Home made camera designs?

[Richard Mellor]

Hi everyone

I think most of our investment in equipment will take us into the future. For example, raw capture on hard drives. Storage will get cheaper, but we will not be able to go back to stuff that is shot in 4:2:2 480p and make it 4:4:4 1080p.

The big camera companies have a different plan in mind. A friend of mine from Avid told me that the biggest concerns of one of the big three camera companies was the 'direct to disk' movement . Camera profits are nothing compared to profit fom tapes. This is also why we want raw capture and they want compression.

I am not in any way anti business, but our goal is the highest quality output with the best componets. Their goal is profitiabilty, and compession sells tapes. These cameras will probably be better than offerings of the big three because we don't have a vested intrest in compression.

Very exciting times!

[Steve Nordhauser]

Wayne, you said: <<Another thing is that you were talking capture rates of 50 Mbps? I heard rumors Canon is coming out with a 50 Mbps HDV camera by the end of the year under 10 grand. If that's true, we all surely would have wasted some serious cash, for what the canon would deliver, complete with lens, would probably be so much more worth the convenience of an extremely cumbersome, and not nearly as feature-filled system. Again 50 Mbps is not theater screen quality really. Might get away with it, but . . . you're pushing it.>>

Gotta watch the bits and bytes. The standard I know is 'b' is a bit and 'B' is a byte. Screen resolution is best described in pixels because you may use 8/10/12 bit data with some cameras. One frame of data times the frame rate gives you the average speed. The pixel clock rate will tell you the maximum bus bandwidth.

For example, the Altasens at 30fps 1920x1080. About 2.1Mpix per frame, that is 62Mpix per second but a clock rate of 75MHz. One thing software can do is pack data - putting 3 12 bit pixels into two 16 bit words. This would give a data rate of 94MB/sec (big B) average and 112.5MB/sec peak. If you don't pack, it is 124MB/sec and 150MB/sec or you stay in 8 bit.

What makes this more confusing is each pathway specs things differently - PCI-32 is max 132MB/sec, gigabit ethernet is 1Gb/sec, USB 2.0 is 480Mbps, Firewire is 400Mbps. Then, the PCI bus has 64 bit and 66MHz and 133MHz varients. The typical next step up from PCI-32 is PCI-64/66MHz - 4x faster bus. There are a number of camera link frame grabbers for this bus and one that I know of (Matrox Helios) that does the full 133MHz.

[Rob Scott]

<<<-- Originally posted by Laurence Maher :
8. If possible, the camera should have some form of MATRIX ADJUSTMENT (probably in the software).
-->>>

Laurence, this is probably a really dumb question, but what exactly is matrix adjustment? Is it on-the-fly color correction or something else?

[Laurence Maher]

Rob,

LOLOLOL . . . Someone in the history of time said, "there are no dumb questions, just dumb answers", And I'm about to prove that guy right.

Actually, I'm not exactly sure what the "matrix" settings are. What I know is that on the CineAlta and the like, they exist, and are used along with a waveform monitor/vectorscope to make sure your video signals are calibrated well. This will enable you to make sure you adjust video levels to get the greatest amount of usable data, so you have the most information in both the highs and lows, the best latitude your equipment can provide given your lighting setup. I'm assuming since adobe premeire and FCP have waveform monitor/vectorscopes that you don't need a "matrix adjustment" on the camera. Somehow you can probably pump signals through these software programs (or hopefully whatever else we would use) via laptop on set and then adjust under "levels" in the software. Of course, I don't quite know the intracacies of just how much data will be lost simply by pumping it thorugh the software to calibrate it. But then, I don't think you'd have to do that, as long as you could just get a feed and monitor the signal before it is fed into the capuring system. Really, it's been 10 years since I graduated, and haven't used a vectorscope or waveform monitor since (don't need that for shooting 16mm film or dv video at weddings), so I'd have to relearn exactly what to do with them.

What I do know is this.....

It is suggested in Scott Billip's Digital Filmmaking book, that no HD for film / theater release production should be without an on-set waveform monitor/vectorscope for calibration, unless you are looking to waste some 20% video image quality. Of course, no man is the end-all, but compared to me, he is.

Where I get this info from is SCOTT BILLUP'S: DIGITAL MOVIEMAKING. I learned more about HD from that book in a few days than I think I learned about video alltogether during my school years (technically, that is) Of course, times have changed sense then, but In my opinion, this should be the HD textbook used in colleges today. Great freaking book, really.

[Laurence Maher]

By the way, someone asked about my price point.

I'd have to say that the invonvenience of Linux to Mac or PC to Mac or Linux to PC would be a monumental thing. This, of course is only based on the problems I've incurred on PC alone. The way I see it is, if I can't get a PC to edit properly, then how could I possibly get a Linux device to work without prior Linux knowledge, let alone interface the linux well with my PC editor after the fact. So I'd be willing to stick in for 10k if you're talking a camera/capture system that would give me files I could read in Mac FCP. That's a lot, I know, but honestly I can't afford the problems of systems duck-taped together.

I'm can't be for sure where everybody else stands. It seems as though everyone is leaning towards some kind of linux setup. I think the best thing is for everyone to give a similar list of what they want performance and price wise. Not to sound to bold, but is the system you guys are talking about really possible for 5k? I mean maybe it is as a semi-working prototype, but in the end, if some are doing high-profile commercials and some are doing features etc., I'm not sure I quite see that 5k will get us where we'd like to be before a year or 2's time, and I predict by that time other options will be available on the market (not the canons at all, but there's buzz about other companies making HD video out of still cameras etc.).

By the way, check this out. This thread says Black Magic Design has got some pretty good capture boards out for as little as $1200. If that is true, I think it's worth a serious look.

http://www.dvinfo.net/conf/showthread.php?s=&threadid=27278

[Laurence Maher]

Whoa! . . .

Just got this info from another thread here under Alternative Imagaing . . .

I really hope my excitement is warranted here. For some reason, I've got a bell ringing that tells me you guys might be able to get one of these cards to work with the upcoming cameras. There's some 10 bit HD cards from Black Magic that are currently on sale for as little as 1k !!!! It's SDI / PCI-X 133MHz!!! , but it works with PC or Mac (everybody's happy) !!!!! Other options too. The HIGHEST on the site I saw was $2500, and those of us that wanted to go that way . . . . ohhhh it's a MONSTER !!!

goto:

http://www.blackmagic-design.com/index.htm

Check out the various options and tell me, Laurence Layman (lol) why it will or won't work, please!!!

Enjoy!

[Jason Rodriguez]

Hey Wayne,

Matrix settings are entirely pointless if you're capturing the RAW information off the A/D converters. In the Cinealta, etc. the color matrix settings (which is basically how the signal is distributed on a various number of vectors on a vectorscope, whether that be 12-pole or 16-pole, etc.) are applied after A/D conversion, so they're working with the same data that you're getting if you getting all the information uncompressed off the A/D converters. Typically matrix settings are what's used to adjust the conversion from 12-bit RGB from the A/D's to 10-bit YUV for the HD-SDI out. And believe me, you don't want to be grading on-set. That's going to require some VERY expensive equipment to make sure that WYSIWYG. Having all the information and then grading in post is perfectly fine, that's what the Viper does, and it can look very film-like (Viper RAW files bypass any matrix settings, sharpening, etc.).

[Laurence Maher]

Cool about the matrix definition Jason,

So you're saying it's better to capture RAW to like 12 bit and then use FCP or similar software to adjust matrix settings via it's built in waveform/vectorscope and colorize down to 10bit? Is this correct?

Okay, you obviously know more than I do about the subject, but then I guess you'd have to be running some seriously powerful software for it to be able to take a 12 bit file and convert it wo 10, wouldn't you? I mean, pumping it through most softwares alone will be a bottleneck, right?

I'll write Billips and see what he does when shooting RAW.

By the way, his site is great:

www.pixelmonger.com

Tbanks!

[Jason Rodriguez]

Combustion or even After Effects can handle these files easily. Shake is a good option too.

BTW, there are no 12-bit image sequence files, typically they'll either be 10-bit log DPX files or 16-bit TIFF's (with padding so they'll look dark).

But yes, the built-in color correction engine of Combustion is plenty for manipulating these files. My personal preference though is Color Finesse from Synthetic Aperture. That plug-in will give you amazing ability.

Also, you're going to have to run these image sequences through a conversion before they can be used propery in FCP. Typically your workflow is going to include first ingesting and doing basic color-correction on the files/converting to low-res Quicktime, and then editing in FCP, with some sort of window-burn on the quicktime so you can match it back to the right file sequence. Then after editing, you'll go back and do a visual match-back to the proper image file sequence, and from there do your final color-correction. That's the match-back method.

The alternative approach is a modified telecine approach, and that's to take the RAW files, do the complete color correction on them, and then convert them to a hi-res Quicktime file that will become your new master. The only caveat with this approach is that you might be spending a lot of time color-correcting the files that you don't want in your final piece.

My advice: Learn Shake, because you can automate a lot of this match-back process with some clever scripting, apply automatic color correction, etc. It's also a pretty good app for the final conform, and it'll read 16-bit TIFF's nicely. Additionally the editing interface in Combustion will make for a nice conform tool, although you can't script that interface, so they'll be a lot of hand-holding through the conversion process. Another app that might come in pretty handy is Sequence Publisher from IRIDAS. You can use that to automate the inital low-res conversion process, with window-burns, and then to automate the hi-res 12-bit Quicktime process with automation also. Then duplicate your project in FCP, name the hi-res files the same as the low-res files, and after unconnecting all your media, reconnect it with the hi-res 12-bit Quicktimes. You can then export an XML file and import that into Color Finesse v2, which will be an awesome CC app, and do your final CC there, render at 12-bit quicktime and import back into FCP or whereever for final output (if you have the hard-drive speed required for this stuff).

Sound complicated-Oh yah, you bet it is, that's why the Kinetta's going to be so awesome, it'll have it's own built-in real-time color corrector with HD-SDI and dual-link HD-SDI ouputs to make life much easier for everybody. You'll also get lots of metadata embedded in your 10-bit log DPX files if you decide to go the image sequence route, and stuff like Sequence Publisher will like that a lot more than TIFF's from an industrial camera that have no embedded timecode, etc. metadata. Additionally Jeff's going to have much better bayer algorithms than I've seen for these industrial cameras.

So things could change here in the near future, but if my post is anything to judge by, if you want to do the industrial camera approach, be prepared for A LOT of rendering!

[Rob Scott]

<<<-- Originally posted by Jason Rodriguez :
So things could change here in the near future, but if my post is anything to judge by, if you want to do the industrial camera approach, be prepared for A LOT of rendering! -->>>

Absolutely true. If you need a simple workflow, professional features, dual SDI, I would definitely recommend a Kinetta or something similar. Don't rely on a homebrew camera that may involve a much more complex workflow.

I suspect that SDI output will simply be too expensive for our project -- our goal (well, MY goal :-) is to use commodity hardware and standards.

If our project *ever* becomes feasible to use on a high-profile project, it will take a lot of work and probably a few years. In the meantime, I think our project will be of value to those of us with VERY low budgets (i.e., a bit more time than money).

[Laurence Maher]

LOLOL

Oh come on guys. I have trouble getting my matrox le to function without crashing on pc, let alone deal with tiff files and biff files and whatever iff files. What I've been saying all along is that anyone who doesn't want to take 3 years making a movie is going to have to fork over some . . . not huge . . . but some . . . cash . . . to make this user-friendly. Here, let's look at it this way....

FOR 5K. . .

. . . By the time any of us build this user un-friendly camera, shoot a film, convert all the files, color correct, render and render and render and then finally edit or whatever, 3 years of our lives will have gone by and the only thing we'll have to show for it is 1 feature film that probably isn't that great because most of what we were doing with this system was experimentation do to lack of features. In the end, after 3 years of hell, you'll also realize just how much time you spent fixing things and patching things and that you probably could have earned a good 30 grand with the amount of fix-it hours spent . . . OR . . . .

FOR 10K

. . . We could put together a system, complete and user friendly, that captures directly into the files needed for editing, open them immediately with user-friendly editing software, and edit immediately, with little rendering time and no problems so you can focus more on the artisitc aspects of your feature film, which you got done from script to screen inside of 1 simple and enjoyable year . . .

THOSE NUMBERS AGAIN . . . .


5k spent = 3 years of your life in hell, 1 finished film, un-reliable equipment your now so pissed at you never want to see it again.


10k spent = 1 year of your life in fun, 1 finished film, reliable equipment you can use again.


Really, by the time 3 years are up, if we go with the 10k option, we'd all hopefully have 3 movies under our belt and be moving on to bigger and better things anyway.

I don't know how many of you have written/directed/produced a full feature of film level magnitude before, but I have, and BELIEVE ME. The last place you'll ever want to be is hanging around in post production hell. This is coming from someone who HAS spent 3 years on a feature in such a fashion. IT . . . SUCKS.


Come on guys, you can't tell me you really want to spend that time and effort.

lolololool---We'd get just as much done all pitching in on 1 camera then traveling around the world helping each other shoot 1 feature a piece!

Really, what is a bigger gamble, forking the cash (for some that means going into a little bank loan/credit card debt, for some, working like an overtime dog several months this year, for some, just cutting back on what you usually don't need to spend your money on, for some, pooling together friendly funds or finding investors for your feature, whatever) or burning those 2 extra years away of your life. You'll live a maximum of 100 if you're really lucky.

Hell, it may not even be 10k, just 7, 8, 9, we don't know for sure.

Everyone here so far says SDI is expensive, but Black Magic Design I think just came out with a card we could use for minimum 1k, maximum 2.5k (unless it won't work for some reason). And it works with both PC and Mac I think. (Unless you guys are talking output to a medium that SDI is very expensive . . . I don't know)

check it out:

http://www.blackmagic-design.com/index.htm

I'm really not trying to knock on anyone/anything here at all, I just think we need to take into consideration the practicalities of our decisions with this. Of course, this is from a feature filmmaker's point of view, not a short project maker. For guys doing shorter projects, some of these arguements would be moot.

I WUV ALL OF UUUUUUUUU!!!!

Comments?

[Rob Scott]

<<<-- Originally posted by Laurence Maher :
I'm really not trying to knock on anyone/anything here at all, I just think we need to take into consideration the practicalities of our decisions with this.
-->>>

It's not for everybody :-)

I don't think it's going to be as bad as you make it sound, but this *is* homebrew stuff. Your same arguments could be made about the GG thread -- why build it yourself? It's a pain in the rear! Why not just pay $8k for one that works?

I'm just saying that there is no telling when this would be a polished, fast, easy-to-use, problem-free solution ... if ever.

Luv U 2 !! :-) Seriously, I appreciate your comments. No hard feelings at all.

[Steve Nordhauser]

Laurence,
Clearly you got your tailfeathers scorched recently. Being outside your industry but in a position where my tools can cost big $$ (a good scope starts at $3k and goes to $50K), I understand the frustration. It is probably not cost effective for each person to venture on their own. That is probably the greatest strength of a forum like this. A grouping of individual skills - some camera, optics, programming, work flow, post production knowledge mixed together.

If the overhead for each person building similar systems is only a 5% learning curve, instead of 75% if they work independently, low cost cameras make sense.

Personally, sales to indies are not going to make me rich - lots of overhead for single unit sales, but this is a place where I can do some helping. Remember not to assume targets are stationary when you look at this. The commercial HD camera will come down in price. Our prices will drop or capabilities will go up. In three years you will be looking at multi-camera systems, long term archiving of raw films plus post processing, everyone will be at 1080p, 12 bit for a reasonable price. I think people in Home Made are the ones that are a bit out in front. If they stand still, technology will pass them.

[Richard Mellor]

camera link card $695

select camera link and click search

http://www.opsci.com/index.asp?pagetitle=FramgrbrSearch&exten=asp

[Laurence Maher]

That's EXACTLY my point, Steve. . . .

A film that takes 3 years to make from this point really acheives nothing. For in the same time, the technology will catch up (perhaps even surpassing the systems we have created before our films are near finished) and the idea of building a hand-crank home grown system will be like trying to take a mule to work now days. You never know, it may happen even faster than we think, and we're stuck, having spent our financial wad on mules while everone around us now is in a Ferrarri. If we are going to do this, we need to find a happy medium between COST and other factors like TIME efficiency, for those of us making features anyway.

Just IMHO