4:4:4 12-bit Uncompressed DVX100

Obin Olson · February 16th, 2004, 09:18 AM

john your going to have to write code to do this? i think firewire 800 for a HD sized image would work ok?

Juan P. Pertierra · February 16th, 2004, 12:50 PM

I beleive it is 966x492, because i found the catalog for Panasonic CCD's and it's the only progressive scan CCD that could possibly be the ones in the DVX. Even though I can see the CCD/prism assembly, i can't read the model number off the CCD's because of the way CCD's are installed. Matsushita doesn't have any PS CCD's that small.

It's clearly wider, but there are usually several dummy pixels in each line, and on top of that there might be several other things. For example, the prism might be designed in such a way that the CCD needs to be wider than normal, or perhaps to compensate for the NTSC pixel aspect ratio if the CCD's have square pixels.

As far as editing it, i know for a fact that final cut pro does 10-bit 4:2:2 uncompressed...however to handle the 4:4:4 uncompressed 12-bit, i'm not sure...i think FCP has an option to import individual frames and compile them into a movie, i know Apple Shake does this. You can import in say, tiff or BMP format.

The sound can be taken directly from the DV tape. Since both recordings will be at the same frame rate, all you have to do is capture the audio from the DV tape(which records at the same time) and align it to the video. It's relatively easy to add sound once i have all of this working, but that's the way it can be done initially.

Juan

Juan P. Pertierra · February 16th, 2004, 12:56 PM

I plan to use a TI Firewire 800 chip along with all the required FIFO's, etc to interface the RGB signals to the transceiver. 800MBps is more than enough, even for the raw signal from the A/D's.

Clayton Farr · February 16th, 2004, 01:02 PM

Thanks for all of the time put into this Juan. I look forward to your results.

Best,
Clayton

Juan P. Pertierra · February 16th, 2004, 01:07 PM

Oh and it should work in all modes, the interface should handle the higher bit rate of 30p/60i, although the driver will have to be programmed to handle these two additional cases....

Obin Olson · February 17th, 2004, 06:41 PM

I am looking forward to this "mod" so so much!

Lucia de Nieva · February 18th, 2004, 02:17 AM

Hello boys,

I really dislike repeating myself, but

as there is already a working DVC50 (6,5 MB/S, 4:2:2) firewire protocol directly supported by FCP4,

and several harddisk recorder units, like for instance the Quickstream DV, are available on the market for some time now,

plus, thanks to your ingenuity, Juan, we know the Y/C stage signal is still uncompressed,

it should be worth a thought whether it might be possible to alter the way the information is processed before it is send over firewire, hence obtaining DVC50 quality with the DVX for a start.

Therefore, Juan, again it would be very helpful, if you could post the names of the various chips within the DVX´s circuit and perhaps a schematic, so that these can be compared with 900´s system.

Best thanks again for all your devotion to this bold project. My suggestions are not meant to question your primary approach, but to open up some sideways for those prefering a lighter and less demanding modification, perhaps for the price of a comparatively lower efficiency.

Juan P. Pertierra · February 18th, 2004, 04:32 AM

Lucia,

The data coming out of my mod can be toned down and piped through watever standard you wish, however there are only a couple of places where you can get the raw data out, and this mod covers both. Anything dealing with using DVC50 over firewire, or any other standard is simply formatting of the data that can be done in the external circuit once I have the data coming out.

The main chip on the DVX is an Altera APEX 20K which is an LPGA, and it's function depends competely on what panasonic programmed into it which we do not know, and for this application I am not interested in as long as I know the data pins.

In other words, this is not some standard chip that you can look up the specs of and know exactly what it's doing...there ARE specs for this chip but everything it does, is based on the software, which of course is heavily protected. But we don't need this to do the procedure.

Juan

John Gaspain · February 18th, 2004, 08:44 AM

http://www.altera.com/products/devic...apx-index.html

Juan P. Pertierra · February 18th, 2004, 10:27 AM

I contacted the company that manufactures the chip-clip, and it seems that they have one that is perfect for this application.

Their design will allow for my idea of having the camera completely assembled and with only a small ribbon cable coming out the back or bottom. I think it even can be routed out without any modification to the camera.

it will take them around 10 days to assemble one, so it's a waiting game for now.

Juan

Lucia de Nieva · February 18th, 2004, 04:57 PM

Juan, thanks to John´s link I am a bit wiser now. Apparently I underestimated the complexity of the issue, so the more your efforts are to be welcome. Did I understood you right that your construction does imply some kind of ASIC to format the data? For any compression technique will need the proper codecs to be recognized from the editing suite and the raw or Y/C output probably does not match any standards according to your results. Or is your favorized method of single frame sequencing only a matter of proper timing alone, for example sending one output sample every 40 or 41,7 ms?

John Alton Disciple · February 21st, 2004, 07:13 PM

Hello all. Just catching up on the posts as I haven't been around in a while.

Juan - Fantastic! I experimented on such a concept in 1996 with another 3 chip from Panasonic (the AG-3). The result on the computer was much better than the camera's taped recording ability. At that time I was recording 4:2:2 (but compressed at 5:1) - which for the day was respectable.

Please keep up the excellent work!

Jason Rodriguez · February 21st, 2004, 11:22 PM

BTW, what are the images straight off the A/D converters going to look like? Will there be problems with noise, banding, etc.? You're probably already going to do this, but could you post some pictures/movies when you finally get this thing working? Also will this mod work with the DVX100A?

Obin Olson · February 22nd, 2004, 09:41 AM

John...do you still have the 4:2:2 version working?

Juan P. Pertierra · February 22nd, 2004, 12:05 PM

Jason,

At this point I am not sure, but from what I've seen in outputs straight from CCD's(such as the Thomson Viper) the raw output needs color adjustments and frame resizing...however if I do use the raw output I plan to have all of these corrections done in software either at capture time or right after.

An interesting note...if we use the raw A/D output, it doesn't matter which camera you have, the DVX100 or DVX100A, because afaik they both have the same CCD's and 12-bit A/D converters. If we endup taking the signal after the Altera chip which does all the color corrections, then the DVX100 will have 10bit color, but the DVX100A will yield 12-bit color.

Juan

February 18th, 2004, 02:17 AM	#82
Lucia de Nieva Regular Crew Join Date: Jan 2004 Location: Germany, Northern Europe Posts: 32	Microcontrollers Hello boys, I really dislike repeating myself, but as there is already a working DVC50 (6,5 MB/S, 4:2:2) firewire protocol directly supported by FCP4, and several harddisk recorder units, like for instance the Quickstream DV, are available on the market for some time now, plus, thanks to your ingenuity, Juan, we know the Y/C stage signal is still uncompressed, it should be worth a thought whether it might be possible to alter the way the information is processed before it is send over firewire, hence obtaining DVC50 quality with the DVX for a start. Therefore, Juan, again it would be very helpful, if you could post the names of the various chips within the DVX´s circuit and perhaps a schematic, so that these can be compared with 900´s system. Best thanks again for all your devotion to this bold project. My suggestions are not meant to question your primary approach, but to open up some sideways for those prefering a lighter and less demanding modification, perhaps for the price of a comparatively lower efficiency.

February 18th, 2004, 08:44 AM	#84
John Gaspain Major Player Join Date: Oct 2003 Location: WA-USA Posts: 371	http://www.altera.com/products/devic...apx-index.html __________________ The glory of the World passes by.

February 18th, 2004, 04:57 PM	#86
Lucia de Nieva Regular Crew Join Date: Jan 2004 Location: Germany, Northern Europe Posts: 32	would have been too easy .. Juan, thanks to John´s link I am a bit wiser now. Apparently I underestimated the complexity of the issue, so the more your efforts are to be welcome. Did I understood you right that your construction does imply some kind of ASIC to format the data? For any compression technique will need the proper codecs to be recognized from the editing suite and the raw or Y/C output probably does not match any standards according to your results. Or is your favorized method of single frame sequencing only a matter of proper timing alone, for example sending one output sample every 40 or 41,7 ms?

February 16th, 2004, 09:18 AM	#76
Obin Olson Trustee Join Date: Jan 2003 Location: Wilmington NC Posts: 1,414	john your going to have to write code to do this? i think firewire 800 for a HD sized image would work ok?

February 16th, 2004, 12:50 PM	#77
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	I beleive it is 966x492, because i found the catalog for Panasonic CCD's and it's the only progressive scan CCD that could possibly be the ones in the DVX. Even though I can see the CCD/prism assembly, i can't read the model number off the CCD's because of the way CCD's are installed. Matsushita doesn't have any PS CCD's that small. It's clearly wider, but there are usually several dummy pixels in each line, and on top of that there might be several other things. For example, the prism might be designed in such a way that the CCD needs to be wider than normal, or perhaps to compensate for the NTSC pixel aspect ratio if the CCD's have square pixels. As far as editing it, i know for a fact that final cut pro does 10-bit 4:2:2 uncompressed...however to handle the 4:4:4 uncompressed 12-bit, i'm not sure...i think FCP has an option to import individual frames and compile them into a movie, i know Apple Shake does this. You can import in say, tiff or BMP format. The sound can be taken directly from the DV tape. Since both recordings will be at the same frame rate, all you have to do is capture the audio from the DV tape(which records at the same time) and align it to the video. It's relatively easy to add sound once i have all of this working, but that's the way it can be done initially. Juan

February 16th, 2004, 12:56 PM	#78
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	I plan to use a TI Firewire 800 chip along with all the required FIFO's, etc to interface the RGB signals to the transceiver. 800MBps is more than enough, even for the raw signal from the A/D's.

February 16th, 2004, 01:02 PM	#79
Clayton Farr Contributor Join Date: Aug 2002 Location: Salt Lake City, UT Posts: 94	Thanks for all of the time put into this Juan. I look forward to your results. Best, Clayton

February 16th, 2004, 01:07 PM	#80
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	Oh and it should work in all modes, the interface should handle the higher bit rate of 30p/60i, although the driver will have to be programmed to handle these two additional cases....

February 17th, 2004, 06:41 PM	#81
Obin Olson Trustee Join Date: Jan 2003 Location: Wilmington NC Posts: 1,414	I am looking forward to this "mod" so so much!

February 18th, 2004, 04:32 AM	#83
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	Lucia, The data coming out of my mod can be toned down and piped through watever standard you wish, however there are only a couple of places where you can get the raw data out, and this mod covers both. Anything dealing with using DVC50 over firewire, or any other standard is simply formatting of the data that can be done in the external circuit once I have the data coming out. The main chip on the DVX is an Altera APEX 20K which is an LPGA, and it's function depends competely on what panasonic programmed into it which we do not know, and for this application I am not interested in as long as I know the data pins. In other words, this is not some standard chip that you can look up the specs of and know exactly what it's doing...there ARE specs for this chip but everything it does, is based on the software, which of course is heavily protected. But we don't need this to do the procedure. Juan

February 18th, 2004, 10:27 AM	#85
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	I contacted the company that manufactures the chip-clip, and it seems that they have one that is perfect for this application. Their design will allow for my idea of having the camera completely assembled and with only a small ribbon cable coming out the back or bottom. I think it even can be routed out without any modification to the camera. it will take them around 10 days to assemble one, so it's a waiting game for now. Juan

February 21st, 2004, 07:13 PM	#87
John Alton Disciple New Boot Join Date: Jan 2003 Location: El Paso TX Posts: 14	Hello all. Just catching up on the posts as I haven't been around in a while. Juan - Fantastic! I experimented on such a concept in 1996 with another 3 chip from Panasonic (the AG-3). The result on the computer was much better than the camera's taped recording ability. At that time I was recording 4:2:2 (but compressed at 5:1) - which for the day was respectable. Please keep up the excellent work!

February 21st, 2004, 11:22 PM	#88
Jason Rodriguez Trustee Join Date: Mar 2003 Location: Virginia Beach, VA Posts: 1,095	BTW, what are the images straight off the A/D converters going to look like? Will there be problems with noise, banding, etc.? You're probably already going to do this, but could you post some pictures/movies when you finally get this thing working? Also will this mod work with the DVX100A?

February 22nd, 2004, 09:41 AM	#89
Obin Olson Trustee Join Date: Jan 2003 Location: Wilmington NC Posts: 1,414	John...do you still have the 4:2:2 version working?

February 22nd, 2004, 12:05 PM	#90
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	Jason, At this point I am not sure, but from what I've seen in outputs straight from CCD's(such as the Thomson Viper) the raw output needs color adjustments and frame resizing...however if I do use the raw output I plan to have all of these corrections done in software either at capture time or right after. An interesting note...if we use the raw A/D output, it doesn't matter which camera you have, the DVX100 or DVX100A, because afaik they both have the same CCD's and 12-bit A/D converters. If we endup taking the signal after the Altera chip which does all the color corrections, then the DVX100 will have 10bit color, but the DVX100A will yield 12-bit color. Juan