4:4:4 12-bit Uncompressed DVX100

Mark Grgurev · May 5th, 2004, 08:01 PM

Its probably impossible to do but I was just wondering. Would you be able to make a memory card reader to make custom gamma curves for the DVX100? If so, it would be great if you make one after ur done with this project.

Sorry for being a little off topic.

Juan P. Pertierra · May 5th, 2004, 09:21 PM

Stephen, Ben, and the rest <g>:

Regardless of the thread getting a little 'heated', i think just about every post contains useful information and good , well informed opinions, even if we feel passionate about our points. I wouldn't expect any of us not to. :)

To answer Stephen's question, yes, the start/stop can be handled over firewire, just like if the capture box circuitry where a PC itself. The circuitry in the box accesses the drive just like a PC and when the recording button is pushed, starts writing files to the drive.

However, i think there's a point we are overlooking here that eliminates SATA directly, and that is that right now the FW800 port on the device can also be connected to a computer directly, for recording on the machine's hard drive. This will be done through a device driver on the machine side, which will allow the camera to access the drive on the PC. This is simply not an option with SATA because it is simply not as flexible as firewire. I can write a device driver on the PC side to do just about anything, but if we go straight to SATA, all we can do is hook directly to hard drives.

SATA is definitely a better hard drive interface, because that's what it is designed to do..however FW800 is not primarily a hard drive interface, but rather a flexible interface that can be used for just about anything...since this app involves more than just writing to a drive, firewire is a good choice.

Let's hear it! :)

Juan

Rodger Marjama · May 5th, 2004, 09:54 PM

Quote:

Originally posted by Stephen van Vuuren : Ben:

That's one problem with this thread - it's so long and has been going on for so many months it's hard to keep all the tangents up to speed.

The Agus35 suffers from the same issues. I would like to see some type of branching thread for these heavy topics here - I'll mention something to Chris as it would make it far easier for people coming into long-standing threads easier.

Anyway, no hard feelings and sorry for the confusion.

If you've got moderator privileges, just lock any threads that reach 500 replies and start a new thread.

For this thread it should be "4:4:4 12-bit Uncompressed DVX100 - Part II"

-Rodger

Stephen van Vuuren · May 5th, 2004, 09:59 PM

There's no official moderator here, so I've been holding off, though Chris had talked about me doing it here. I sent him an email earlier.

Rodger Marjama · May 5th, 2004, 10:07 PM

<<<-- Originally posted by Stephen van Vuuren : There's no official moderator here, so I've been holding off, though Chris had talked about me doing it here. I sent him an email earlier. -->>>

So I see...

Make sure to link to the original thread on the opening post if you guys do this. We have a few threads for our products on some different boards that went over 2000 replies if I remember right. It also helps indexing and search engines if you limit thread sizes. Setting an arbitrary post number seems to work well.

-Rodger

David Warrilow · May 5th, 2004, 11:12 PM

Hi,

fist I have to say thankyou to Juan for his brilliant virtuoso work on this project, which is not only very exciting but is also giving all of us a rare glimpse 'behind the curtain' on our camcorders. What we pay for is not always what we 'get' with these units...

As I understand it, the normal lattitude rating on prosumer camcorders is only a few stops (5-8?) with the higher end cameras pushing 8-11 stops (correct me if I'm wrong on the numbers here). But it's apparent that the latitude coming off the DVX100 chips is far greater than anything that we've been accustomed to in the past with our cameras. Does anyone know how to measure this difference in range? Even an educated estimation would be enlightening since it would give us a better indication of what our cameras are capable of, rather than what we're used to getting from them.

Again, great work Juan.

Best,

DW

Juan P. Pertierra · May 6th, 2004, 04:01 AM

One of the tests I am planning to do today or tomorrow, is to take outdoor and indoor snapshots of the same scene with my 35mm camera, and the DVX raw. This will give us a better idea of what the latitude is and i for one can't wait to play withe footage and see if I can get it anywhere close to the look of film.

I am going with ISO400 film because i read somewhere this is what most CCD's correspond to roughly...

I will have the film images scanned off from the negatives at 2K resolution and put on CD at the lab so i can post them...

Juan

Luis Caffesse · May 6th, 2004, 08:58 AM

Juan,

Most people seem to agree that the DVX100 is roughly the equivelant of 320ASA (when shooting at 24P with 0db and a 1/48 shutter).

http://www.dv.com/jive3/thread.jspa?...000694&start=0

http://www.dvxuser.com/cgi-bin/DVX/Y...num=1068605987

I did see one post on Dvinfo which claimed a 640ASA equivelance, but the majority of posts claimed 320. I think the 640ASA may be true when shooting at 60i with a 1/60th shutter speed, which is not the case here.

I realize you can't get 320ASA film for your still camera, but 400 should be close enough.

-Luis

John Cabrera · May 6th, 2004, 08:06 PM

Juan,

Did you ever figure out why the darker areas of the DV clip you uploaded are considerably brighter than the darker areas of your raw capture? The explanation about the distribution of 24bits across 8bit codec doesn't seem to explain why those areas would have to be brightened by the codec as well. Only why information in the brightest areas of the frame might be sacrificed. It would make more sense that panasonic brightens the footage at the same stage that color correction functions occur in the camera in order to help out overall with low lighting. I don't know... I may be way off... but if you captured a raw clip and then captued the same thing with a ND filter on enough to match the darks and midtones of your raw footage, we'd get a better idea of exactly how much information in that sky is actually being lost... please explain what I might be missing if you can.

John

Juan P. Pertierra · May 7th, 2004, 10:38 AM

John,

All DV/RAW comparisons where frames shot under the exact same conditions, so you can see what the DV output for the specific RAW data is.

I'm not sure about the darks either, but do note that the adjustment from 12-bits to 8-bits is not the only correction done by the camera. There are tons of other modifications done to the video which could be causing this...it could also be my settings on the camera which i haven't checked in a while since the joystick is disconnected.

It is possible, however, that since the color is not paletted, they moved the entire limited 8-bit range (which doesn't cover the entire latitude) to somewhere in the middle, sacrificing a little low end, a lot of the high end, but compressing the 8-bits in the brightnesses that they thought you would work under...this would give more color precision in that specific range, while sacrificing high end and low end latitude.

Just a theory...

Juan

Juan P. Pertierra · May 7th, 2004, 11:06 AM

Oh, and an update:

I started on the code work yesterday, but finally gave in to a couple of days without sleep <g>

However, i did snap pictures of the outdoor shot with my 35mm camera and the DVX at the same time...i got em developmed but even though i was using the on-camera spot meter, they came out a bit over exposed, and to all honesty the DVX captured a way better scene but is not a good measure.

I will try to do it again today or tomorrow, and try some slightly underexposed shots by the camera's spot meter. I do have a couple of indoor shots of my lamp which makes a good test of the range in brightness, so I am going to take some shots with the DVX of that as well, and see what it looks like...

Juan

Leif Thomas · May 7th, 2004, 11:48 AM

I'd like to post my try of the colorcorrected cap9_RAW:

http://www.filmerforum-koeln.de/user...cap9_RAWcc.tif

the colors now seem more like the DV (although - of course - more natural) und I also lightend it up a bit...

I'd LOVE to see your 35mm / RAW / DV comparison! But why scanning the 35mm on 2K? A simple NTSC-Scan would be better to compare.

BTW:
Do you already have a name for your system? I think it's time :)

Nicholi Brossia · May 7th, 2004, 12:23 PM

Juan,
Would it be possible to write your code so that the hard drive records a flipped image, or even a way to toggle back and forth from normal to flipped modes?
Like many others on this board, I think your 4:4:4 system will coordinate perfectly with a 35mm adapter. Recording the image with correct orientation would helpful in the longrun. Although, if its too much of a hassle, flipping in post is pretty easy too.

Juan P. Pertierra · May 7th, 2004, 12:34 PM

Nicholi,

That's not a hassle at all, and in fact already implementing it...the function i wrote to deal with the inverted blue channel can invert images in any orientation, and thus you can save them inverted vertically if you'd like.

Luis Caffesse · May 7th, 2004, 12:40 PM

Just to make sure everyone is saying the same thing:

The 35mm adapters flip the image vertically AND horizontally, right?

so the image would have to be flipped not only vertically (like the blue channel you mention Juan) but horizontally as well.
(or rotated 180 degrees, same thing)

As was already mentioned, this would be simple enough to do in post, but it would be a nice addition if possible.

-luis

May 5th, 2004, 09:59 PM	#529
Stephen van Vuuren Space Hipster Join Date: Oct 2001 Location: Greensboro, NC Posts: 1,508	There's no official moderator here, so I've been holding off, though Chris had talked about me doing it here. I sent him an email earlier. __________________ stephen v2 www.insaturnsrings.com

May 5th, 2004, 10:07 PM	#530
Rodger Marjama Regular Crew Join Date: Sep 2002 Location: Loveland, CO Posts: 101	<<<-- Originally posted by Stephen van Vuuren : There's no official moderator here, so I've been holding off, though Chris had talked about me doing it here. I sent him an email earlier. -->>> So I see... Make sure to link to the original thread on the opening post if you guys do this. We have a few threads for our products on some different boards that went over 2000 replies if I remember right. It also helps indexing and search engines if you limit thread sizes. Setting an arbitrary post number seems to work well. -Rodger __________________ Rodger Marjama www.speedwing.net

May 6th, 2004, 08:06 PM	#534
John Cabrera Regular Crew Join Date: Apr 2004 Location: California Posts: 67	latitude Juan, Did you ever figure out why the darker areas of the DV clip you uploaded are considerably brighter than the darker areas of your raw capture? The explanation about the distribution of 24bits across 8bit codec doesn't seem to explain why those areas would have to be brightened by the codec as well. Only why information in the brightest areas of the frame might be sacrificed. It would make more sense that panasonic brightens the footage at the same stage that color correction functions occur in the camera in order to help out overall with low lighting. I don't know... I may be way off... but if you captured a raw clip and then captued the same thing with a ND filter on enough to match the darks and midtones of your raw footage, we'd get a better idea of exactly how much information in that sky is actually being lost... please explain what I might be missing if you can. John

May 7th, 2004, 11:48 AM	#537
Leif Thomas Tourist Join Date: May 2004 Location: Germany, NRW Posts: 1	Color Corrected I'd like to post my try of the colorcorrected cap9_RAW: http://www.filmerforum-koeln.de/user...cap9_RAWcc.tif the colors now seem more like the DV (although - of course - more natural) und I also lightend it up a bit... I'd LOVE to see your 35mm / RAW / DV comparison! But why scanning the 35mm on 2K? A simple NTSC-Scan would be better to compare. BTW: Do you already have a name for your system? I think it's time :)

May 7th, 2004, 12:23 PM	#538
Nicholi Brossia Major Player Join Date: Jun 2003 Location: Oklahoma City, OK Posts: 389	Juan, Would it be possible to write your code so that the hard drive records a flipped image, or even a way to toggle back and forth from normal to flipped modes? Like many others on this board, I think your 4:4:4 system will coordinate perfectly with a 35mm adapter. Recording the image with correct orientation would helpful in the longrun. Although, if its too much of a hassle, flipping in post is pretty easy too. __________________ Nicholi Brossia

May 5th, 2004, 08:01 PM	#526
Mark Grgurev Regular Crew Join Date: Dec 2003 Location: Fairview,nj Posts: 137	Its probably impossible to do but I was just wondering. Would you be able to make a memory card reader to make custom gamma curves for the DVX100? If so, it would be great if you make one after ur done with this project. Sorry for being a little off topic.

May 5th, 2004, 09:21 PM	#527
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	Stephen, Ben, and the rest <g>: Regardless of the thread getting a little 'heated', i think just about every post contains useful information and good , well informed opinions, even if we feel passionate about our points. I wouldn't expect any of us not to. :) To answer Stephen's question, yes, the start/stop can be handled over firewire, just like if the capture box circuitry where a PC itself. The circuitry in the box accesses the drive just like a PC and when the recording button is pushed, starts writing files to the drive. However, i think there's a point we are overlooking here that eliminates SATA directly, and that is that right now the FW800 port on the device can also be connected to a computer directly, for recording on the machine's hard drive. This will be done through a device driver on the machine side, which will allow the camera to access the drive on the PC. This is simply not an option with SATA because it is simply not as flexible as firewire. I can write a device driver on the PC side to do just about anything, but if we go straight to SATA, all we can do is hook directly to hard drives. SATA is definitely a better hard drive interface, because that's what it is designed to do..however FW800 is not primarily a hard drive interface, but rather a flexible interface that can be used for just about anything...since this app involves more than just writing to a drive, firewire is a good choice. Let's hear it! :) Juan

May 5th, 2004, 11:12 PM	#531
David Warrilow Regular Crew Join Date: Apr 2003 Location: Brisbane, Australia Posts: 106	Hi, fist I have to say thankyou to Juan for his brilliant virtuoso work on this project, which is not only very exciting but is also giving all of us a rare glimpse 'behind the curtain' on our camcorders. What we pay for is not always what we 'get' with these units... As I understand it, the normal lattitude rating on prosumer camcorders is only a few stops (5-8?) with the higher end cameras pushing 8-11 stops (correct me if I'm wrong on the numbers here). But it's apparent that the latitude coming off the DVX100 chips is far greater than anything that we've been accustomed to in the past with our cameras. Does anyone know how to measure this difference in range? Even an educated estimation would be enlightening since it would give us a better indication of what our cameras are capable of, rather than what we're used to getting from them. Again, great work Juan. Best, DW

May 6th, 2004, 04:01 AM	#532
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	One of the tests I am planning to do today or tomorrow, is to take outdoor and indoor snapshots of the same scene with my 35mm camera, and the DVX raw. This will give us a better idea of what the latitude is and i for one can't wait to play withe footage and see if I can get it anywhere close to the look of film. I am going with ISO400 film because i read somewhere this is what most CCD's correspond to roughly... I will have the film images scanned off from the negatives at 2K resolution and put on CD at the lab so i can post them... Juan

May 6th, 2004, 08:58 AM	#533
Luis Caffesse Major Player Join Date: Apr 2004 Location: Austin, Texas Posts: 704	Juan, Most people seem to agree that the DVX100 is roughly the equivelant of 320ASA (when shooting at 24P with 0db and a 1/48 shutter). http://www.dv.com/jive3/thread.jspa?...000694&start=0 http://www.dvxuser.com/cgi-bin/DVX/Y...num=1068605987 I did see one post on Dvinfo which claimed a 640ASA equivelance, but the majority of posts claimed 320. I think the 640ASA may be true when shooting at 60i with a 1/60th shutter speed, which is not the case here. I realize you can't get 320ASA film for your still camera, but 400 should be close enough. -Luis

May 7th, 2004, 10:38 AM	#535
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	John, All DV/RAW comparisons where frames shot under the exact same conditions, so you can see what the DV output for the specific RAW data is. I'm not sure about the darks either, but do note that the adjustment from 12-bits to 8-bits is not the only correction done by the camera. There are tons of other modifications done to the video which could be causing this...it could also be my settings on the camera which i haven't checked in a while since the joystick is disconnected. It is possible, however, that since the color is not paletted, they moved the entire limited 8-bit range (which doesn't cover the entire latitude) to somewhere in the middle, sacrificing a little low end, a lot of the high end, but compressing the 8-bits in the brightnesses that they thought you would work under...this would give more color precision in that specific range, while sacrificing high end and low end latitude. Just a theory... Juan

May 7th, 2004, 11:06 AM	#536
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	Oh, and an update: I started on the code work yesterday, but finally gave in to a couple of days without sleep <g> However, i did snap pictures of the outdoor shot with my 35mm camera and the DVX at the same time...i got em developmed but even though i was using the on-camera spot meter, they came out a bit over exposed, and to all honesty the DVX captured a way better scene but is not a good measure. I will try to do it again today or tomorrow, and try some slightly underexposed shots by the camera's spot meter. I do have a couple of indoor shots of my lamp which makes a good test of the range in brightness, so I am going to take some shots with the DVX of that as well, and see what it looks like... Juan

May 7th, 2004, 12:34 PM	#539
Juan P. Pertierra Major Player Join Date: Jul 2003 Posts: 479	Nicholi, That's not a hassle at all, and in fact already implementing it...the function i wrote to deal with the inverted blue channel can invert images in any orientation, and thus you can save them inverted vertically if you'd like.

May 7th, 2004, 12:40 PM	#540
Luis Caffesse Major Player Join Date: Apr 2004 Location: Austin, Texas Posts: 704	Just to make sure everyone is saying the same thing: The 35mm adapters flip the image vertically AND horizontally, right? so the image would have to be flipped not only vertically (like the blue channel you mention Juan) but horizontally as well. (or rotated 180 degrees, same thing) As was already mentioned, this would be simple enough to do in post, but it would be a nice addition if possible. -luis