waiting for that image post!! shoot some stuff with LOTS of colors in it for the tests/image grabs! need server space? I can give you some!

So Juan.. what is the answer then..

either the PAL DVX doesnt share chips with the NTSC one..

the PAL camera somehow upconverts the raw image to the pal 576 standard..

or you are capturing a "raw" signal that has somehow already been downsized from a 576+ signal?

I 100% believe you when you say that is the size of the frames you've captured, I just dont understand how that can be, when pal cameras do 576?

::very excited to see some frame grabs:: ..or even some short video. If you need somewhere to host the video or frames (regardless of file size), I can handle that for you.. just drop me an email (adam@adamgeek.com)

Adam:

I am 100% that what I'm getting is the raw CCD size.(unless the service manual I have is part of some evil scheme by panasonic).

However, I think i've found evidence that the CCD's are indeed different between the NTSC and PAL versions, other than the frame size. The CCD sensors do not seem to be square, but rather have the NTSC aspect ratio...the raw frames of resolution charts having circles in them come out as slight ellipses, which are turned into circles when the NTSC pixel aspect ratio is applied. The difference is small, so I didn't see it until I used a rez chart.

I am very excited about this thread. I just read it all, but some things aren't clear to me. Maybe you guys could explain a bit.

So actually the quality of a raw footage taken by a Panasonic DVX100 or a smilar camera is great, but because the camera saves the shot on the DV format is loses quality. So now we are trying to do something so we can direct store the raw format on for example on a HD without losing quality?

But then, if so.. when I search on google there are many products (especially the brand matrox pops up everwhere) which will transfer a footage from a mini DV to the PC.. why not use these then?

sorry, kinda new to this :)

Carlos,

Because those products store the compressed DV stream that comes out of the camera, which is the exact same thing that is saved to tape.

Manufacturers do NOT allow you to access any other data stream other than the compressed DV stream that is recorded to tape, or the analog output which is horrible.

BTW, sorry for the lack of updates, right after i got done with my 3 tests i left for spring break to visit my family since i haven't seen them in a while...will get back to work soon...it is almost done.

Juan

i'm very curious to see some examples too. this threas is breathtaking! i thought that i already found some fantastic threads here on dv i community, but this one is just unbeleivable!

juan,

does this mean - in your opinion, that i.e. beta sp, or digibeta can give us even more superior pictures? if dvx panasonic is cappable to "see" more then it "can" record - does it mean that broadcast cameras can "see" in "raw mode" more than just pal/ntsc, say - something closer to hd?

just curious

filip

Fillip:

The output of this experiment yields 4:4:4 uncompressed color in ANY camera. however, there is one thing that you do get with a more expensive camera: better CCD's. This translates into larger frame sizes(i.e. closer to HD) and less noise...also with larger CCD sizes you get a shallower depth of field that better mimics film.

So yes, you do get advantages of doing this procedure on a better camera, however, i think it is up in the air how good actual DigiBeta(4:2:2 3:1 compression) will look compared to the raw DVX output.

Juan

so, do you think your technique could be adapted to other cameras then? any chance of releasing pictures showing us the inner workings for DIY'ers?

Not to rain on anyone's parade, but if you're gonna go through all this trouble, why don't you just buy the CCDs, circuit boards, lens, etc., and design your own camera? You could probably do a better job than Panasonic, and you'd have exactly the features you want.

Peter,

As soon as you find a source for state of the art CCD's such that I can buy a small amount for less than $10,000 a piece, i'll happily do it.

Good Luck! :)
Juan

:)

My understanding is you can get them via the astro-photography community, which frequently mod's their telescopes with CCDs and cameras. They are expensive, but most are much under $10,000.

Peter,

I am an astrophotographer myself, and unless i'm mistaken somewhere, there is a huge difference between a broadcast video CCD and the CCD's commonly used for astrophotography. In fact, there are all sorts of different types of CCD's.

Up to now, what i've seen is that manufacturers such as Panasonic, do not sell the cutting edge CCD's they are using in their cameras(such as the DVX). They can afford to sell the DVX at a relatively small price because they build the CCD's themselves and in large quantities. They aren't dumb enough to sell the heart of their awesome cameras for other companies to exploit, and probably outsell the DVX or their VariCam for example.

However, nobody can stop us from breaking the camera open once we buy it.

Once again, if you find a concrete source for progressive scan video CCD's of matching or better specs to the DVX's, i'll be all over that. Trust me, i thought about it and researched it before starting on the DVX mod. :) I mean, i've found some progressive CCD's but they are horrible, nowehere near the output of the DVX.

Juan

Are they unable to be adapted for video use instead of stills? If so that's a shame that you can't otherwise get good CCDs to use to mod your video cameras.

So what are all the different CCD types besides just the number of pixels? The only technology difference I've heard of is CCD vs. CMOS in still cameras.

Peter,

Besides the frame sizes, there are different things to consider like pixel aspect ratios, sensitivity, chip sizes, and the rate at which the chip can read complete frames and how it does it.

For example, for astrophotography, you probably want a very sensitive CCD with the least amount of noise possible...most people also incorporate some method of cooling to reduce thermal noise. however, the CCD might not read enough complete frames per second to actually put out video...which is usually the case with cheaper CCD's. On still cameras you can have tiny size chips with large pixel counts...however the readout times are long because you only want one frame.

So in essence, CCD's with good sensitivity, low noise and fast readout times, as needed for video, are among the most expensive to come by. i think camera/chip manufacturers are plenty aware of how easy it is to hookup CCD's to A/D converters and record the output, so they are doing their best to keep the cutting edge stuff innaccessible to all but the large companies who can pay them the $$$.

Take for example, some video cameras that record DV, but also take hi-res pictures. The CCD's have the pixel count, but it probably can only read a fraction of those pixels fast enough to put together video....the complete frame is read out slower, which is only good for stills or extremly slow, SLR burst-like video.

Sorry if this sounds a bit like gibberish, it's been a long day...

Juan

Fascinating stuff Juan, thanks. (and I did understand the technobabble :) )

Too late, you'll have to start this experiment over with HDV chips.

http://www.camcorderinfo.com/content/sony-hdv-prototype-camcorder-03_17_04.htm

I'm not an owner of DVX100, but I find this thread very interesting. Someone mentioned grabbing their own CCDs , lenses, etc. and building their own camera. Rockwell (I think) is selling their own HD CCDs (actually CMOS image senser) and they will definitely sell prototypes for something like US$550 I believe.

And for other people who are getting excited about this MOD, there will be a camera announced at NAB, an HD camera that reads the raw data off of the sensors directly to a hard disk. It is designed to mimic a film camera's operation (someone finally gets it!) and the signal will be "graded," similar to a DP grading a film with a colorist, in the computer in post. I'm not sure about cost, or when it will be available, but it should be an interesting camera.

mg

Matthew, at least one other camera (the viper) already exists like that, andprobably a few more. I believe it was mentioned earlier in this thread. I'm really looking forward to attending NAB, and getting a chance to see all sorts of cool new toys i can't afford, haha.

I know all about the Viper. AFAIK, this camera will operate like a film camera (ie. push to run, push to stop, use manual lens to control focus, exposure, etc.). Also, I don't believe it will be gigantic and unwieldy like the Viper (ie. the hard drive is mounted in like a film mag and swapped out when full, not some tethered monstrosity like what I've seen of the Viper).

mg

You can read about this camera in the last couple of paragraphs here http://millimeter.com/ar/video_roll_own/index.htm.

guys what about the CMOS chips that can be had from Rockwell for about $500 each? they are being used in the JVC HD box cam NOW and are HD res.....

check this
http://www.rsc.rockwell.com/procam/

I would think this is the way to go ...why not make our own cameras?

Here's the thing about that:

I am unsure whether this setup will be worth it with a color CCD like the JVC uses, instead of 3-CCD's. This is because in a mosaic pattern, adjacent pixels are combined to approximate the color of the center pixel, so there is a PHYSICAL approximation of color going on, which is why going 4:1:1 or even 4:1:0 is not that bad.

When doing 3-ccd's which are all monochrome, there is precious information in each pixel, and doing 4:1:1 quantization looses quite a bit of information.

What is the semiconductor site for these CCD's? I'll check...the main points are that the chip needs to be progressive, monochrome and read fast enough to do 30 OR 24fps.

Remember also that another problem is the optical front end...where do we get prism assemblies for 3-way beam splitting?

Juan

I would say your best bet is to keep plugging away at your dvx100 mod and forget all else...we are willing to pay you MONEY for your mod NOW

oh don't worry about that :) I got back sunday noon after 24 straight hours of driving from Miami, and then i had to work until 5pm, so i was 'out of order', but work continues as soon as i get back from work today.

idea. hmmm....put a rockwell ProCamHD CCD into an old Arri type 35mm film camera!

This would cost soooooo much. I think the point here is that there are great cameras out there, they are just way to expensive....the point here is to look for a cost-effective solution and get the highest quality possible.

Whoa, never mind, i've been thinking about the wrong JVC camera! I looked at the site and these are clearly monochrome CMOS sensors. I am going to inquire about pricing, maybe there is something here.

Probably, however, they will NOT sell a small amount to an individual for anything less than thousands of dollars.

Juan

<<<-- Originally posted by Juan P. Pertierra : Whoa, never mind, i've been thinking about the wrong JVC camera! I looked at the site and these are clearly monochrome CMOS sensors. I am going to inquire about pricing, maybe there is something here.

Probably, however, they will NOT sell a small amount to an individual for anything less than thousands of dollars.

Juan -->>>

Don't be to sure about that. Especially if you can convince someone it's in their best interest to keep prices at some point where you will choose to look at there CCD's further. It could also help if they understood you may represent a large number of people who are interested in your current work on the DVX.

BTW - Toward the bottom of that linked page, they offer a single CCD for testing setup to interface board for the PC. I suppose you've already noticed though.

-Rodger

yes like I said $500 each

I dont know how it will help.. but i'll be attending NAB, if you need me to stop by the booth of those Rockwell people (who make the procam HD chips) to talk to them, take pics/video, etc.. just let me know.

I dont know how easy it would be to interface something like that to an existing camera setup.. but if you have the technical knowhow that would really empower dv filmmaking on a budget, heh

Assuming those Rockwell chips can be obtained for a decent price, a simple camera can be made from them very easily.

The cool thing about the rockwell chips, is that they have built-in timing/AD conversion, so the chips actually output digital signals. So right out of the chips I am getting the signal that I am getting out of the DVX's A/D converters.

You can make a camera by obtaining some casing that can house optics, and mounting the chips accordingly(only hard part is obtaining the 3-way splitter). After that, we can handle the output from the chips just like the output from my mod.

Of course, the bandwidth will be much larger since its HD, but we can always go dual FW800 or maybe SDI. I've tried to stay away from SDI because the recording equipment is expensive, and besides i haven't found the specs anywhere.

I emailed rockwell but haven't gotten a reply regarding low-volume sales...not sure if they will be carried by any online retailer...CCD's/CMOS sensors of this kind are generally hard to come by in this way.

Juan

would it be feasible to just mount them in an XL1S.. rather than making a new casing.

i can stop by their booth next month and talk to them if you want.. they might be more receptive in person.

Assuming you gut all the electronics(which you won't need), the main constraint is the fact that the 3-way beam splitter in the XL1s(or any 3 chipper) is probably made to the chip size/aspect ratio that is being used. Therefor, since the rockwell chips are larger, chances are they will not quite work well with the beam splitter.

However, if you find another camera casing, like say from an old ENG system that uses similarly sized(16:9?) chips, it should be incredibly easy to mount the rockwell chips and even put hard disks and the rest of the electronics in the camera, since ENG shoulder bricks tend to be pretty large.

Juan

This is a very interesting topic! Here is a system that is HD. Take a look!

http://www.siliconimaging.com/SI1280%20spec.htm

http://www.siliconimaging.com/


Michael Pappas
PappasArts Entertainment
http://www.Pbase.com/arrfilms
http://www.pbase.com/pappasarts15

so then to "build your own" hd camera using those chips (assuming they were financially viable) and the dual FW800 setup (or similar).. one would still need a casing, beam splitter, a way to mount a lens to the casing the appropriate distance from the film plane, power supply for the chips or any other onboard electronics.. and what else?

That's pretty much it...it depends on how you intend to capture the data. If you want on-board hard disks, then you can use some sort of programmable logic chip to interface the data from the chips to, say, a serial ata controller. likewise, for a firewire 800 interface, you can have the programmable chip interface the data to firewire link layer->physical layer. The programmable chip can also selectively interface to both, making the system more flexible.

There's not that much to it. Of course, there is some PCB work to be done, in order to design and solder the board that has the programmable chip and interface chips, but the circuitry is extremely simple and there is little if any 'glue logic', as long as you use the rockwell chips.

if you use standard CCD chips, for example, then you need independent sample-and-hold chips for each CCD, together with timing chips and A/D converters...this complicates things a bit.

Juan

When are you gonna post the uncompressed pics?

Right now I am literally working on the bad connection issue that is causing noise in the capture...once i get a good connection i will post the 10-bit uncompressed color captures.

juan, what is the thing called that splits the light into RGB colors? it's not a beam splitter is it? or do you need a beam splitter AND color filters to do it? this is with 3ccd or cmos chips that are Black and White

I have read about the chroma phasing problem with the dvx100 during pans which can make keying difficult with the camera. Do you notice this problem still using your method of pulling the image from the chips or does that seem to clean it up?

Obin:

Afaik, there is a three-way beam splitter which basically splits the light coming from the lens into three(i.e. the exact same image x3 with 1/3 the original intensity) and in front of each monochrome CCD, then there is a color filter for R,G or B.

Thomas:
I am not aware of the details, but it is very possible that this solves the problem. I think the chroma phasing could be a side-effect of how the electronics deals with the signals, perhaps when they are transformed to Y-C.

Unless this chroma phasing is a physical defect in the design, or in the timing characteristics of the chips(doubt it) it probably won't be present in the raw output.

Juan

ok in that case edmund optics has what you need a RGB filter set ...cost, about thirty bucks and I think they have a beam splitter too......http://www.edmundoptics.com/

Thanks! That's fantastic, they do have the beam splitters....now the only hurdle is obtaining the rockwell chips at a decent price, no reply from them yet.

An update on the DVX:
Last night i fixed the false connection problem, i was still getting bad connections but i found out that the culprit was a surface mount soldering i did that was getting loose. Tonight i will resolder it.

I managed to capture one out of focus full frame before the connector completely fell off, but the software was set to single channel so it captured a red 10-bit frame...looks noise free...

I got a response from Rockwell...it seems what was said on this thread before is true! For a 1-10 unit order, the ProCamHD 2560 is a bit over $500, while the ProCamHD 3560 is about $1500.

The ProCamHD 2560 would be the PERFECT choice for making a 3CCD HD box camera...it is 1280x1024, and can do up to 60fps progressive so slow motion is definitely in :)

Juan

dude that is soo great! I sure hope that this can be done and we can bypass the $60,000 + lenses why not bypass tape too? and use some sorta firewire800 capture ??


don't get ahead of yourself yet though! we all still want dvx100 4:4:4 !!!!!!


what about computer processing ?? don't you need lots of stuff to get the data from 3 cmos chips into a color image format? that can be saved?

I could not find a 3 way beam splitter...only 2 way...I sent an email to find out if they have what you need ;)

<<<-- Originally posted by Juan P. Pertierra : I got a response from Rockwell...it seems what was said on this thread before is true! For a 1-10 unit order, the ProCamHD 2560 is a bit over $500, while the ProCamHD 3560 is about $1500.

The ProCamHD 2560 would be the PERFECT choice for making a 3CCD HD box camera...it is 1280x1024, and can do up to 60fps progressive so slow motion is definitely in :)

Juan -->>>

so you would need 3 CCDS at $500 x3, then a beam splitter +RGB $100-200?, then a suitable box $120? then a lens...$200? then engineer the interface...$$$, I gather this is going to be in the area of $2000 to $3000usd to prototype. Not bad for an HD cam I guess...we are appoaching small car territory.

Oh yea, what size are the CCD's 35mm?

The future is in one chip designs. All current DSLRs use one chip. Why not implement a 1 chip HD design, if anything? If I knew anything about electronics on this level, I'd probably attempt to roll my own HD camera using a single Rockwell chip, some sort of hard drive storage and interchangeable lenses. Just my .02 cents.



mg

B&W HD, lol

Who said anything about black and white HD? Just ask Arri or Kinetta about 1 chip HD designs.



mg

First of all, right now the only one-chip design that would perform equally as well as a 3-chipper is something that is not exactly standard like the chip in the Sigma SD10(photography). It is a single chip but it effectively has three layers of sensors so it's basically 3 chips in one. It is waaaay to slow for HD video use anyway.

I think the idea that single chippers will perform equally as well as 3-chip cameras is an expression directed towards the consumer market...consumers are getting compressed 4:1:1 anyway, so when going to HD, even 4:1:1 yields fantastic color(comparable to a SD 3-chip) because there are so many more pixels sampled.

However, in the professional world, where you might want to project your footage on a huge theater screen, every bit of color counts...and right now, in the video world, unless there is something like the Sigma chip, i don't think anything but a 3-chipper is an option for professional use. If this was not the case, why aren't the Thomson Viper or Cinealta, or Varicam 1 chippers?

If I am wrong, someone please correct me.

Juan