I have a question for those in the know. I have ordered a Blackmagic Intensity Pro card to capture in HDMI. My main reason for buying the card is for the clenest best quality capture I can get . At the Blackmagic website it says you can capture from your HDV camera to HDMI and import with no compression.
I have been told by several people here that thats not true unless you are shooting live.
I called Blackmagic and the Tech on the phone says that its true that when you record to tape it is compressed to Mpeg2, but when you import through the intensity card and HDMI it is "uncompressed " back to its original state.
I dont have the expertise to know which is correct so Im wondering if I should keep the card ( it hasnt arrived yet).
My main goal was to import uncompressed and start playing around with Green Screen so I wanted a wider color band.
I do realize the Tech works for Blackmagic so Im wondering if I will see any benifit as I will never be shooting live.
What do you think ?

Thanks

If the video has been compressed to go to tape, how could it ever have the results/damage of that compression removed? If you uncompress data that has gone through a lossy compression, you can't get back what was lost. There is no magic that can bring back what wasn't put into the data stream on the tape.

Don't confuse HDV compression with something like .ZIP compression. When a data file is compressed in a LOSSLESS manner, it can be reconstructed perfectly. HD video is so compressed when put in the HDV format that there is information lost. From what I have seen, it seems that lossless compression with video can reduce data to about a 3:1 level. That is, lossless compression can make the data one third it's uncompressed size. HDV cameras typically have enough data coming off the imaging sensor to make about 8 Gigabytes of data per minute if my calculations are correct. Even if I am off a bit, HD has a tremendous data flow. Without compression, an HDV tape would only hold a couple minutes of video. At 3:1 lossless compression, a tape would still only hold less than ten minutes of footage. The only way to store all that data in a workable format is to use a "LOSSY" compression format.

Correct me if I'm wrong but the issue is, if you record to tape the data is compressed and as Marcus said you can't get back what isn't there, but if you capture via HDMI directly to a computer(aka live), it is not compressed.
Now finding a computer to keep up may be an issue.

We built a system to capture via this card to demo at broadcast live (Which, incidently, started about 10 minutes ago) and it needed A Quad mac pro with 8Gb Ram running FCPs2 for starters to handle the live uncompressed HD.

So if you're around London's earls court in the next few days head to the apple stand to see this in action :)

We built a system to capture via this card to demo at broadcast live (Which, incidently, started about 10 minutes ago) and it needed A Quad mac pro with 8Gb Ram running FCPs2 for starters to handle the live uncompressed HD.


Not really, you'll be fine with a lower end Mac Pro as well with 4 GB of RAM which is the recommended RAM amount for HD by Apple.

I will be sending this card back and I feel the advertising at the blackmagic site is somewhat missleading. I posted here to get the real story . You cant capture uncompressed once you go to tape so why advertise it can? Perhaps the website wording should read "if you are shooting live you can capture uncompressed."
That way people wont be missled. And next time I will ask questions before I pull out the credit card

Hi,

The HDMI output is 4:2:2 uncompressed video. Even if the footage is compressed to tape, it is not sent out as native HDV video. The camera applies filtering and sends the video out as an uncompressed video stream which means you can capture to an uncompressed codec or into a more edit friendly codec like DVCPRO HD.

Similar, when you capture off XDCAM or HDCAM, are you capturing compressed or uncompressed given that all data going to tape is compressed lossy data but the SDI / HD-SDI output is uncompressed 4:2:2 in full raster?

Hi,

The HDMI output is 4:2:2 uncompressed video. Even if the footage is compressed to tape, it is not sent out as native HDV video. The camera applies filtering and sends the video out as an uncompressed video stream which means you can capture to an uncompressed codec or into a more edit friendly codec like DVCPRO HD.

Similar, when you capture off XDCAM or HDCAM, are you capturing compressed or uncompressed given that all data going to tape is compressed lossy data but the SDI / HD-SDI output is uncompressed 4:2:2 in full raster?
Hi Kristian,
Could you possibly suggest the best way to get the highest quality through FCPS2? I am using JVC GY-HD111E ,new macbookpro 4GB FCPS2 sorry to be upfront but it sounds like you know your stuff.
Mick

Sorry John, but i've read the PDF brochure on this product....
And it's pretty obvious...They categorically state that you can divert the HDV tape compression, and shoot straight to the Hard drives uncompressed....

I'm sorry you were lead to believe otherwise..
But as for your intended workflow goes, i don't see any other method of getting 4:2:2 colour, except via live capture....

But the main intention for this card (IMHO), is for HD monitoring..And of course....People who film with their camcorders hooked up to their computers.
Great for studio work...Impossible for event videographers..

Now, if your workflow is intended for NLE's, compositing,and chromakeying online for further editing, then perhaps you can look into shooting one of the variants of DVCPro50, or something that allows for extra chroma information...

"And next time I will ask questions before I pull out the credit card"
Yes, of course....
But investments like this always need to be seriously double checked and referred back to people on these forums and such...
Especially Blackmagic's Decklink cards..They get quite confusing, and it's very easy to make an incorrect purchase..

I'm a bit of a savage when it comes to compression codecs and computer cards, but I know a good looking picture when I see it.

I just got my new MacPro (quad 2.66) running with the Blackmagic card and captured a project shot on my XL-A1 via the HDMI out of my HV-20 into FCP with the ProRes422 codec., and I think it looks fantastic. Using the 3-way color corrector (I'm saving learning Color for when I have some down time), I can already see a massive difference vs. DV, not just in the resolution but the way the color space handles the color correction without blocking up.

As far as I'm concerned, the Blackmagic card is delivering great results for what I wanted to do with it. But again--I haven't tried any chromakeying.

I'm a bit of a savage when it comes to compression codecs and computer cards, but I know a good looking picture when I see it.

I just got my new MacPro (quad 2.66) running with the Blackmagic card and captured a project shot on my XL-A1 via the HDMI out of my HV-20 into FCP with the ProRes422 codec., and I think it looks fantastic. Using the 3-way color corrector (I'm saving learning Color for when I have some down time), I can already see a massive difference vs. DV, not just in the resolution but the way the color space handles the color correction without blocking up.

As far as I'm concerned, the Blackmagic card is delivering great results for what I wanted to do with it. But again--I haven't tried any chromakeying.Charles, I'm sorry but I'm a little confused by this post. I imagine you mean an XH-A1 (those Canon model numbers are truly annoyingly close). Also do you really mean "DV," which is standard definition video or are you comparing to "HDV" which is high def?

Finally, but most importantly: Are you saying that you get better results playing a previously shot HDV tape in your HV-20 and outputting via HDMI to an Intensity card, instead of just capturing from FireWire? If that's the case, that IS very significant, IMHO. And it seems to imply that the HV-20's HDMI is performing some kind of favorable up-resing when HDV tapes are played.

But in all honesty, I've never heard of this, and it seems to go against what I've heard so far about how HDMI in the HV-20 works.

Thanks for having me clarify, Peter. I wrote that too quickly, obviously.

Yes of course I meant XH-A1. Having spent the day with a producer discussing the various merits of the HVX200 vs HDX900 (thankfully not mentioning the HVX500 in there), I'm at meltdown when it comes to this camera designations. I wish Panasonic had learned from the Varicam's success and given 'normal" names to the rest of their product line. Sure it may have been cutesy, but we wouldn't be wrestling with this stuff.

I did actually mean DV in the sense that I've been used to seeing the DV codec respond a certain way to color correction and I would expect at least the same or worse from HDV, but migrating (what's a better term? transcoding? recapturing?) the HDV footage to ProRes422 seems like a visible improvement over the original codec for color correction work.

I thought I had read that bringing the footage out through the HDMI port does result in better quality than through firewire (and subsequently transcoding it to a more robust format). This is consistent with what Kristian has said here. I guess the question is, what ultimately is the difference between the two processes, seeing as they both start with HDV and end with ProRes422 (or whatever final format is selected), the difference being the pipeline that they are pushed through. Kristian, can you enlighten us as to what exactly makes the HDMI capture better?

(like I said, I'm a savage, and just being able to work with HD in any form is enough of a miracle for me at the moment--the old G4 just wasn't cutting it. BTW, site sponsor Zotz Digital configured the Mac Pro for me and did a nice job).

Charles,

I'm no expert myself, by ANY means, but this how I understand HDMI vs Firewire:

If you are capturing LIVE from the HV-20's HDMI port to a computer with an Intensity card, you will get a 1080 by 1920 image with 4,2,2 color sampling. This is because you have avoided the HDV codec entirely. To handle this data stream you will need to do some on the fly compression by high quality codec like Cineform or ProRes422.

The lower resolution 1080 by 1440 and 4,2,0 color sampling is a result of the HDV codec; they're part of its specs. So if you record to TAPE using an HDV camcorder, then 1080 X 1440 and 4,2,0 color is what's laid down on the tape and that's all you get.

It makes no difference what port you subsequently use, HDMI or Firewire, to capture the tape to your NLE. You will still get an HDV quality stream because that's all that's on the tape. (If this is not the case, and Canon's HDMI port somehow intelligently upreses HDV taped footage, I'd be really really surprised.)

So it seems like the improvement you're experiencing is from getting out of the HDV codec as soon as possible and color correcting, etc footage compressed by the ProRes422 codec.

I would imagine if you captured via Firewire and converted to ProRes422, you'd get the exact same results.

Hi,

When you capture HDV or DV video via firewire, what you're doing is essentially copying data that is already stored on tape to the computer hard disk. There is no change to the signal. It's simply moving 1s and 0s across the firewire connection.

However, HDMI is not data transfer. It is an actual video signal and in order for the camera to play out your HDV tape as a video signal, it has to up res the footage to 1920x1080 at 4:2:2. This is the only way it can transmit that information to the Intensity card or any HDMI capable display. This is done regardless of whether you send video out via HDMI or the analog component connection.

When you are capturing 'live', the camera bypasses the HDV compressor chip and sends the signal out via HDMI.

When you are capturing off a HDV tape, the camera decompresses the data on tape, filters and resamples the data before sending it out as 1920x1080 4:2:2 via HDMI.

Either way, you're getting a 1920x1080 4:2:2 signal off HDMI. The question is how good a job the camera is doing when it is up res-ing the HDV video compared to capturing native HDV via firewire and then transcoding it to ProRes 422. In this instance, it may be possible (and I'm speculating here) that the HV20 is doing a better job than going via the software transcoding route.

Well this is actually REALLY interesting, and I wholeheartedly apologize to Charles.

So the HV-20 intelligently upreses when playing an HDV tape out of its HDMI port. And there is a good chance that HV-20 >>> HDV tape >>> HV-20 HDMI port >>> good quality codec yields a significantly better result than HV-20 >>> HDV tape >>> good quality codec. (Though this would still not be as good doing a live HDMI capture.)

Very intersting. Anyone else doing this?

Thanks very much Charles and Kristian!

Yes the up res-ing does indeed seem to smooth the mpeg encoding in the footage. The real hit is in storage. It requires a lot of space.

My system, a dual quad core handles it fine capturing to a dedicated raid. The only thing is it seems a little finicky. When I switch back and forth between the Matrox card and the infinity card.

The captured footage has to be rendered and does not play in realtime. That is a pain. I stopped using the intensity for that reason.

I hope that new drivers will come out that makes this thing work better. I do wish there was a better explanation on the available codecs for capture into Premiere Pro.

Kristian, I have an email out to your tech support dept...when the Mac Pro goes to sleep, the card loses the handshake to the monitor (JVC HD CRT, through the component outs on the BM Intensity Pro) and won't reconnect unless I restart the computer. Worse, if I try to work in FCP with the board and the monitor not talking, it causes FCP to hang up and I have to force quit.

help?!

Yes the up res-ing does indeed seem to smooth the mpeg encoding in the footage. The real hit is in storage. It requires a lot of space...Ron, you could use a really good codec when you capture, like Cineform's, if you want to save some space.

The real ? seems to be "Does capturing an HDV tape out of the HDMI port to a high quality codec give a better result than capturing an HDV tape out of the Firewire port to a high quality codec?"

The answer seems to be "Yes."

The question is how good a job the camera is doing when it is up res-ing the HDV video compared to capturing native HDV via firewire and then transcoding it to ProRes 422. In this instance, it may be possible (and I'm speculating here) that the HV20 is doing a better job than going via the software transcoding route.

If the HV20, with its teensy tiny processor, can up-convert in real-time better than a full-size computer can do with unlimited time, then the software is braindamaged. However, both should do an equally excellent job given that up-conversion has been a solved problem for a long time.

The real question seems to be "Does capturing an HDV tape out of the HDMI port give to a high quality codec give a better result than cpaturing an HDV tape out of the Firewire port to a high quality codec?"

The answer seems to be "Yes."

If the answer isn't "they are the same", then one of them is poorly done.

If the HV20, with its teensy tiny processor, can up-convert in real-time better than a full-size computer can do with unlimited time, then the software is braindamaged. However, both should do an equally excellent job given that up-conversion has been a solved problem for a long time.

If the answer isn't "they are the same", then one of them is poorly done.

Fair point, Daniel. However, since there is no fixed formula for this conversion process, certain manufacturers may employ different algorithms which will inevitably affect the final quality.

...
However, HDMI is not data transfer. It is an actual video signal and in order for the camera to play out your HDV tape as a video signal, it has to up res the footage to 1920x1080 at 4:2:2. This is the only way it can transmit that information to the Intensity card or any HDMI capable display. This is done regardless of whether you send video out via HDMI or the analog component connection.

When you are capturing 'live', the camera bypasses the HDV compressor chip and sends the signal out via HDMI.

When you are capturing off a HDV tape, the camera decompresses the data on tape, filters and resamples the data before sending it out as 1920x1080 4:2:2 via HDMI...Kristian,

Do you know how audio is handled?

When capturing live via HDMI, what sampling and bit rates come out?

And when HDMI capturing from an HDV tape, the MP2 audio must be decompressed. Is it up-resed to the same sampling and bit rates that live HDMI capture stream out?

Thanks much!

Kristian,

Do you know how audio is handled?

When capturing live via HDMI, what sampling and bit rates come out?

And when HDMI capturing from an HDV tape, the MP2 audio must be decompressed. Is it up-resed to the same sampling and bit rates that live HDMI capture stream out?

Thanks much!

Hi Peter,

The audio in the HDMI stream is 48 kHz and 24 bit so it it is coming off tape, it will get resampled if necessary.