There has been some disagreement about the potential for generation loss in DV video via Firewire in Avid. The idea is that DV video, as data, is transferred and ‘captured’ as a simple data stream via a Firewire connection – there should be no loss, as it is simply moving data back and forth.
But is that the case?
I created an image in Photoshop of a variety of test patterns, imported it into Avid as 1:1 media, then output from Avid to a DVCAM tape. Then I proceeded to capture from the tape, put the captured footage back on the tape, and capture off again.
With these clips I then exported, from Avid, single TIFF frames from each clip and compared those frames in two ways:
In Photoshop – Placing Generation 1 (the first capture from the tape) on Layer 1, I then stacked up each of the other generations above it, set their transfer mode to ‘Difference’ and applied an Adjustment Layer with Levels set with 1-24 (to highlight any differences).
With a Checksum – Using WinMD5Free I got the MD5 checksum value of each frame.
My conclusion? There’s no difference. No generational loss is introduced with DV25 capture from a DV source via Firewire into Avid and subsequent layback via Firewire to DV.
However it is important to note, that any changes to the picture will introduce variations. In Avid, it is best to ensure that all new media (Titles, Effects, Imported graphics) is at a higher quality than DV25 (Ideally 1:1 or alternatively DV50).
I won’t post any picture, they are boring, and all the same, however the checksum details are below:
Generation 1 (first capture from Tape):
4128ef910dbe661d4238574ed3b3fbe7
Generation 2 (Gen 1, put on tape, then captured again):
4128ef910dbe661d4238574ed3b3fbe7
Generation 3 (Gen 2, put on tape, then captured again):
4128ef910dbe661d4238574ed3b3fbe7
Generation 4 (Gen 3, put on tape then captured again):
4128ef910dbe661d4238574ed3b3fbe7
#1 by Matt Dwen - November 13th, 2007 at 12:27
You must have been bored. But good to know.
#2 by Patrix - October 10th, 2009 at 11:42
Came across this blog thru the AVID forums.
Generation loss (however slight) occurs when you output from the AVID to DVCAM via any conduit. DVCAM has a native compression ratio of 5:1. When recording to DVCAM the sampled video data is reduced by a factor of 5:1 using bit rate reduction. This loss has nothing to do with AVID output, rather the media you choose to output to.
#3 by Dylan - October 10th, 2009 at 18:21
Capturing any DV25 footage (DVCAM, Standard DV or DVCPRO) into Avid, via Firewire in the correct format (DV25 420 for PAL DVCAM for example) introduces no loss at all. That is because a Firewire transfer is not moving video as such, but digital data. It moves the DV data from the tape to the harddrive without changing it.
However if you take that same tape and capture it by SDI then you’re capturing a baseband video stream and recompressing it. The worstcase scenario basically being capturing DV25 footage into a DV25 Avid resolution using any method other than Firewire. In that scenario the DV25 compressed footage is being unpacked into a baseband video stream, sampled by the Avid and recompressed back to DV25. This will create significant loss that will become visible generational loss quite quickly.
If the source footage is DV25, and is captured (transferred really) via Firewire into DV25 Avid media, edited and output again via Firewire to a DV25 deck there is absolutely no generation loss. The data written to the tape is exactly, bit for bit, the same as what was ‘captured’. However as soon as the footage as any translational effect (scaling, resizing) applied, or has to undergo and pixel manipulation (colour correct, transitions) then new media is created and will be compressed with a DV compression at some point (assuming it’s going back to a DV tape) – at which point change has occurred and generational compression becomes an issue.
#4 by Patrix - October 13th, 2009 at 06:27
http://www.sony.ca/dvcam/pdfs/dvcam%20format%20overview.pdf
…or you can do a Google search for DVCAM format overview to read SONY’s white paper document on the DVCAM format. Once you have the document, see the Video Signal Processing section, page 6. It describes the sampling, blocking shuffling, etc. processes of how the bits are stored on the tape.
It doesn’t matter how the bits of data are recorded to the format, the compression process is applied during the ‘Sampling’ (“The sampled video data is reduced by a factor of 5:1 using bit rate reduction…”) and ‘Shuffling’ (“In the DVCAM/DV formats, the compression process is applied over five macro blocks gathered from five different super blocks.”) phases of the signal processing, hence the 5:1 compression.
#5 by Dylan - October 13th, 2009 at 18:19
Patrix, I am very familiar with DV compression.
The point is that compression, if you’re shooting on a DV25 format, is applied at the time of acquisition. What is stored on the tape is a 25Mbit/s stream of compressed video data.
In ‘capturing’ the DV25 footage into an Avid DV25 resolution via Firewire you are transferring that video essence (the 25Mbit/s compressed video data) as a data stream onto the computer, where the data is wrapped in an MXF or OMFI wrapper. It is lossless – the data is transferred from the tape (or disc, or drive) without change. No further compression or recompression occurs.
Whereas, if you were to capture the same DV25 footage using a baseband connection (SDI or Analogue Component) then that video would be decompressed from the player, and then recompressed (the same 25Mbit/s Intraframe DCT compression) by the Avid software. For a DCT compression this can be quite damaging, especially if any translational shifts occur, in which case the DCT blocks (between acquisition compression, and subsequent NLE compression) become misaligned will significantly degrade the image.
If capturing via Firewire in Avid, from a DV25 source, then a matched DV25 resolution is fine, and infact lossless (if possible all subsequent rendering should be at a greater resolution, such as 1:1). However if the capture is taking place via SDI or Analogue, then it’s vital that a greater resolution be used (1:1 ideally).
However if you capture DV 25 footage
#6 by Patrix - October 16th, 2009 at 12:24
Hey Dylan, same here, very familiar, from an engineering perspective. Final word for me on this issue is the fact that when footage is recorded to DVCAM, compression is native to the tape format, no way around it, native to the format, whatever resolution you’re talking about recording to the tape, including 1:1.
#7 by Dylan - October 16th, 2009 at 13:16
Indeed DV is a lossy format (indeed, see my much older post Avid Resolutions Compared) but once that compression has been committed to tape there has been many discussions online about how best to deal with it – as DV captured with Firewire, as DV captured Analogue or SDI, as 1:1 captured with Firewire, or 1:1 with Analogue or SDI.
In my opinion, as was the point of this post, the best option is to acquire as DV25, via Firewire. The original DV material can be moved to and from DV tape via firewire infinitely without any further loss of quality – generationless.