Is DV lossless once captured? at DVinfo.net

Aviv Hallale · October 28th, 2006, 07:25 PM

This is a question I've never had answered, but is DV a format that can be rendered out of an NLE multiple times and brought into other software or projects without any noticeable quality loss?

Vincent Rozenberg · October 28th, 2006, 07:51 PM

If you stay in the DV codec without any rendering caused by colour correction or dissolves or so, yes, there will be no loss in quality since you're just coping data.

But if you do any re-rendering after you've put any effect on your material you go back in quality theoretically.

Guillaume Iacino · October 28th, 2006, 09:16 PM

DV format has a color sampling of 4:1:1, meaning that it has less information is available for the color than other formats such as DigiBeta or BetaSX (4:2:2). It isn't the best format for chroma-keying and intense image manipulation - For that, you might actually want to capture your DV as analog Y-C or interpolate the chroma prior to apply the keying or use another format to start with (quoting Adam Wilt). That said, if you are just planning to do simple color correction and fades, you should not see noticeable artifacts.

If you output your DV file back to a DV file, the quality will be the same (except where the luminance and chroma was changed).

Chris Hocking · October 28th, 2006, 10:02 PM

Guillaume, I don't quite understand why capturing your footage as analog Y-C would make any difference. For example, say you shoot footage of a person dancing in from of a green screen with a DV camera. You are recording 1's and 0's on a MiniDV tape. Then when it comes to capturing you have two options. First you capture using Firewire. All the 1's and 0's are "transfered" from the MiniDV tape to the computer. No loss in quality. After that, you capture via an analogue capture card. The 1's and 0's are converted to an analogue siginal in the camera. This signal travels along a wire to an A/D converter. The newly created 1's and 0's are then stored on the computer. There may be some quality loss during the conversion, depending on the quality of the electronics - probably not. Now, I don't see any difference?

I can understand that if you took the video signal directly from the camera (without storing to tape), you could probably get more colour information from the camera (if the analogue output of the camera was BEFORE the compression processing). But I've never been able to understand why CAPTURING from a DV tape will look better if done with analogue hardware. The way I see it is you can't create information that isn't there. I'm under the impression that the loss of information occurs before the images and sound are written to tape.

So, what's the truth? More than likely, I'm completely missing something vital! If you could please explain in more detail that would be fantastic! Thanks in advanced!

Glenn Chan · October 29th, 2006, 01:09 AM

When you need to process the image, you usually need to resample/"uprez" the chroma. i.e. You need to stretch it out horizontally.

If you think in terms of Photoshop:
You need to change the image size of the chroma.
Photoshop has different options for resizing an image. What FCP does is the nearest neighbour resampling method, which yields poor quality results for chroma keying purposes.
Linear interpolation is a little better. I'm not sure, but Photoshop's equivalent might be bilinear??? (Again, not sure there.) But the point is, there are better resampling methods than nearest neighbour.

2- Some of the high-end decks will apply linear interpolation to the upsampled chroma, when converting from 4:1:1 to something else like SDI (which carries a 4:2:2 signal). Doing this can have a workflow advantage because there's no rendering; the upsampling is real-time.

SDI is preferable over Y-C.

Alan James · October 29th, 2006, 05:26 AM

Chris your right. Capturing information from a tape via firewire is better then from S-Video cables. However if your camera did compress the information to your S-video port then on set capturing from your s-video port is best, but cameras do compress video before it gets to your S-Video port. Out of the 2 capturing to tape and transferring to your computer is best. There is a third option but it is only available to DVX-100 users. The Reel Stream Andromeda is a warranty voiding modification to the camera. A usb 2.0 port is installed at a point before the compression takes place. You capture 4:4:4 uncompressed video straight to your hard drive then down convert and edit. But this costs 3000$ probably a new computer and you have to have a DVX-100. Anyway record to tape then transfer to your computer via firewire.

Aviv Hallale · October 29th, 2006, 05:43 AM

At my school, we capture from Mini-DV tapes through a Deck with B-Y-R Cables into an SDI format that takes up about 30 gigs for ten minutes or so. Our editor explained something about that workflow increasing the resolution and uncompressing the video from DV limitations, or something to that effect, but he wasn't really clear.

Guillaume Iacino · October 29th, 2006, 10:17 AM

Quote:

Originally Posted by Chris Hocking

I don't quite understand why capturing your footage as analog Y-C would make any difference

I was quoting Adam Wilt and his article on color sampling.
http://www.adamwilt.com/DV-FAQ-tech.html#colorSampling

"Some people capture DV for chroma keying using an analog Y/C feed, since the analog connection prefilters and smooths the chroma."

Earl Thurston · October 29th, 2006, 11:07 AM

Quote:

Originally Posted by Chris Hocking

Guillaume, I don't quite understand why capturing your footage as analog Y-C would make any difference.

Here are some explanations:

http://www.neopics.com/bluescreen/
http://www.philipwilliams.com/greenscreen.aspx

My Neopics page is somewhat out of date. Fortunately, some keyers now include built-in chroma smoothing features for DV and HDV.

Gene Brockhoff · October 29th, 2006, 12:41 PM

Here is a deep exploration of chroma sampling from Master Nattress http://www.kenstone.net/fcp_homepage..._nattress.html

Vincent Rozenberg · October 29th, 2006, 04:34 PM

Quote:

Originally Posted by Aviv Hallale

At my school, we capture from Mini-DV tapes through a Deck with B-Y-R Cables into an SDI format that takes up about 30 gigs for ten minutes or so. Our editor explained something about that workflow increasing the resolution and uncompressing the video from DV limitations, or something to that effect, but he wasn't really clear.

What you do at school is capturing DV at an uncompressed format. In theory this will improve picture quality after putting effects/dissolves on your material. So that will be the reason to do this.

@ the others: I do not know why the whole discussion of chroma sampling etc. belongs in this tread..?

Mark Williams · October 31st, 2006, 03:26 PM

To add to this if I am using Edius 4 and project setting are Canopus lossless AVI will there also be a loss in quality when rendering?

Regards,

Vincent Rozenberg · October 31st, 2006, 04:03 PM

Theoretically there's always a loss with rendering. But with lossless and uncompressed codecs it supposed to not show up.

Alan James · November 1st, 2006, 01:28 AM

This probably belongs in the After Effects thread but I’m gunna ask it here. Tell me if any of you have had this problem. When I do effects in AE and export using the DV codec it always comes out darker. All the information is there just everything is darker. Why is that?

Emre Safak · November 1st, 2006, 07:26 AM

After Effects renders DV using the crappy Microsoft codec and unfortunately it looks different from others. Try rendering to uncompressed first before handing it over to After Effects.

October 28th, 2006, 07:25 PM	#1
Aviv Hallale Major Player Join Date: Mar 2004 Location: Cape Town, South Africa Posts: 276	Is DV lossless once captured? This is a question I've never had answered, but is DV a format that can be rendered out of an NLE multiple times and brought into other software or projects without any noticeable quality loss?

October 29th, 2006, 12:41 PM	#10
Gene Brockhoff Regular Crew Join Date: Aug 2003 Location: Oakland CA Posts: 46	Here is a deep exploration of chroma sampling from Master Nattress http://www.kenstone.net/fcp_homepage..._nattress.html __________________ "We are basically free, but who we think we are is not" Ram Dass

October 31st, 2006, 03:26 PM	#12
Mark Williams Trustee Join Date: Dec 2002 Location: Deep South, U.S. Posts: 1,526	To add to this if I am using Edius 4 and project setting are Canopus lossless AVI will there also be a loss in quality when rendering? Regards, __________________ Mark videos: http://vimeo.com/channels/3523 Stock: http://www.pond5.com/artist/mark29

October 28th, 2006, 07:51 PM	#2
Vincent Rozenberg Major Player Join Date: Jan 2006 Location: Rotterdam, The Netherlands Posts: 475	If you stay in the DV codec without any rendering caused by colour correction or dissolves or so, yes, there will be no loss in quality since you're just coping data. But if you do any re-rendering after you've put any effect on your material you go back in quality theoretically.

October 28th, 2006, 09:16 PM	#3
Guillaume Iacino New Boot Join Date: Oct 2006 Location: NY Posts: 10	DV format has a color sampling of 4:1:1, meaning that it has less information is available for the color than other formats such as DigiBeta or BetaSX (4:2:2). It isn't the best format for chroma-keying and intense image manipulation - For that, you might actually want to capture your DV as analog Y-C or interpolate the chroma prior to apply the keying or use another format to start with (quoting Adam Wilt). That said, if you are just planning to do simple color correction and fades, you should not see noticeable artifacts. If you output your DV file back to a DV file, the quality will be the same (except where the luminance and chroma was changed).

October 28th, 2006, 10:02 PM	#4
Chris Hocking Major Player Join Date: Dec 2005 Location: Melbourne, Australia Posts: 595	Guillaume, I don't quite understand why capturing your footage as analog Y-C would make any difference. For example, say you shoot footage of a person dancing in from of a green screen with a DV camera. You are recording 1's and 0's on a MiniDV tape. Then when it comes to capturing you have two options. First you capture using Firewire. All the 1's and 0's are "transfered" from the MiniDV tape to the computer. No loss in quality. After that, you capture via an analogue capture card. The 1's and 0's are converted to an analogue siginal in the camera. This signal travels along a wire to an A/D converter. The newly created 1's and 0's are then stored on the computer. There may be some quality loss during the conversion, depending on the quality of the electronics - probably not. Now, I don't see any difference? I can understand that if you took the video signal directly from the camera (without storing to tape), you could probably get more colour information from the camera (if the analogue output of the camera was BEFORE the compression processing). But I've never been able to understand why CAPTURING from a DV tape will look better if done with analogue hardware. The way I see it is you can't create information that isn't there. I'm under the impression that the loss of information occurs before the images and sound are written to tape. So, what's the truth? More than likely, I'm completely missing something vital! If you could please explain in more detail that would be fantastic! Thanks in advanced!

October 29th, 2006, 01:09 AM	#5
Glenn Chan Inner Circle Join Date: Jun 2003 Location: Toronto, Canada Posts: 4,750	When you need to process the image, you usually need to resample/"uprez" the chroma. i.e. You need to stretch it out horizontally. If you think in terms of Photoshop: You need to change the image size of the chroma. Photoshop has different options for resizing an image. What FCP does is the nearest neighbour resampling method, which yields poor quality results for chroma keying purposes. Linear interpolation is a little better. I'm not sure, but Photoshop's equivalent might be bilinear??? (Again, not sure there.) But the point is, there are better resampling methods than nearest neighbour. 2- Some of the high-end decks will apply linear interpolation to the upsampled chroma, when converting from 4:1:1 to something else like SDI (which carries a 4:2:2 signal). Doing this can have a workflow advantage because there's no rendering; the upsampling is real-time. SDI is preferable over Y-C.

October 29th, 2006, 05:26 AM	#6
Alan James Regular Crew Join Date: Jul 2006 Location: Tampa, Florida Posts: 261	Chris your right. Capturing information from a tape via firewire is better then from S-Video cables. However if your camera did compress the information to your S-video port then on set capturing from your s-video port is best, but cameras do compress video before it gets to your S-Video port. Out of the 2 capturing to tape and transferring to your computer is best. There is a third option but it is only available to DVX-100 users. The Reel Stream Andromeda is a warranty voiding modification to the camera. A usb 2.0 port is installed at a point before the compression takes place. You capture 4:4:4 uncompressed video straight to your hard drive then down convert and edit. But this costs 3000$ probably a new computer and you have to have a DVX-100. Anyway record to tape then transfer to your computer via firewire.

October 29th, 2006, 05:43 AM	#7
Aviv Hallale Major Player Join Date: Mar 2004 Location: Cape Town, South Africa Posts: 276	At my school, we capture from Mini-DV tapes through a Deck with B-Y-R Cables into an SDI format that takes up about 30 gigs for ten minutes or so. Our editor explained something about that workflow increasing the resolution and uncompressing the video from DV limitations, or something to that effect, but he wasn't really clear.

October 31st, 2006, 04:03 PM	#13
Vincent Rozenberg Major Player Join Date: Jan 2006 Location: Rotterdam, The Netherlands Posts: 475	Theoretically there's always a loss with rendering. But with lossless and uncompressed codecs it supposed to not show up.

November 1st, 2006, 01:28 AM	#14
Alan James Regular Crew Join Date: Jul 2006 Location: Tampa, Florida Posts: 261	This probably belongs in the After Effects thread but I’m gunna ask it here. Tell me if any of you have had this problem. When I do effects in AE and export using the DV codec it always comes out darker. All the information is there just everything is darker. Why is that?

November 1st, 2006, 07:26 AM	#15
Emre Safak Trustee Join Date: Feb 2004 Location: Brookline, MA Posts: 1,447	After Effects renders DV using the crappy Microsoft codec and unfortunately it looks different from others. Try rendering to uncompressed first before handing it over to After Effects.