4:4:4 12-bit Uncompressed DVX100

John Gaspain · March 26th, 2004, 12:04 AM

<<<-- 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?

Matthew Groff · March 26th, 2004, 08:09 AM

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

John Gaspain · March 26th, 2004, 08:43 AM

B&W HD, lol

Matthew Groff · March 26th, 2004, 09:45 AM

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

mg

Juan P. Pertierra · March 26th, 2004, 10:12 AM

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

Milosz Krzyzaniak · March 26th, 2004, 10:43 AM

A very interestning thread indeed.

I have a suggestion over your project. I think it would be interestning to consider deliver a way to someway compress of the data to reasonably lower rate with yet maintaing good quality, 4:4:4 and 12-bit color depth benefits. I think cutting the stream by 2 or even 3 times would do no harm to its quality. If recorded on a laptop, wouldn't be a good idea to write a codec that would employ the MPEG-2 I-Frame compression, or at least MJPEG compression with a reasonable ratio of eg. 2:1 or 3:1.

regards

Juan P. Pertierra · March 26th, 2004, 10:52 AM

I am keeping all of this in mind. The important thing right now is to get raw frames and see if the output is good enough that people are interested in me doing a portable mod. At the very least, the final design allows you to select different decimations (4:1:1, 4:2:2, etc). Compression is an option as well, thoiugh i haven't implemented that yet in my design.

Update:
I am letting a pro do the soldering this time, i dropped it off this morning with a friend of mine who does soldering for the EE department, it should be ready this afternoon. After that it should be a matter of hooking it up and capturing frames tonight.

Juan

Adam Burtle · March 26th, 2004, 11:19 AM

Glad to hear it, Juan. If you need somewhere to host video or pics, just let me know and i can take care of it for you.

Obin Olson · March 26th, 2004, 04:10 PM

dude check this out:

http://www.photonics.com/

they should have a beam splitter that goes 3 ways for 3 cmos chips..I don't have time to look at the site but someone told me they should have what you want

Jason Rodriguez · March 26th, 2004, 05:13 PM

<<<-- Originally posted by Juan P. Pertierra : If I am wrong, someone please correct me.
-->>>

All the next generation Digital Cinema cameras which are inherently designed to be blown up to theatre-size screens are going to be one-chippers: Arri D20, Dalsa Origin, Kinetta HD and the Aaton D-Minima.

Juan P. Pertierra · March 26th, 2004, 05:19 PM

Oh i don't doubt that...however matthew was sugggesting using a 1-chip design for the HD box camera we are talking about here, which, unless the new generation of chips in these cutting edge cameras are cheap, is really not an option. I was talking about currently available chips that we can purchase as consumers for a relatively low price.

The whole point of the rockwell HD chips is that they can be had for relatively low price, and they already include all the necessary circuitry which further lowers the price of implementation as well as design complexity.

Obin Olson · March 26th, 2004, 05:34 PM

exactly.

Obin Olson · March 26th, 2004, 06:49 PM

when will the frame grabs be ready?

Juan P. Pertierra · March 26th, 2004, 07:26 PM

I am hooking it all up right now. It is 30 wires, but it takes time because if i'm not careful the surface mount connector can come off again and delay everything.

However, I decided to upload a very partial image. This image contains ONLY 27% of the imformation of the complete raw frame. ALSO: it still quite a lot of noise that can be seen as dark dots in sharp edges about the image...it's not perfectly focused either because the LCD screen is disconnected, and I can't put my eye against the viewfinder.

Once again, this is only the 10 red bits, coming off one CCD only. I will try and upload more partial frames as i carefully hook up more wires.

The noise is due to the fact that I am forced to use long thin wires with my test setup, because of the location of my computer. I will try and get completely rid of it once i have all colors hooked up.

http://expert.cc.purdue.edu/~pertierr/dvxcap.tiff

Juan

Juan P. Pertierra · March 26th, 2004, 09:00 PM

Correction...i misplaced one of the wires and that image is actually 9-bit ...bit 7 was mistakedly hooked up to ground so that image contains actually even less information...also this bad hookup could be the source for the speckled noise....

Eating dinner...then going back to work. :)

Juan

March 26th, 2004, 12:04 AM	#196
John Gaspain Major Player Join Date: Oct 2003 Location: WA-USA Posts: 371	Re: Rockwell chips <<<-- 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 glory of the World passes by.

March 26th, 2004, 08:43 AM	#198
John Gaspain Major Player Join Date: Oct 2003 Location: WA-USA Posts: 371	B&W HD, lol __________________ The glory of the World passes by.

March 26th, 2004, 10:43 AM	#201
Milosz Krzyzaniak Regular Crew Join Date: Mar 2004 Location: Germany Posts: 64	Suggestion A very interestning thread indeed. I have a suggestion over your project. I think it would be interestning to consider deliver a way to someway compress of the data to reasonably lower rate with yet maintaing good quality, 4:4:4 and 12-bit color depth benefits. I think cutting the stream by 2 or even 3 times would do no harm to its quality. If recorded on a laptop, wouldn't be a good idea to write a codec that would employ the MPEG-2 I-Frame compression, or at least MJPEG compression with a reasonable ratio of eg. 2:1 or 3:1. regards

March 26th, 2004, 08:09 AM	#197
Matthew Groff Regular Crew Join Date: Mar 2003 Location: New York, NY Posts: 132	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

March 26th, 2004, 09:45 AM	#199
Matthew Groff Regular Crew Join Date: Mar 2003 Location: New York, NY Posts: 132	Who said anything about black and white HD? Just ask Arri or Kinetta about 1 chip HD designs. mg

March 26th, 2004, 10:12 AM	#200
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	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

March 26th, 2004, 10:52 AM	#202
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	I am keeping all of this in mind. The important thing right now is to get raw frames and see if the output is good enough that people are interested in me doing a portable mod. At the very least, the final design allows you to select different decimations (4:1:1, 4:2:2, etc). Compression is an option as well, thoiugh i haven't implemented that yet in my design. Update: I am letting a pro do the soldering this time, i dropped it off this morning with a friend of mine who does soldering for the EE department, it should be ready this afternoon. After that it should be a matter of hooking it up and capturing frames tonight. Juan

March 26th, 2004, 11:19 AM	#203
Adam Burtle Regular Crew Join Date: Oct 2003 Location: Los Angeles, CA Posts: 173	Glad to hear it, Juan. If you need somewhere to host video or pics, just let me know and i can take care of it for you.

March 26th, 2004, 04:10 PM	#204
Obin Olson Trustee Join Date: Jan 2003 Location: Wilmington NC Posts: 1,414	dude check this out: http://www.photonics.com/ they should have a beam splitter that goes 3 ways for 3 cmos chips..I don't have time to look at the site but someone told me they should have what you want

March 26th, 2004, 05:13 PM	#205
Jason Rodriguez Trustee Join Date: Mar 2003 Location: Virginia Beach, VA Posts: 1,095	<<<-- Originally posted by Juan P. Pertierra : If I am wrong, someone please correct me. -->>> All the next generation Digital Cinema cameras which are inherently designed to be blown up to theatre-size screens are going to be one-chippers: Arri D20, Dalsa Origin, Kinetta HD and the Aaton D-Minima.

March 26th, 2004, 05:19 PM	#206
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	Oh i don't doubt that...however matthew was sugggesting using a 1-chip design for the HD box camera we are talking about here, which, unless the new generation of chips in these cutting edge cameras are cheap, is really not an option. I was talking about currently available chips that we can purchase as consumers for a relatively low price. The whole point of the rockwell HD chips is that they can be had for relatively low price, and they already include all the necessary circuitry which further lowers the price of implementation as well as design complexity.

March 26th, 2004, 05:34 PM	#207
Obin Olson Trustee Join Date: Jan 2003 Location: Wilmington NC Posts: 1,414	exactly.

March 26th, 2004, 06:49 PM	#208
Obin Olson Trustee Join Date: Jan 2003 Location: Wilmington NC Posts: 1,414	when will the frame grabs be ready?

March 26th, 2004, 07:26 PM	#209
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	I am hooking it all up right now. It is 30 wires, but it takes time because if i'm not careful the surface mount connector can come off again and delay everything. However, I decided to upload a very partial image. This image contains ONLY 27% of the imformation of the complete raw frame. ALSO: it still quite a lot of noise that can be seen as dark dots in sharp edges about the image...it's not perfectly focused either because the LCD screen is disconnected, and I can't put my eye against the viewfinder. Once again, this is only the 10 red bits, coming off one CCD only. I will try and upload more partial frames as i carefully hook up more wires. The noise is due to the fact that I am forced to use long thin wires with my test setup, because of the location of my computer. I will try and get completely rid of it once i have all colors hooked up. http://expert.cc.purdue.edu/~pertierr/dvxcap.tiff Juan

March 26th, 2004, 09:00 PM	#210
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	Correction...i misplaced one of the wires and that image is actually 9-bit ...bit 7 was mistakedly hooked up to ground so that image contains actually even less information...also this bad hookup could be the source for the speckled noise.... Eating dinner...then going back to work. :) Juan