Any new infos about the PD1/HD1?

Marco

yeah, Leo, it's been ages - have you managed to unlock anything else since unlocking the menu entries for colour bars and audio rec level indicator?

We're very excited.

I'm very curious if there is a way to get 1280x720 out of the PD1 via firewire just like the HD1 does.

Marco

Hi there! I am back...

It's been awhile but I am looking at GR-PD1/HD1 again. I believe there could be something done to make them better!

I have stopped the research after I was not able to get access to HD1/HD10 or their memory dump but now I had to put some logical end to it and have ordered a new HD10.

For the last year I was shooting with EE completely turned off and I like the effect so much.

My biggest problem is that the camera that I have (PD1) was bought second-hand and I believe was not colour-adjusted properly - colour saturation changes with the luma. I.e. when exposure level drops, colours seems to become more vibrant. Exactly like if you had linearity problems in luma vs chroma channel on an analogue equipment.
This, obviously, makes chroma noise more pronounced in the dark areas...

Hi Leo

Welcome back.

Could be the extra poor low light sensitivity of the JVC sensor. With the GR-DV3000, lower light produced unbalanced colour (unless you use the presets) but mostly grey, that seems like the test frames of the HD10 too. On cheaper cameras there is a trick to increase gain and saturation in low light. The way the JVC sensor works makes red noise worse as well.

I have been putting out feelers for a low priced secondhand JVC PD1/HD1/10, but no response.


Anyway, have you seen my thread here on JVC mod ideas:

http://www.dvinfo.net/conf/showthread.php?t=64636


Another camera that could produce better results then the JVC with modification, is the Sanyo HD1 for $799, and it is probably not even the best one coming.

Wayne,

Email me, hmcknight@mac.com, about an HD10. I may be able to sell it (it's a lease that I'm paying off this month), depending on several factors with the leasing company.

heath

There's a HD1 on eBay for $412, bidding. Seller still has box, juding by the photo they posted, not that this means anything of course..

Thanks guys, I had pretty much given up on that idea (my post in the classifieds forums didn't attract anything) and the H264 cameras are nearly here, the Samsung in August, the AVCHD, I don't know. So I am only looking as a cheap fill in camera. I have even been thinking of that $799 Sanyo HD1 (posting about a possible solution to the macro orientation bug over there).

So I will try the email now, Keith.

Graham, thanks, $412 would be great, but I am in need of sleep, so I'll have to check it latter. I did look at ebay previously, but the cameras went for a lot more than that by the time the auction finished.

Hey Leo, good to hear your back in action. In your soon to be extended tinkering if you ever come accross a variable for any of the presets, that would be great. For example the "Sports" preset sets the shutter to 1/250th and up. If this default could be changed to the far more usefull 1/60th and up, it would become thew single most usefull feature on the cam. Couple this with EE reduction and these used HD1's would become an amazing bargin.
Keep in touch and good luck.

Don't forget low light gain, and manual functions (especially locking the shutter down to 30/60fps).

Wayne, Ken, I will definitely try to find anything of interest. I have here default and AGC low light limits but, obviously, this needs controlled experiments. I also have few sound parameters but I haven't looked at them closely enough. Looks like they are low and high-pass filters, gain and audio AGC timing.

I agree that tweaking one of the [unused] presets is an easy way to get some manual control over exposure. I will look at it more closely.

Thanks for everybody.

As I understand, there is virtually no gain in HD mode (that famous 36 horror lux rating, as deadly as no 25fps support). This is due to noise problems with the design, which I always thought was not as good as what was needed. But, for run and gun, footage with noise is better then no footage at all.

The sensor has likely been greatly improved since it was released, so this might prove to be better then expected on some cameras.

Negative gain to extend exposure latitude is also another trick.

As long as you can controll light with variable polarizer's (ND's) or ND filters themselves, the cam always reverts to 1/30th in HD mode. At times you can finesse it to 1/60th and maintain an exposure lock, but it is tricky. If I knew I could just flick on the sports "preset" at get a 1/60th lock and then was free to lock exposure, it would save so much time and effort. One would still need to apply filters if needed to prevent it from climbing higher then 1/60th (or the old 1/250th) if there is too much light comming in. But a least you know it would never drop down to 1/30th.

I personally do not think a gain boost is of much use. In low light the cam is far too noisy for it to be much use. Lowering the shutter and leaving it in auto exposure as well as shooting in the SD HDV mode are far better options. (The SD mode looks almost every bit as good as the HD mode except for medium to long shots where you want background detail. It intercuts seemlessly with HD footage)

Yes, it is totally possible to shoot 25p or even 24p, on this camera.

I was thinking of an old idea I had for fitting 25p in a 30p stream, for a universal camera footage. I contemplated on how using this technique you could do 25p on this camera, but using a external shutter working at 50th of a second, moving the timing around (smooth, but slight variations) so that 25 frames would fall in 25 of the thirty frames (I know you guys are interested in 24fps, but it applies the same). As there is nothing to record in the other five frames, they get marked as blanks and hopefully that extra bandwidth is freed up for the other 25 frames.

This works because a 1/50th a second shutter can be moved around inside a 1/30th a second frame. We move the frame so it gradually goes up against the hard edge of the beginning of the frame to be skipped, and frame after at the hard edge of the previous frame. This reduces the jump across the skipped frame. The camera shutter has to be 30th a second so the sensor is nearly always on, which might increase dark current noise. It then occurred to me, hey if it is always on then we can simply dump the 1/50th shutter image the sensor at anytime, treating the camera frames as a continuous storage device. Some 1/25th of a second images will be split across two 1/30th a second frames, but this can be joined together by software and the blanks removed. There might be a problem of a lag in between the 1/30th second frames (not true 1/30th a second) or variation in brightness because of readout, but adjusting the external shutter slightly to avoid this should eliminate it. the is adjustment, if at all required, would probably be much less then then the adjustment needed in the first and you might be hard pressed to notice it even if you were looking for it. With the first scheme, you would probably notice it, if you were looking for it, but even fi you did it would be so subtle as not to concern you anyway.

The external shutter always does 1/50th/s stopping variation in light response in the sensor as the frame period changes. You might be able to do this electrically by hacking the camera, but shutters for movie cameras start at over $100. If you can replace the lens with an SLR lens through a condenser, which would give you a number more stops lowlight, you could add the shutter there and a variable ND to reduce blowout from the extra brightness of the SLR lens.

Now, about that gain, yes please, with an SLR lens, like I described above, you could get true low light ability. About the noise, is it much better on a cool day compared to a hot day, then it is probably thermal noise and dark current. Removing the lens and applying a cooling method would reduce the low light noise problems.

So would anybody like to try these with their own camera (you don't have to replace the lens to do it)?

Re-edit: Whoops already covered.

I remember why I was posting here, I found what looks like a service manual for the HD1 going on ebay, if anybody would like one.

http://cgi.ebay.com.au/JVC-GR-HD1-GR-HD1US-DIGITAL-SERVICE-MANUAL-REF-420_W0QQitemZ7510895289QQcategoryZ21167QQrdZ1QQcmdZViewItem

Out of curiosity, I know the GYDV500 is supposed to have a hidden uncompressed video port, are there any in these cameras, any hidden ports/sockets/pins at all on the boards? You guys have the schematics is there anything there?

I wonder if formatted uncompressed information can be tapped along the process path without having to go as far as Andromeda?

There's a HD1 on eBay for $412, bidding. Seller still has box, juding by the photo they posted, not that this means anything of course..

I looked but couldn't find anything, do you have a link Graham?

The camera produces a lot of heat inside which is designed to be dissipated by the aluminium frame (there is a heatpipe going to it from the DSP/encoder chips) - this surely drives up the dark current and noise of the CCD imager. I have some ideas around the CCD frontend and did some work in bypassing primitive one-BJT current sink JVC put for some reason at the output of the CCD before emmiter follower. No controlled tests yet but it surely did not make the situation worse.

The simplest way to shoot 24p on 30p would be to swap all crystal resonators in the camera for 20% slower ones and then play back the tape in the similar camera for normal acquisition of 30p source which is now in fact 24p timebase. All operations are controlled either directly from XTAL frequencies or via PLL controlled circuits. Shutter speed, frame rate, tape speed, drum RPMs, etc.

Re-edit: Whoops already covered.

I remember why I was posting here, I found what looks like a service manual for the HD1 going on ebay, if anybody would like one.

http://cgi.ebay.com.au/JVC-GR-HD1-GR-HD1US-DIGITAL-SERVICE-MANUAL-REF-420_W0QQitemZ7510895289QQcategoryZ21167QQrdZ1QQcmdZViewItem

I've got it. It's not that useful.

You wouldn't believe I posted a more primitive form of this idea before and forgot all about it, but I now remember I posted a link to it, I thought here, so you guys could look at it, but I don't see any.

Here is the thread, and it adds a little more to the Scheme:
http://www.dvinfo.net/conf/showthread.php?p=460793#post460793



Leo,

This sounds good (if programmable even better), the trick is the compression engine works at 30fps. If the engine timing can't be changed, it might be possible to do the sensor frames at 24fps out of sync with the 30fps compression engine, similar to the shutter idea except sensor replaces external shutter, and frame buffer replaces 30 fps sensor storage. This could be done by giving it a blank frame every sixth frame from the frame buffer (either by changing buffer address, or filling with black, or inserting a fake black signal). A little more complex but more compact than external shutter

Patrick,

Is there anything to trace on there that might represent an output, or a place to tap, uncompressed video.

In furthering my investigation into this hacking project, a very knowledgeable JVC dealer told me that there have been modifications to the HD10U since its release. From trying to read between the lines of what the dealer was 'saying' without saying it (NDA?? He would neither confirm nor deny this), several hardware changes have been made without the model number changing. This would lead to different firmware floating around.

What sort of modifications was he talking about Xander?

As you can tell from all the posts, sorry for that, I have been re-reading right through the thread. One thing I came across, was the mention of a "change" etc manual for changing the firmware, did anybody find anything? That would probably describe all the useful locations.

Thanks for all your patience.

MPEG encoding engine is synchronous with frame rate. I am sure that inserting random black frames in MPEG stream is going to wreck havoc and significantly drop overall image quality. MPEG-2 compression relies heavily on interframe correlation, compressing only differences between new and predicted frame contents. But DV compression might be OK as each frame is compressed independently.

There is no easy place to tap into the digital stream as PCB has a range non-standard busses ranging from direct 10-bit data off the imager to data flow to and from MPEG-2 codec chip. All of it is synchronous data which also needs clock.

It looks like component Y'PbPr outputs have uncompressed signal on them and that's the best you can get - the component encoder is JVC's own chip and even then, it is in BGA package.

The component out is uncompressed, but it is at SD 60p. I have never been able to verify if it is the raw 1040x480 or if it is the condended 720x480 wide screen.

I'll bet the JVC HD cam has one lurking somewhere too. People should try powering up with all sorts of key combo's held down, it isn't that many possibilities and the rewards might be adjusting the ccd gain, and turning off/down the edge enhance on the HD1.
Who knows what else!

Who will be the first to find it? ;)

-Les
What does "Self destruct sequence activated" mean???

The component out is uncompressed, but it is at SD 60p. I have never been able to verify if it is the raw 1040x480 or if it is the condended 720x480 wide screen.
The dot clock for the D/A chip that produces component output is 27.75MHz. Given NTSC 60i video line timing is 51.2uS this results in 1420 dots per 60i line. I don't know timing for 60p signal but I assume it is roughly twice faster so it looks like 720 fits better.

Many people have reported that using their PD1's that their HiRes SD mode, captures at 1040x576, which is what the square pixel format is in wide screen, befor it is compressed anamorphically with wide pixels at 720x576. I would assume when you capture uncompressed (component out) you would get the raw widescreeen 16:9 SD which my guess would be 940(ish?)x480.

MPEG encoding engine is synchronous with frame rate. I am sure that inserting random black frames in MPEG stream is going to wreck havoc and significantly drop overall image quality. MPEG-2 compression relies heavily on interframe correlation, compressing only differences between new and predicted frame contents. But DV compression might be OK as each frame is compressed independently.

The image should never be worse, and hopefully better as it moves the data stream around to take advantage of the space freed up from a blank frame. But I see what you mean, it might make the mistake of saying that the blank screen has to be recorded as a difference from the last frame, rather then a big black rectangle, and the next frame the difference from the rectangle, inefficient. Some testing would be needed to see how smart it is in this. Even MiniDV can do this to a certain extent, even though it is inter frame, adjacent frames can carry extra data if some are left over (I read the specs).


The component out is uncompressed, but it is at SD 60p. I have never been able to verify if it is the raw 1040x480 or if it is the contended 720x480 wide screen.

Have you tried a res chart with a resolution synchronising monitor? Even a oscilloscope with a horizontal line pattern would tell (print lines, size to match expected camera resolution, see if it goes off on in sync) providing you have one of high enough sampling frequency.

I have never tried component uncompressed out from these cams.

Hi Leo,

I have totally disassembled my PD1... There are 2 PCB Boards. On the top, one named "Digital" with the Firewire connector, and on the bottom, one named "Analog". They are striped. Do you think we could get a Hi Def signal from this internal connection ?

My goal is to capture the raw data from the sensor and then apply a better demosaicing process. (I try to do the same thing with the Sony HC1)

How do yo connect your camera on a PC ? (cable pins and Software...)

Thanks.

Good find Rodolphe. Brave, the problem I find in taking cameras apart is getting them back together in working order (my old sharp currently is not working and had some case circuit scheme that also seems to be at fault).

There is a guy that has experience with the HC1 firmware around, Wolfgang Winne, he is in Germany. It would be very good to ask him about this stuff in relation to the HC1, and you might ask him about the PD1.

http://www.dvinfo.net/conf/showthread.php?p=425035#post425035
http://www.schnellsuche.de/kostenlos/DV-IN/BauAnleitung.htm
http://www.fxsupport.de/hc1/index.html
http://babelfish.altavista.com/babelfish/trurl_pagecontent?lp=de_en&url=http%3A%2F%2Fwww.fxsupport.de%2F01.html

His static 35mm adaptor looks interesting.

http://www.fxsupport.de/hc1/hc1_9.html

Wolfgang is a very thorough person, like me, less the sickness, which you can tell by his site, he even looks a bit like me, poor lad ;) .

Actually, Juan from Andromeda, might also be interested. Not that it is worth doing for him, but maybe he might sell configurable recording units/software and let you figure it out.

Actually, there are even more with the 5100....such as adjusting color rotation in EXTRA 3. AND.....hidden menu in the 3000 deck that allows LP as well as SP mode..........................................
Lou Bruno



Interesting!

Check this out regarding hidden menus on JVC cameras:

http://www.abcdv.com/article/articleview/78/1/74/

http://www.abcdv.com/article/articleview/169/1/74/

http://www.offworld.com/dv500/troubleshooting.html#firmware

No easter eggs for the HD1/10. They should have to put at least one in by law!

Does any of this HD100 stuff help:

Menu and time-laps:

http://www.dvinfo.net/conf/showthread.php?t=68885

http://www.dvinfo.net/conf/showthread.php?t=70103

Even MiniDV can do this to a certain extent, even though it is inter frame, adjacent frames can carry extra data if some are left over (I read the specs).

You probably misunderstood it. It can reallocate bits, but it won't use previous frames data to create a new one.

Inserting black frames into mpeg2 material will destroy the image quality. Like white flashes will create very nasty blocking.

You probably misunderstood it. It can reallocate bits, but it won't use previous frames data to create a new one.

Inserting black frames into mpeg2 material will destroy the image quality. Like white flashes will create very nasty blocking.

From memory that is what I read from the specs, that there is a limited portion of the frame data that can be given to the next frame (but maybe that was field).

With the black frame, as I said before, you might be right.