4:4:4 10bit single CMOS HD project

[Ben Syverson]

True, if I wanted to build a very low cost camera with a direct HD interface, fixed res viewfinder, fixed file format - I could do it very well with FPGAs (and the right designer).

Yup. I just don't know why you need flexibility. Mark the product up $1000, move a thousand units (which would be easy), make a million dollars. If the needs of the market change, make a new camera. Lather, repeat, retire.

I'm not advocating any pre-processing of the image; do some simple lossless compression on the raw data and pass it off to an in-camera RAID, a la Kinetta..

[Eric Gorski]

I'd buy a system for under $5000... at $10,000? No. I'm sorry but if you want to appeal to the type of independent filmmaker who would stray from big established companies like sony/canon/panasonic then you have to be competitive on price.

The thing you have to understand about these camera systems is that they'll never be mainstream. There's too many unknowns. The functionality of this $10,000 piece of equipment is dependent on inconsistent things like the speed and stability of your computer, the compatibility of the software, and the reliability of harddrives which are seen by many to be unreliable. On top of all this, nobody wants to drag a PC around with them on their shoot and people will be hesitant to give $10,000 to some company that they've never heard of. At this pricepoint, the risks start to outweigh the rewards. Sony offers a camera for $3000 that is 1080i, shoots on cheap mini-dv tapes, includes a lens, its footage can be edited in realtime with realtime effects on most NLEs, it records sound, has a shit load of features that will never be available on these cameras and with a $500 micro35 lens adapter shoots some pretty convincing filmlike images.

Your profit margin might be higher with a $10,000 system but I promise you that you will sell so many more if its a few thousand dollars. It doesn't have to be 1080p uncompressed. Honestly, I think the most appealing aspects of a camera system like this are the adjustable aspect ratio, variable framerates (super slow-mo), interchangable lenses, and the opportunity to have something that you just plug into a laptop and you're good to go. The higher the resolution and the less the compression the greater the strain on your system until it bogs the process down.

Offer a camera with software for less then $5000 that can run from a laptop and you will make a killing. I know because I'm your target market. You're trying to appeal to low-budget independent filmmakers. I think the thing you have to realize is that at $5000 you're competing with minidv and mini-hdv cameras. At $10,000 you're competing with 35mm film. You can shoot a feature film on 35mm for less then $10,000.. The other thing is that the technology is moving so fast that in another year or two this $10,000 camera will be obsolete. You have to offer something cheap and quick.

This is just my two cents but independent filmmaking is all about weighing the quality vs. the convenience vs. the price and you have a rare opportunity to offer all three, if you don't get greedy.

[Ben Syverson]

Word.

Steve, the indy market is $5k and under. The pro market wants real pro features, and will never be comfortable shooting with an industrial cam. $10k is no man's land.

[Matthew Wauhkonen]

You can shoot a feature on 35mm for under 10 grand? Please inform me as to how.

[Aaron Shaw]

I believe he was meaning digital.

[Radek Svoboda]

Originally Posted by Steve Nordhauser

The main issues are bandwidth - camera head to the PC, PC bus in and back out to the RAID and RAID continuous throughput.

1920x1080, 12 bit, packed 24fps is about 75MB/sec average. This is a 64bit/66MHz PC, 2 drive SATA RAID. No compression in recording. Sometime later maybe.

Our camera head should be equal to or better than another Altasens 3562 design. We will have a 3570 when available.

On Bayer, I've heard the 25% number before, also 30%. It depends on what color you look at. For red or blue primary colored objects, it is much worse. For something with a mix of RGB, it might be nearly zero loss with the proper algorithm. So, I'll give you a very qualitative 'it depends'.

Steve,

I'm not computer expert, I'm student filmmaker. Please explain me following:

1. Resolution should be equal to Sony CineAlta recorded to tape 1440x1080p because Bayer is known lose about 25%. Or does Bayer loses vertical resolution too and resulting effective pixels be 1440x810?

2. Color will be better, so will lattitude, because is 12 bit, Sony is 8 bit system. If you record at 10 bit, how better lattitude will you have compared to CineAlta?

3. The shutter is some kind of rolling type, which creates unnatural effect on movement so mechanical shutter must be integrated into camera? How much extra is with the mechanical shutter? What is mechanical shutter reliability, noise level?

4. Sony has X300 and X310 box cameras with HDSDI, even HDV output. You can feed HDSDI to computer with dual compressor chip and compress everything real time 6:1 with Prospect HD. Is there way get 10 bit signal out your comera, use same computer system? Would RAID still be needed? Filmmakers need complete solutions. Why don't offer small factor dual processor computer with Prospect HD with your product, and LCD? I posted info on such computer in HDV main acuisition forum.

5. You have very nice product but are in group of buyers that is buying 3K USD HD cameras, 500 USD 35 mm adapters, makes wedding videos or student type films on most part. That why your low price is not appreciated.

6. I think you should develop your product further. There will never competing products from Sony etc., because they need sell 10x more expensive products, which market bear. If product is less convenient than Sony's product, no big deal, indie filmmakers will buy, if image quality is there.

7. You have finalized system; you should ask DVinfo to give you forum on that product. It seems that here your product not very appreciated.

8. If I were you, I would continue investing in indie filmmaking market segment. It would be very profitable. You just need market it to the segment where buyers are.

9. Is chip 2/3"?

10. You should work with P+S, have them develop series 400 35 mm adapter for it.

11. You will not gain profits by lowering price but by adding features and promotion to different market segment, product improvement, cooperation with DP's who give their opinions.

Radek

[Kyle Edwards]

Originally Posted by Radek Svoboda

10. You should work with P+S, have them develop series 400 35 mm adapter for it.

Expect to pay 10g more for the product if so.

[Eric Gorski]

you can indeed shoot a movie on 35mm film for under 10grand. The movie 'Primer' was shot for around $7,000 on 35mm and it happened to be one of the best movies made last year...

http://www.primermovie.com/

[Radek Svoboda]

Originally Posted by Kyle Edwards

Expect to pay 10g more for the product if so.

I would not buy P+S. I would buy one of cheap ones but SI needs it if they want attract rather pro customers. That's where their profit will be, once they can start make money there, they can develop better cheaper products to us.

Radek

[Matt Champagne]

I think the thing you have to realize is that at $5000 you're competing with minidv and mini-hdv cameras. At $10,000 you're competing with 35mm film. You can shoot a feature film on 35mm for less then $10,000.

I couldn't disagree more. At $5000 you are blowing minidv out of the water, and definately not getting what you deserve for your efforts. Maybe at $35,000 we are talking about competing with film (and that's low)...but while a $10,000 film can be made...you should budget the extra $15,000 for therapy.

However more importantly...at 4:4:4 10bit...you are competing with film in image quality and that is really saying something for $10,000. Especially since once you have the equipment paid for (lets not forget that equipment can be rented or financed while filmstock itself cannot) then your tape or other capture media is about 1/100th of the cost of film...litterally making movies for pennies on the dollar.

This is just my two cents but independent filmmaking is all about weighing the quality vs. the convenience vs. the price and you have a rare opportunity to offer all three, if you don't get greedy.

That's alot easier said than done. These guys have been working on this project for some time...and all that labor is worth something. Personally, I would charge more (but also I'm lazy and getting me to do any work means overpaying me). Also there is the issue of production numbers. Too much demand can be a very bad thing if they were to create such a camera and sell it for $5000. Could a small company fulfill 30,000 orders in one week? Supply and demand curves aren't perfect curves in the real world...they are stair stepped at the various levels of production efficiency that can be achieved with real world machinery. The amount of money it would cost to fulfill the demand a $5000 price point would require the use of a production facility of unreasonable proportions for a risky venture such as this. They aren't Sony....they just have better engineers :}

[Eric Gorski]

Originally Posted by Matt Champagne

I couldn't disagree more. At $5000 you are blowing minidv out of the water...

that was my point. at $5000 you are offering the best option. the thing about these systems is that they're not complete out of the box. a $10,000 camera still requires a very high-end computer and lenses and tons of storage. we're talking close to $15,000+ when all is said and done :(

This is the thing that concerns me about all this. Almost a year ago I saw Obin capturing 14fps from a thousand dollar camera at uncompressed 8bit and that footage video looked excellent. It just looked so much more vibrant and organic then standard high-definition. There were problems with that camera concerning the shutter, etc. but it was close. Now a year later all anyone is focusing on is these $10,000 solutions. I don't know how i'm going to capture 4:4:4 12bit uncompressed. Obin spent a year trying to build a computer that could do it and it still hasn't happened. How about focusing on a data stream that the average person can capture to their laptop? If you could get uncompressed 8bit from a thousand dollar camera, that still kicks the shit out of high-def. Go ahead and mark that one up $4000 and i'll still buy it. Maybe i'm totally clueless to how supply and demand works but I know how much independent filmmakers are willing to spend. I guess maybe a company like silicone imaging would rather sell 10 cameras at $10,000 then 100 cameras at $5000 because its less work to make?

[Rob LaPoint]

I agree with Matt, developing something like this is an unbelievable amount of work. Just flooding Steve with posts saying "cut the cost in half" is not helpful. The Altasens camera that they are basing this system around costs $5000 for just the box, so unless SI is supposed to throw in the software and buy a capture card from Epix to give away there is no way this camera will cost under $5000.

That being said, it is a very interesting point of what niche this camera would fulfill. I have always considered this project as a 35mm substitute, and when you shoot 35mm many things must be considered. It is not an easily portable solution and it takes 3 people to run a camera. I see this camera being very similar to that. The only real difference between the two solutions is that with this you do not have to pay the extrordinary cost of developing/color timing/printing/ect..

I am a professional videographer and would LOVE to have a shoulder mounted system shooting 1080p uncompressed that ran off a battery for 6 hours, but right now its just not gonna happen. I believe that a totally hardware run solution is the ultimate answer but that takes time and tons of money, Obin is working on it. The software solution gives a solution to us now however. The one thing that I am wondering is the possibilities using the cheaper SI-3300-RGB, then it would be possible to have a solution for around $6000 I am guessing. But the real stitch is Panasonics new baby coming out this fall that will shoot in DVCProHD straight to solid state memory. That camera will be about as expensive as this project and it won't have as nice images, but it will be MUCH MUCH MUCH easier to use.

I am really pulling for this project in whatever way it manifests itself I have been following for over a year. Good luck to both Steve and Obin, but I think it will be a lot easier for people to pay $9000 for an Altasens if they can buy a $5000 SI-3300-RGB solution, try it out, and realize they want the upgrade of the better chip.

[Noah Yuan-Vogel]

Ive mentioned this idea before, but didnt get a lot of responses. Everyone is talking about $5000-10,000 camera systems which would be in the range to compete with the upcoming panasonic and jvc hd(v) cameras, but couldnt it be done a lot cheaper?

Personally I would not spend 10s of thousands on a homemade camera due to the high probability of something not quite working correctly and lack of funds in the first place. Also, it seems to me that there is a point where it becomes exponentially more expensive to build your own HD camera system, that happens when you have to build portable raids or use cameralink. This means if you want to keep costs down use gige and keep it to a data stream that is possible on a single HDD. Tomshardware shows there are plenty of big, relatively low cost 250-400GB 3.5" SATA drives that can run with a MINIMUM practical write speed of 37MBps.

The silicon video 9T001C camera from EPIX uses the same micron sensor as the si-3300 (steve verified this) and it comes bundled with a gige framegrabber and cable and software for $995.

Hook this up to a small formfactor (probably shuttle) computer with an athlon64 winchester core CPU (from what i understand the best power consumption in its class, although i havent looked into the new venice core) which all together is likely to cost around $600 once you add in ram etc. Make that more like $1000 if you add in a 12v power supply and a high capacity heavy duty battery belt.

However, assuming the bandwidth limitations are the 37MBps hard drive and 33MHz PCI and the gige. It should still be possible to get 1920x800 8bit at 24fps for 2.35:1 and you still get the flexibility of programmable framerates, ROI, binning (for better than bayer color?) and 10bit color so you can choose your balance of color, resolution and framesize. Also it gives you compatibility with various really fast c-mount lenses for shallow DOF. So throw in another $500 for a nice set of angenieux f0.95-f1.3 lenses for up to a stop and a half better low light and shallower dof than any 3ccd lens.

The main problem as i see it besides getting used to XCAP (which i havent heard amazing things about) and finding a good resolution to shoot at being limited to 37MBps (although i wouldnt want to have to deal with much more than that in post anyway assuming 3x more data for rgb but then lossless ~2.3:1 compression would bring it to about 50MBps). ROI at 720p would probably leave the usable sensor area closer to 1/3" but would allow more flexibility in frame rate and the ability to use 10bit color and that could be fixed with a $100 homemade 35mm GG adapter (but that would likely introduce a loss in sharpness and a lot of loss in light).

Overall cost ~$2000-2500 for a camera that could run tethered (gige can get pretty long) to the computer or possibly be shoulder mounted. A homemade shoulder mount that could optionally hold the computer case and/or the camera head would be few more $$, but then it might also be possible buy (or use since i have a bunch sitting around) an older CPU and shuttle barebones as I am not sure of the system requirements for capturing over gige in XCAP.

Anyway I'm kind of obsessed with this idea and am seriously considering going ahead with it, but tell me if you see big problems I missed. But I tend to consider 1080p to be more than I need and would probably prefer to shoot in 1.85 or 2.35 aspect ratios anyway. tapeless 10bit uncompressed 24p recording from a 1/3"-1/2" CMOS chip with fast manual movie lenses seems pretty awesome to me especially at a price less than a dvx100. Of course the dvx100 takes less than 30sec to power up... but then it cant double as an editing station :)

Oh yeah and another $200-300 for the 7" touchscreen LCD and whatever shipping charges I didnt take into account in the prices.

[Noah Yuan-Vogel]

eric, I checked out the primer movie website and it says the film was shot on super16mm and blown up to 35mm.... and its only 77min, which im sure cut the cost down a bit. I'm pretty sure shooting a >90min 35mm feature would be pretty impossible for $7000 or even $10,000

[Adam Burtle]

Originally Posted by Noah Yuan-Vogel

I'm pretty sure shooting a >90min 35mm feature would be pretty impossible for $7000 or even $10,000

this is all off the top of my head, so i am sure i'm not exact on figures, but i'm probly pretty close. 1000' of vision2 is around $600 (something like $.60/ft iirc). 1000' mag of 35mm is roughly 11mins. (90min / 11min)x$600 = ~$4900. So.. even if you already owned all the camera gear, editing equipment, etc.. and got everyone to work for free, with even a 2:1 shot ratio you'd be pushing $10k.

(someone feel free to correct me if my math was insanely wrong.. it is 8am and i havent slept yet)