phone is still a phone, and a camera a camera, apart from turning the phone and data off, just the doubling handling could lead to more damage. The Red product undubtedly will break the mobile mild and do great filming, but to get the best results requires the multipoint camera module, which on top of the nearly $1200US-$1600US phone could be rather costly. However, using mobile parts could a multipoint thin camera could be.dine cheaper. The multipoint allows you to emulate a larger lens in suitable light, inside a thin tiny camera, suitable for drone, action, pov work.

The issue BM has had before is use of FPGA. Red also used this strategy on its way to Application Specific Integrated Circuits (ASIC's) which are custom chips many times more efficient than FPGA's. There is only one FPGA technology I know of that came anywhere close to ASIC's, and that was snapped up by the military. The next FPGA technology, magnetic FPGA,, can
overcome the issues, whenever it finally gets here. FPGA's are slower and uses many times more energy, going to heat. But in previous generations a customisable FPGA was more convenient to get performance from than a normal microprocessor based system, but I now believe that has changed.

Nvidia has made great strides in low energy integrated processor General Purpose GPU chips. There next Xavier chip series is one of the most complex CPU's ever made, but uses very low power. There is normally a mobile version. It should be capable of handling 8k and multiple camera vision for multipoint cameras. The previouse boards for multicamera AI drone use were credit card sized. An entire processing platform for a new camera. Because these parts are bulk made to sell to a number of manufacturers they gain lower cost, like mobile hardware. So, a multipoint camera phone might be less than $1000 Australian, for something that could displace low end cameras for professional use. The Ambarella ASIC chipset technology could make a even lower cost multipoint camera than this, but tends to be set up with lower specs than what we would like.

Sometime back I suggested the Video Assist could be made to have full camera control and GUI system (like a large Pocket Camera) to use any of the BM POV cameras, or competitors. But I also suggested that the Video Assist (or in this case, Video Director) could have 2 or more mobile like sensors to film computational 3D photography like video). 4-5 sensors is preferable, and they can be put on IMax an larger Medium format spacing to emulate that format). The Light 16 camera is an example of this. While the best mobile sensors, like used in the LG and Google Pixel cameras would be good, they are not necessary. However, if we look above we see the Nvidia product allows for another way to make the Video Assist with full Linux computer programmability. Now, video assist could be made into a tablet format with multipoint camera system.

Going a step further, a credit card sized thin camera can be made, that attaches into a video assist tablet case. The case merely presenting interfaces, battery, storage, display and controls. The credit card like thin camera could even record itself on a drone, or in your hands.
Offline
User avatar

Frank Glencairn

  • Posts: 1570
  • Joined: Wed Aug 22, 2012 7:07 am
  • Location: Germany

Re: Suggesting: The thin camera (A new camera)

PostWed Oct 18, 2017 6:17 pm

If my camera would be phone sized, I would kill myself.
http://frankglencairn.wordpress.com/

Set your zebras to 100%, ETTR and you're golden - and NO, you can't use TB as output for an external monitor, and you can't download the footy via TB ether.
Offline
User avatar

rick.lang

  • Posts: 7571
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Suggesting: The thin camera (A new camera)

PostWed Oct 18, 2017 7:49 pm

Miniaturization of tools no doubt will continue, but the properties of the light we try to capture have remained constant for about 14 billion years in this universe. The wavelengths and photon energies are the same and how they behave when they encounter other materials is the same. To capture quality, rather than merely quantity, we are at the mercy of physics.

Most people here are coping with that limitation using tools that provide high quality including those who require light weight. You can also argue these phone cameras are marvels of design and engineering and produce good images, but generally a thin camera will produce a thin image that breaks the moment you stretch it beyond its design target.


Sent from my iPhone using Tapatalk
Rick Lang
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostThu Oct 19, 2017 4:25 am

Firstly, negative bombing threads first up, in reality, is counter productive. It also lacks the imagination to overcome problems. So, is there anybody here with a real mind for science?


A worthy quote..

Now, Rick, we are not talking generally, we are talking quality parts in descent lights (though regular dim light also doable). We are talking sensors that do 10-12bits plus, and 14 stops native (though HDR video modes gets you into 20 stops territory, but likely anything over 16 stops will be a stretch grading. These are the specs of sensors used on some phones, car recorders and action cameras.

Even though the laws of physics, debstably if you are a real scientist, may not have change, technology and materials used to take advantage of them have. For a very long long time there has been small sensors with more than 27stops with could color over the range. They have sensors that will pick up a single photon and with a very low noise floor. All these things have been done by others. And I would not put a 4k mini up against some phones for photo pad quality.

Lens quality, a good hybrid liquid lens will give you it.

Remember we are talking about a camera you use instead of anything under a mini 4k, for the convenience of it in various personal and professional roles, such as print journalism, that you can on sell the design into a phone company's phone if you wish. We would be talking raw or pro video rates and modes.

Now, look at the pictures from Light 16 as an example, and realise you can go beyond that with more recent sensor technology.

Now, the areas you could have attacked, are fill factor (though foveon has no real issue with this for years) aggressive antialaising filtering, abberations from the microlensing affecting pixel purity (though removable).
Offline

Tristan Pemberton

  • Posts: 571
  • Joined: Fri Aug 08, 2014 6:07 am

Re: Suggesting: The thin camera (A new camera)

PostThu Oct 19, 2017 8:09 am

Frank Glencairn wrote:If my camera would be phone sized, I would kill myself.

Yeah, I'd ask you to take me with you.
Director
Australia
www.flywirefilms.com
Offline
User avatar

rick.lang

  • Posts: 7571
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Suggesting: The thin camera (A new camera)

PostThu Oct 19, 2017 6:44 pm

Wayne, not meaning to attack your ideas, but very skeptical the things you mentioned will deliver in a credit card sized camera at normal operating temperatures.

There’s such a history of researchers building something in the lab and achieving astounding results in a particular area, with projections of it going mainstream in less than five years. Most times we are still waiting for these miracles to be delivered in a practical application.

Must have been about five years when we were told memory access times (in the lab) were going to improve performance in your computer a thousand fold real soon now. Still waiting.

Back to your topic though, a thin camera...


Sent from my iPhone using Tapatalk
Rick Lang
Offline

Anatoly Mashanov

  • Posts: 444
  • Joined: Sun Sep 28, 2014 11:36 am
  • Location: Russia

Re: Suggesting: The thin camera (A new camera)

PostFri Oct 20, 2017 2:30 am

I think a thin camera will never appear.

Yes, you could make ASICs, use specilaized GPUs etc, but don't forget that Canon which uses specialized ASICs is a monster company and the camera business is a small part of it. And they have a lot of experience in producing ICs. Nvidia has about 15+ years in IC business. Both Canon and Nvidia produce trillions of chips and R&D costs per chip are negligible.

BM produces thousands of cameras (I have no precise numbers). So R&D is a big part of costs and it would be even bigger with ASICs instead of FPGAs.

Then, BMPCC consumes 7 watts. The FPGA has NO radiator (at least on published photos). But the cryogenic device is cooled with a pair of heat pipes. The FPGA is NOT a main power consumption device, cryogenic device is. It's just meaningless to economize on FPGAs.

And the last: the ergonomics. If you use a stills camera then you either keep it in hands and look at the screen, or look into a viewfinder pressing the camera to your forehead. You could do it for some stills but NOT for a long cinema shot unless you are a SuperHero. So the ENG cameras are kept on shoulder. But it means that the camera should have a viewfinder brought forward or a screen brought even more forward, and it also means that the viewfinder should have a big ugly fat connector (I mean SDI, not micro HDMI).

If you use a small camera connected with a screen that is a recorder too (Video Assist etc), then the big ugly fat connector becomes twice bigger and twice uglier simply because it passes not video or Prores, but RAW. Or, if you want, the system becomes limited to Prores only.

Can you imagine a smartphone with full size BNC connector?
Offline

Denny Smith

  • Posts: 7084
  • Joined: Thu Aug 01, 2013 4:19 pm
  • Location: USA, Northern Calif.

Re: Suggesting: The thin camera (A new camera)

PostFri Oct 20, 2017 5:18 pm

You could use the DIN mini BNC connector.... :roll: But, yes a small thin camera for shoulder work would be counter-intuitive. In 2001 new conference scene. they showed video news cameras (the size of a Micro camera) “floating” around the room, once you get to this point, then small would work.
Cheers
Denny Smith
SHA Productions
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostSun Oct 22, 2017 2:06 pm

Now, improving memory times as you described, will if itself not improve general computing processing a thousand times. I belong to a branch of computing which demands high speed random access memory access. So, such memory speed up, as wonderful as it is, won't improve things anywhere near that much generally (lost track of what I was going to say) without specific architecture. So, would it be under special workloads or architectures Rick? In the hundred K plus processing arrays I look at, a thousand fold increase on main access would be a great benefit. I have to go through great lengths organising a scheeme for global access. If it is used much on conventional memory you get slow downs from processors waiting for access. A thousand fold increase would mean you could possibly divide up the capacity to avoid much stalling, but obviously a 100k+ speed up would be better. I've got memory design proposals for that, but doing standard memory is another few levels of complexity more. But in everyday reality, unless you increase processing speed to match, you are unlikely to maximise general performance. I'm interested, which memory technology is this. Now, my guess is if you apply such a thing to a GPU or processor array of 100 or more units you could maximise performance. But, magnetic processing units are one of the closest expected next gen technologies last Iheard (these days I don't know) so shouldn't be dismissed.

I have seen the expected end of silicon advancements go ahead continually. So, I'm basing present day expectations on the expected 5nm technology, which is comfortable for our apps.

Now, when I say credit card, I'm not expecting it to be as thin as a credit card anytime soon. I did not say how thick. Obviously anything thinner than a pocket is thin, it doesn't need to be as thin as a mobile, but why not. So, you start at whatever thickness makes sense, and work from there in future. We pretty much have the cameras now in our pockets, just without the multipoint (and even at 720p or fullHD, multipoint is usable. The Nvidia credit card sized Jetsons computers, are basically this camera already, with computational drone AI video from many camera inputs. But even.thiugh they might be closer to credit card, you still have to add screen, battery, controlls, sensors and lens in a case which is going to add thickness.
Last edited by Wayne Steven on Sun Oct 22, 2017 3:38 pm, edited 1 time in total.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostSun Oct 22, 2017 3:06 pm

This was supposed to be posted first but got missed:

Look I know people in industries, and this industry, and as a researcher I have a very relative idea of what it possible. But everytime something realistic and positive is said, visionary, the first things that turns up in threads are negativity. It kills the joy.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostSun Oct 22, 2017 3:30 pm

Anatoly Mashanov wrote:I think a thin camera will never appear.


Anatoly, see my above post.

Ambarella has had 8k capable designs for years, and as far as leading standard low power video technology, they may be it. People here keep confusing BM cameras as examples of what you can do in low power technology. But their designs are not about low power or low cost chips, but their own concerns/needs, that technology is many times lower power. Presently I would guess their designs have moved suitability from pocket like sizes to action camera, and will continue down. But ambarella is orientated towards consumer video, but with datarates maybe useful professionaly in the video feild. However, being unfamiliar with the design internals, it may or may not be reprogrammable, but as long as you can get a recordable raw stream from it. They could probably do a 8k raw camera in a size between the pocket and a phone. However in this case, pro video is good enough. But for something reprogrammable now that handles video, Nvidia, who do mainstream good performance for energy designs.

Interesting note, after Ambarella released their design, they licensed the technology to many top camera makers. Before that many cameras compression was rubbish. I wouldnt be surprised if many Canon designs use this technology. That is why they are a leader in lowenegy camera chip design.

Don't get confused about company sizes as to how good their chips are, it:s their expertise that does it. Ambarella had leading people and did leading graphic processing in the 90's.

So, I didn't think the pocket had a cryogenic cooler?

Your point about thousands of BM sales, is the point. Getting high end low energy ASIC processing solutions by the thousands is not so cheap and easy per unit as by the 100k or milion, but getting FPGA is probably closer. Getting thousands of Nvidia Jetsons boards is probably easier too. The advantage with mass produced high volume designs in low volume, is being able to use and adjust the reference design.

Now, ergonomics, Anatoly, we are talking about people with blogs, and newspaper reporters that use phones, and their photographers, if they still have one, expected to produce video clips fur the website. In feild tv local news, you may use a shoulder rig with extra battery, controls whatever, with the camera placed where the viewfinder normally is, but there are other rigs, and horrors, gimbals. :) . But I am talking about something below a mini, not an Eng camera, instead of camcorders, prosumer, stills, drive, action, and lower quality phones. Something in the BM pocket camera market. Sure, with some effort you could do more than that aswell. Even camcorder formfactor or shoulder case. So, at the same time many connections become irrelevant and uneeded, when you can use USBC under USB/TB, wireless etc. You don't destroy progress because of things like connectors that should be remade. We are talking about two different product categories here.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostSun Oct 22, 2017 3:31 pm

Well, it's been a nice couple hours guys. Thanks.
Offline
User avatar

rick.lang

  • Posts: 7571
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: Suggesting: The thin camera (A new camera)

PostSun Oct 22, 2017 6:14 pm

Apologies, Wayne. I misspoke. The 1000x improvement was in memory performance, which certainly doesn’t improve overall compute performance anywhere near that with so many other components involved. My point was that claim was probably two years ago and not a hint of it has appeared in memory that we mere mortals access in computers as far as I know. Hypothetical and theoretical can be there, but practical is a ways off. When you said credit card size, I assumed you mean as thin as a credit card given your topic title. But I can see it’s more feasible if you consider a small stack of credit cards. Hope you’re having a good weekend trying to expand our horizons.


Sent from my iPhone using Tapatalk
Rick Lang
Offline
User avatar

Robert Niessner

  • Posts: 1261
  • Joined: Thu Feb 21, 2013 9:51 am
  • Location: Graz, Austria

Re: Suggesting: The thin camera (A new camera)

PostSun Oct 22, 2017 8:42 pm

Wayne Steven wrote:Interesting note, after Ambarella released their design, they licensed the technology to many top camera makers. Before that many cameras compression was rubbish. I wouldnt be surprised if many Canon designs use this technology. That is why they are a leader in lowenegy camera chip design.


I don't think Canon is using Ambarella's technology. They have always designed their own custom chips called DIGIC DV
-> https://en.wikipedia.org/wiki/DIGIC

And I have to say when comparing similar compression codecs and bitrates, then Canon had always the edge over e.g. Sony.

Wayne Steven wrote:Don't get confused about company sizes as to how good their chips are, it:s their expertise that does it. Ambarella had leading people and did leading graphic processing in the 90's.


Ambarella was founded 2004.
-> http://www.ambarella.com/uploads/docs/c ... -sheet.pdf
Robert Niessner
LAUFBILDkommission
Graz / Austria
---------------------------------------------------------------------
Blackmagic Camera Blog (German):
http://laufbildkommission.wordpress.com

Read the blog in English via Google Translate:
http://tinyurl.com/pjf6a3m
Offline
User avatar

Frank Glencairn

  • Posts: 1570
  • Joined: Wed Aug 22, 2012 7:07 am
  • Location: Germany

Re: Suggesting: The thin camera (A new camera)

PostSun Oct 22, 2017 8:52 pm

Why would anybody have a credit card sized camera? Folks build Frankenstein rigs around their Pocket cameras, cause it is to already to small/light to hold it right, same goes for DSLRs - there is a wohle rig industry just for small cameras and phones. Making cameras even smaller, just because you can, just doesn't make any sense. So what's next? Credit card rigs?
http://frankglencairn.wordpress.com/

Set your zebras to 100%, ETTR and you're golden - and NO, you can't use TB as output for an external monitor, and you can't download the footy via TB ether.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostMon Oct 23, 2017 3:18 pm

rick.lang wrote:Apologies, Wayne. I misspoke. The 1000x improvement was in memory performance, which certainly doesn’t improve overall compute performance anywhere near that with so many other components involved. My point was that claim was probably two years ago and not a hint of it has appeared in memory that we mere mortals access in computers as far as I know. Hypothetical and theoretical can be there, but practical is a ways off. When you said credit card size, I assumed you mean as thin as a credit card given your topic title. But I can see it’s more feasible if you consider a small stack of credit cards. Hope you’re having a good weekend trying to expand our horizons.


Sent from my iPhone using Tapatalk



Rick. Thanks for that. I still would like to know which.memirt that was, as I have not been keeping up with upcoming developments for years. Unless they state something is going be their product in X years time, it could take a very long time to reach commercialisation and then longer to come down the lower end of the market. When the manufacturers can see the end of the production development process coming they can then say they are planning an introduction in X period. But even then things can stall or get cancelled fur a number of reasons. Look at OLED TV, Samsung and LG were planning much better yeilds then it turned out. They would have done production development and planned for refinement, but it didn't turn out , so Samsung pulled out. I remember reading about new OLED development in the early 90's irlate 80'dms from Cambridge I think, but it took around decades from the lab.

Now, with the 1000x memory, very useful for me, but even if it is a product expected to go into production all the things previously mentioned can happen, it can get canned or sold on, or somebody like intel buys it and introduces it on a new mass array processor. I think I remember mention of Intel having something, but don't Rembert it was memory related. But you might be talking about the Intel micron flash alternative, definitely good, but the 1000x performance improvement would be for write. For storage that's fantastic, for processing not much benefit except in quantity. That memory is probably under production cost refinement, and I believe has-been out for a while, but upmarket. If it was processing memory, a 100Ox speed up is useful for the military Service well as processing centers veting data for security. A far superior almost ASIC like in performance FPGA years ago I wanted to use didn't come to the market because the military bought exclusive use. The flat folded light path lens I wanted to use was also snapped up by the military, even the 9 lens sebsur technology. There is a stack of tech that the military has we can't get hold of. We had done guys developing volumetric image projection dissapear in America, then some people I know we're invited to develop optical processors, and they announced holograms would be used in the battle feild. One guy from intelligence I knew was telling me of talking computers back before it was supposed to be so. So, just because something hasn't come out dont think it means that it was unreliable development, it maybe we just missed out.


Now, another example is the 1 nanometer transistor in the lab in recent years. Even if it proves to be tla worthy option, it could take a while to perfect. Whereas the recently proven 5nm nanoscale flat transistor we might see in future years.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostMon Oct 23, 2017 3:40 pm

Ambarella is a team that had a previouse company I believe. I can't remember it's name but it was doing cutting edge graphics technology.

Whatever Canon maybe doing, a lot of these custom Japanese camera brand processors are at least using licensed Ambarella technology, if not cell designs on the ASIC, or ambarella custom design, or ambarella chips (I think less so in these brands).

Frank, a credit card sized camera is just a different market. It's not the tool, but Huw you use it. At that size it is useful in pocket, on tripod, in drone. If you want offer, put inside a case, a camcorder, or prosumer case with conventional controls. As I said, a step up from your phone, or as a small stills. Now as a shoulder rig, you can mount it to bars with handles for controls and with connectors. Batteries or storage can go over the shoulder to hook in and balance weight. The thing is so light it can be carried around with the battery/storage hooked over your shoulder resting there. But you can also get the bars and fold it up etc, to a size lessthen a gh5. The whole point is it's just a cheap option in place of theickey with lots of alternative uses. Red are planning a phone and tablet sized versions (likely at a price).

Now, wherever you want credit card firm factor or ursa sized eng camera, the Nvidia credit card sized Jetson platform could do it.

Now, I think there is at least one credit card formfactor rig out there :) folds out for use.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostMon Oct 23, 2017 3:42 pm

I was going to write about something I missed in the first post, on making low volume chips and sensors cheaper. But it is getting too late.
Offline

Anatoly Mashanov

  • Posts: 444
  • Joined: Sun Sep 28, 2014 11:36 am
  • Location: Russia

Re: Suggesting: The thin camera (A new camera)

PostTue Oct 24, 2017 12:40 pm

Wayne Steven wrote:Ambarella has had 8k capable designs for years, and as far as leading standard low power video technology, they may be it.

We are a collective of cinematographers. Ambarella is irrelevant, you will be assimilated.

Ambarella codec is used for consumer grade video. There is a terribly wide market for it: there are lots of consumer grade video players and consumer grade video recorders. Their task is to compress data as much as possible while preserving just enough quality for the consumer grade consumer.

Blackmagic cameras (as well as RED cameras and possibly ARRI) are absolutely different class of equipment.Their task is to receive the data from the as good sensor as physically possible and record them as is, compressing data only as it's possible to do without losses. All other processing is to be performed offline. You don't need much computing power, but you need a fast storage. Really fast storage. Ambarella uses 240 Mbits/sec to store 8K video, while BMPCC uses about 240 Mbits/sec for 1920x1080 only, and Ursa Mini uses 2 cards in parallel for 4K video just because the single storage with required speed just doesn't exist.

From my personal point of view, codecs similar to Ambarella are [censored] [censored]ing [censored]. I just see the compression artifacts on everything, and it's why I, a total amateur, decided to buy BMPCC.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostSun Nov 12, 2017 2:59 am

I've lost track of where we were with this Anatoly, but I'll give it a crack.

Most of the market is about professional use non raw video, not elitest low wirk flow cinema cameras. Did BM stick with RAW only? Why didn't they?

It is dreaming to think that something is impossible because somebody else didnt do it.

You are applying dreaming to my answers that show that there is a clear superior edge out there beyond what we see from lower technology manufacturers.

Now, the ambarella codec is actually a nice version of h264. Don't mistake datarate for a codec. Many pro cameras with pro datarate, will be using licensed ambarella technology for h264 versions etc, just at higher data rates and higher h264 profile spec. I can't speak for the other codecs they do or may have in the background. But they have been the heavy weight leaders. The h264 videos were rubbish on many cameras before them.

The older chip in the Yi 4k+ action camera runs at 120mb/s, and bueve me, I have seen people claiming to use 100mb/s professionally :( , but is capable of over 173 or 200 mb/s I understand. That is acceptable, especially if they turn 4:2:2, and I forget if the chip has more than 10 bit more. The 2015 Sony sensor does a nice job on that camera too. The picture color/levels rendering is done better than most video I have seen from M4/3rds stills cameras. It is about his you implement things. Many things you see are artificially limited, otherwise BM cameras and Red COULD NOT be this successful in the market. But they are there to take advantage of. This is actually reality, bit some dream that it isn't there.

Now, those images are pulled in raw off the sensor, and could be recorded uncompressed if they wished. They could, if they wish compress the raw, maybe they do already in designs in certain cameras, but that would be behind closed doors in their licensing at the moment, if they are doing that. The point was, not some dystopian allegation on whatever I may write, but a lot more, in reality, can be done then currently thought from viewing low volume companies trying to spin their own.

Now, as for the Nvdia system, I imagine it can be programmed to compression professional video or raw from multiple 8k sensors (and at high frame rate if supported at sensor interface).

So, tomorrow can be an exciting day, with pro quality data rates, features and rigs for new cameras, which we should embrace.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostFri Nov 24, 2017 12:07 am

I want to cover something I meant to put in. Alternatives to custom circuites, and image zebsirs.

There are small foundry setups that use stamp technologies that could possibly be done in house for low numbered runs of low ASIC's and sensors.

These systems use a physical contact printing process. They are ongoing refinement, and one of the few things presently for potential in company use.

As a side note:
Magnetic FPGA and other magnetic circuites are going to kick the issue. They are talking up to a million times less power consumption then using a processor (3-7 or so years back I think). So, the processing done inside the present camera could easily be done in a phone at 9 cameras of 8k 16 bit, 5 color, the proposed 240fps 3D standard, each and barely kick up a sweat (from the processing).

A side note:

There are various processing array technologies, some aimed at replacing FPGA. I'm not going to get into it. But the Green Array chips undisclosed advanced processing proposal is probably a leading design. Their extremely light weight 18 bit version has set industry records in the past, without even the best manufacturing process (but an unusual thing to program).

Now, there are two other sorts of things to consider:

GPU and Memory array Processors. Now, memory array processors have a high volume parralel data processing ability. They use memory fabrication processes to drive down costs and increse performance. But these fabrication processes use less layers and are not suitable for full cpu's etc, but accelerating simpler data processing functions, which might, it might not, be excellent for the sort of compression we are talking here. BM engineers probably didn't know about them, or most of the non-gpu technologies mentioned here.

Thanks.
Offline

Richard Thompson

  • Posts: 6
  • Joined: Fri Sep 01, 2017 12:01 am

Re: Suggesting: The thin camera (A new camera)

PostFri Nov 24, 2017 2:41 am

Well I had a studio camera in the early 80s, in my living room. My dad brought home a Sony Beta Max and camera and I thought wow, this thing is tiny, then I bought a panasonic VHS Camcorder and thought wow, can these modern cameras get any smaller. Fast forward 30 years and a $200 Cell phone will blow the away any of the cameras I had as a young adult. To say that Cameras are not going to get smaller, better and more efficient is to ignore the recent past. We cant even imagine what sensor and processing tech is 20 years down the road. But I welcome them
Offline
User avatar

Frank Glencairn

  • Posts: 1570
  • Joined: Wed Aug 22, 2012 7:07 am
  • Location: Germany

Re: Suggesting: The thin camera (A new camera)

PostFri Nov 24, 2017 7:32 am

Why would you want such a camera in the first place Wayne?
http://frankglencairn.wordpress.com/

Set your zebras to 100%, ETTR and you're golden - and NO, you can't use TB as output for an external monitor, and you can't download the footy via TB ether.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostFri Nov 24, 2017 11:08 am

Why not, good business tool.
Offline
User avatar

Frank Glencairn

  • Posts: 1570
  • Joined: Wed Aug 22, 2012 7:07 am
  • Location: Germany

Re: Suggesting: The thin camera (A new camera)

PostThu Nov 30, 2017 8:08 am

What kind of business you are talking?
http://frankglencairn.wordpress.com/

Set your zebras to 100%, ETTR and you're golden - and NO, you can't use TB as output for an external monitor, and you can't download the footy via TB ether.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostThu Nov 30, 2017 1:05 pm

See above Frank.
Offline

Denis Kazlowski

  • Posts: 51
  • Joined: Tue Nov 28, 2017 12:54 am
  • Location: NJ - USA

Re: Suggesting: The thin camera (A new camera)

PostFri Dec 01, 2017 2:53 am

If the discussion is ASIC vs. fPGA , it's the Field Programmable part any pro camera vendor desires ability to re-load it with better code, during development, and again when units are fielded - ASIC's are pretty much baked-in when completed and are expensive to produce in smaller quantities presently. The defense community, and even the machine vision to some degree is able to afford in-house ASIC production, no really they are, I don't think even the most popular camera vendors combined together have that pull, unless it translates into millions of units, like mobile phones, and unlike any consumer and pro camera systems.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostFri Dec 01, 2017 3:59 am

Lower volume chip making method, and higher performance fpga (5 GHz 2005) which the military has, has been covered.

Camera manufacturers do have asci's, most consumer, and probably prosumer etc cameras. They can use somebody else's ASCI designs, or chips, like ambarella. The other way is they can do a lot of different asci chips and co-inhabit them on a wafer, reducing the costs. There are lots of processing/compression designs out there that can be use of off the shelf and be reprogrammed.

BM does a lot of different product, so I imagine they might be thinking of asci too. However, licensing various aspects of up and interfaces might be too much.
Offline

Denis Kazlowski

  • Posts: 51
  • Joined: Tue Nov 28, 2017 12:54 am
  • Location: NJ - USA

Re: Suggesting: The thin camera (A new camera)

PostFri Dec 01, 2017 7:55 am

Wayne Steven wrote:Lower volume chip making method, and higher performance fpga (5 GHz 2005) which the military has, has been covered.

Camera manufacturers do have asci's, most consumer, and probably prosumer etc cameras. They can use somebody else's ASCI designs, or chips, like ambarella. The other way is they can do a lot of different asci chips and co-inhabit them on a wafer, reducing the costs. There are lots of processing/compression designs out there that can be use of off the shelf and be reprogrammed.

BM does a lot of different product, so I imagine they might be thinking of asci too. However, licensing various aspects of up and interfaces might be too much.


Yeah, the licensing of ASIC code and platforms is about $100,000 to $1,000,0000 per design variant, a company that needs to adjust image pre-processing or demosaicing and post-processing in their R&D lab and then do it some more to customers in the field is not going with ASIC designs - they are going with fPGA due to orders of magnitude of cost.

Sure maybe some tiny cam vendor licensed the Sensor and the ASIC and the OS from a company under a white label and put their logo on it. But they were expecting to sell a million units for under $299 USD price, I cannot see your suggestion as being realistic in terms of internal cost. Even fPGA costs a ton of money in "pre built" code you can use on it, but 1000x less than ASIC designs.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostSun Dec 03, 2017 8:28 am

Yet, many cheap cameras do this with ambrella chips for less than $30, some even cameras around $30. Then they go up in price and complexity. And yet, canon and the rest did do their own ASIC and when ambarella came along with they're behind closed doors top designs, companies started licensing them. Real history. Fpga is a red herring for small low low volume companies. Six actually, lowest. I hear a lot of big rotting cinema camera chest bulging over the years, but often the reality is not able to afford top of the line hardware techniques. Red is the only independent company that has been able to break the mold.

Now, what I have said is realistic, it is actually what has happened, and if you did want to make a camera "properly" you will be able to sell it by the tens of thousands to hundreds of thousands, maybe above a million, to consumers as well. If people are extremely ignorant, or just don't care, they will sell it only to cinema people for some fantastically high price innthe thousands. That's the reality. You think $100k-$1million for an asic is a high price, you don't know how expensive it is to do things (btw, from how low ambarella profits are, I would think they are on the lite side of charging). I got quoted over a million to fully develop a design (based on previous work) which was considered a bargain years ago. Divide by 1000 cameras, that's $1000 for the design, plus the chip. For 10k cameras that's just $100. For 100k cameras, where to Sim for, that's $10 a camera, which will be less than the chip price. But in a full mtjet, once you get over 10k units if cameras, you had better look at getting price down value up or selling to wider markets like consumers. At 100k units, you had better have something that most cinema and videographers will buy, or be selling to consumers. There are only so many customers.

However, with the programmable asic solutions out there, you don't even need your own asic's. I think I calculated that nvidia's Xavier solution should be able to wavelet cineform record 8k. That is potentially a great price depending on how large their volume gets with that solution. The truth is a lot more out there then people usually think. FPGA, needs to make great strides (that military one is actually one that had something). Magnetic fpga and magbetic ASIC, would be the end of conventional ASIC. However, that is expected early next decade.
Offline

Denis Kazlowski

  • Posts: 51
  • Joined: Tue Nov 28, 2017 12:54 am
  • Location: NJ - USA

Re: Suggesting: The thin camera (A new camera)

PostMon Dec 04, 2017 9:19 am

Wayne Steven wrote: If people are extremely ignorant, or just don't care, they will sell it only to cinema people for some fantastically high price in the thousands. That's the reality.


That is the reality actually. Not a lot of people make films or content and the barrier to entry for imaging and dissemination of images in now a iPhone or Android. This is why low level photographers, some 'videographers' are out of business, pro-sumer camera sales are in decline. The ubiquity of it all. Average consumers are just that.. average, if you need anything a step above that 'mass market' you're immediately in a small minority of professional consumers with specific needs. So economics of scale do not apply.

The very post production houses whom used to buy BM products in bulk for their edit bays have been decimated by iMovie, tape delivery is almost gone so $50,000 decks are for sale on eBay for under $1000. Professional imaging in general has sustained a serious financial hit, rates are dropping like there is no tomorrow. Moving professional images are on the rise due to the new delivery methods, and will be for a while, until it evens out. Leaving only high-end operators and tablet owners in the future.

ambrella chips for less than $30

Ambarella, Inc. (NASDAQ:AMBA) - Did you buy their stock or something? They did have a big loss in Q3.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostMon Dec 04, 2017 4:41 pm

A loss, whooho, about time. Indus tell you that they weren't profiteering (that's why there stuff is so cheap). I guess I won't be running into Diddier on holidays here for a while (that's actually a sad point). Now is time of motivation to a new direction for them. I have admired their advancement, but noticed the certain limitations (like all the ambarella products I have have a post 8 bit 4:2:0 HDMI output, so forget about a recorded on it) nothing like vc5/cineform or any pro version codec.

The issue is that ambarella tech was very good in the consumer space, it could be used on DVD or phones etc, of gpu cards (I imagine it probably is). But any crappy chip could do for DVD or other, and there are already supply lines of existing suppliers of crappy solutions. Now, phone chipset manufactures are delivering rubbish codec rates, so does it matter if they design their own, or go fur a low level bidder. Maybe the phone GPU ASIC design has one as almost an afterthought. So, another market not so open. They were doing TV infrastructure, as their solutions gave storage and transmission quality a major kick in the pants in those days (the quality of digital TV here with older solutions, were terrible). But that market has probably bought their stuff and dried up a lot aswell. So, where to now. I know with action cams I would want visually lossless motion, so a cineform/vc5 solution would be good. They could do it a lot better than most, and seek the cell design to phone, camera, gpu and TV station infrastructure manufactures. Yes, I'm advocating a version of cineform as a storage and as a transmission technology. As sturage, it very editable, but as transmission, that's difficult. The problem with TV, is to many channels diluting content quality and profit. By have single channels of 19mb/s and above. The industry could get rid of a lot of expenses and concentrate profit. It would take the TV industry a long time to do that and they would have to rework confirm to increase quality per but a lot (there is a formula to part of it), but just one cable win for new setop boxes would put quality way up compared to TV or Netflix. But the issue is, these guys have been going a long time.
Offline

Wayne Steven

  • Posts: 374
  • Joined: Thu Aug 01, 2013 3:58 am

Re: Suggesting: The thin camera (A new camera)

PostMon Dec 04, 2017 5:11 pm

Denis, didn't you read what I wrote about what you just talked about. That was my point (apart from things tanking in the GFC, likely driving down rates as well).

Somebody somewhere wrote that there were enough Red cameras out there to do every production on the planet. If that's so, where are the sales going be? It makes it difficult for others to sell. But what I suggested was to instead, sell to a wider professional audience and consumers. I have given some numbers above. If you go past the cinema crowd into videographers, then you have multiplied sales, but if the gear is also as easy as consumer cameras, and basically a pro version of consumer camera then you can multiply sakes again, giving volume for a special consumer camera better than all others (but really, using this volume to reduce the price of professional product).

Don't think of a professional product as the hand held equipment, there is much more to it that a professional contributes for paid work. Here is a movie on a shoe string budget, it is an hour and a half of watching a shoe string sitting on a surface (ohh, if I was at Uni, that would be a good arts film project, people may rave about it and use it in film studies for years afterwards (I am poking fun at it)). But that is nothing, hiring real professionals to do a good job, gets you a lot further to a quality product. Don't think otherwise, or it will get you down. Now, if it was me, the shoestring would be a Samaria Shoe string, going on his adventure slashing his way to victory with his razor sharp shoe string ends. Maybe he falls in love with a tadpole, but it can never be, because his end is too sharp, and holding her dead lifeless corpes in his arms (strings) he yells out, and the plot takes another turn. Now, in an amatures hands, the 3D animated string interwoven with live action, will look lifeless enough to look like a comedy act, but in the hands of professionals, it can look like it not only exists, but is larger than life (and I'm not talking about closeups or zooms here Denis) with purpose, intent, wisdom and honour (lots of framing, positioning and lighting). In the armatures hands Larry String will look like a string, but in a pros hand, Samaria honor string will look like a Samaria string.

Return to Cinematography

Who is online

Users browsing this forum: Google [Bot] and 8 guests