Convergent Designs Flash XDR

[Mike Schell]

Originally Posted by Barlow Elton

Speaking of bit rates...is there any chance, given that the XDR can remove pulldown from the H1's 24F 29.97 1080i HD-SDI output, the 50 mbs mode could be applied to the 24 progressive frames *only*, just like Canon's 24F HDV mode works? This would add 20% more bandwidth per frame and I think might be the actual sweet spot for most shooters.

If the 50 mbs mode looks fine and is truly motion artifact resistant (given the higher bit rate plus VBR and progressive encoding), this would probably be the mode I would use most. If the quality is nearly indistinguishable from the 160 mbs mode, that'd be good enough for me. 50 mbs to 16 GB CF cards would mean about 32 minutes of footage per card, (as opposed to just over 10 min. in 160 mbs) if I'm not mistaken. That would mean signifigantly less card swapping throughout a day's shoot.

Audio question--Is it possible to extract the H1's audio via firewire, even if it's post-compression? I think this might be a better solution than trying to accomodate the analog RCA ports on the camera.

Yes, the 50Mbps rate can be applied to the 24p mode. Yes, I do think the 50Mbps 4:2:2 rate will look excellent. You will be able to select Long-GOP or I-Frame only. Long-GOP will likely look better, but I-Frame would be easier to edit.

My calculations show 42 minutes per 16Gbyte card at the 50Mbps rate.

Mike Schell

[David Heath]

Originally Posted by Barlow Elton

50 mbs to 16 GB CF cards would mean about 32 minutes of footage per card, (as opposed to just over 10 min. in 160 mbs) if I'm not mistaken. That would mean signifigantly less card swapping throughout a day's shoot.

AFAIK, at the moment, the largest high performance CF card (eg Extreme IV) capable of recording 160Mbs is 8GB, not 16, so at that data rate you are currently restricted to about 5 minutes/card.

Which makes the 100Mbs long-GOP option doubly intriguing. Should be extremely high quality, and being 100Mbs, it allows the use of Extreme III, so cheaper anyway, as well as being available now in 16GB sizes. That means 16 minutes per card now, with 32GB cards forecast soon.

Mike - I think my confusion is that I'm thinking of PAG batteries, with V-lock connector one side, but plain on the other. What you say makes sense if I assume such as IDX with male connector one side, female the other.

[Michael Galvan]

Yes, Barlow makes an interesting point here. Extracting audio from the firewire port is an interesting paossibility, although it would be post compression.

Your thoughts Mike?

Originally Posted by Barlow Elton

Speaking of bit rates...is there any chance, given that the XDR can remove pulldown from the H1's 24F 29.97 1080i HD-SDI output, the 50 mbs mode could be applied to the 24 progressive frames *only*, just like Canon's 24F HDV mode works? This would add 20% more bandwidth per frame and I think might be the actual sweet spot for most shooters.

If the 50 mbs mode looks fine and is truly motion artifact resistant (given the higher bit rate plus VBR and progressive encoding), this would probably be the mode I would use most. If the quality is nearly indistinguishable from the 160 mbs mode, that'd be good enough for me. 50 mbs to 16 GB CF cards would mean about 32 minutes of footage per card, (as opposed to just over 10 min. in 160 mbs) if I'm not mistaken. That would mean signifigantly less card swapping throughout a day's shoot.

Audio question--Is it possible to extract the H1's audio via firewire, even if it's post-compression? I think this might be a better solution than trying to accomodate the analog RCA ports on the camera.

[Mike Schell]

Originally Posted by David Heath

AFAIK, at the moment, the largest high performance CF card (eg Extreme IV) capable of recording 160Mbs is 8GB, not 16, so at that data rate you are currently restricted to about 5 minutes/card.

Which makes the 100Mbs long-GOP option doubly intriguing. Should be extremely high quality, and being 100Mbs, it allows the use of Extreme III, so cheaper anyway, as well as being available now in 16GB sizes. That means 16 minutes per card now, with 32GB cards forecast soon.

Mike - I think my confusion is that I'm thinking of PAG batteries, with V-lock connector one side, but plain on the other. What you say makes sense if I assume such as IDX with male connector one side, female the other.

David-
I agree, all things considered, the 100Mbps long-GOP may be the overall best choice. Certainly the image quality will be outstanding - 4X the bit-rate of HDV with 4:2:2 color space and full-raster (1920x1080). Using the 16G CF will minimize card swaps and save some cost.

FYI, we are working with Anton Bauer, IDX and PAG on various battery combinations. IDX looks the simplest, but I think we can work out suitable mounts for the the other batteries.

Mike Schell

[Mike Schell]

Originally Posted by Michael Galvan

Yes, Barlow makes an interesting point here. Extracting audio from the firewire port is an interesting paossibility, although it would be post compression.

Your thoughts Mike?

Well, we have actually considered this same approach. It offers a lot of advantages. We will discuss the design with Canon engineers, as we need to align the compressed audio out the 1394 port with the HD-SDI signal. We'll also need to decompress the audio before writing to CF (at least for the QT file format).

Fundamentally, I think this approach will work.

Mike

[Dan Keaton]

Dear Mike,

Before developing a method to extract off the audio via the firewire port, I would want to know the signal path of both options.

I assume that the audio line outputs from a camera, such as an XL H1, would be full analog signals, If so, then you could capture the sound in whatever format you liked, such as 16/48k, 24/48k, or any other flavor.

If the above is true, then the audio would be much better than compressed audio from firewire. Excellent audio would be a great feature, to match the overall purpose of your unit, which is to improve upon the standards available in camera (which may be HDV or some other standard).

[Tim Kolb]

Originally Posted by Dan Keaton

Dear Mike,

Before developing a method to extract off the audio via the firewire port, I would want to know the signal path of both options.

I assume that the audio line outputs from a camera, such as an XL H1, would be full analog signals, If so, then you could capture the sound in whatever format you liked, such as 16/48k, 24/48k, or any other flavor.

If the above is true, then the audio would be much better than compressed audio from firewire. Excellent audio would be a great feature, to match the overall purpose of your unit, which is to improve upon the standards available in camera (which may be HDV or some other standard).

I suspect they're aware of this caveat, but I would agree that it's significant.

I'm going to guess (and Mike can certainly correct me if I'm wrong) the rub is that re-clocking analog audio to the HDSDI video signal, while the HDSDI signal is being digitally transcoded and the analog audio is going through a quantization step would be trickier than bringing in another digital signal.

The fact that Canon didn't include embedded audio on the XLH1 is really unfortunate. They have included it now on the smaller HDSDI-capable HDV camera, but I suspect their original intention was to streamline monitoring with that jack and little else at the time...

[Nikol Manning]

Just so we are clear the XDR will record what datarates to compact flash in what formats?

Also what gave you folks the idea to make this product. Keep up the good work.

[Mike Schell]

Originally Posted by Nikol Manning

Just so we are clear the XDR will record what datarates to compact flash in what formats?

Also what gave you folks the idea to make this product. Keep up the good work.

Flash XDR can record (and playback) using the 422P@HL Profile which is 4:2:2 MPEG2 at full raster (1920 x 1080, 1280x720) at the following max data-rates:

100 Mbps in Long-GOP
160 Mbps in I-Frame Only

You can use the lower cost SanDisk Extreme III cards (133X) for 100Mbps (and lower), but will need the Extreme IV Compact Flash (266X speed) for the 160Mbps rate. (Note CF speeds are based on multiples of 150K bytes/sec, the original CD playback rate. So, 133X = 20 MBps or 160 Mbps).

Regarding our idea for this product, after the experience of our HDMI to HD-SDI converter (nanoConnect) we realized that the quality of the HD-SDI output from most cameras was far superior to the recording mechanism, even for HDCAM. Also, we saw a trend that file-based workflow, would replace tape-based ingest. So that gave us the motivation to build an external add-on box which would increase the video quality and enable fast transfer of video footage to the NLE.

Compact Flash speeds and capacities have increased dramtically and offer relatively low costs, high reliability and low-power. So, we realized that by using CF memory and the high-quality Sony MPEG2 CODEC we could build a small portable HD recorder that could mount on the back of any HD-SDI based camera.

In a nutshell, that's how we decided to build Flash XDR. (No, we didn't come up with the idea after visiting a coffee shop in Amsterdam).

Mike Schell

[Barlow Elton]

Originally Posted by Tim Kolb

I'm going to guess (and Mike can certainly correct me if I'm wrong) the rub is that re-clocking analog audio to the HDSDI video signal, while the HDSDI signal is being digitally transcoded and the analog audio is going through a quantization step would be trickier than bringing in another digital signal.

Interesting take. I think you're also correct about Canon's original intentions with the H1's HD-SDI output. As inconvenient as it is to not have embedded audio and timecode like the G1 or JVC 250, IMO it's still worth jumping through a few extra hoops with regards to audio in order to be able to exploit the uncompressed video output.

HDV audio: I've never had a problem with it. It of course depends so much more on the quality of the pre-amps, converters etc, onboard the camera than the fact that it's moderatley compressed by HDV. I've even rescued some recordings that required heavy background noise reduction processing in Soundtrack Pro, and got very acceptable results. The compressed MPEG2 audio was a non-factor.

[David Heath]

Originally Posted by Mike Schell

Flash XDR can record (and playback) using the 422P@HL Profile which is 4:2:2 MPEG2 at full raster (1920 x 1080, 1280x720) at the following max data-rates:

100 Mbps in Long-GOP
160 Mbps in I-Frame Only

What about 50Mbs, as used in the PDW 700?

(Though again, I see you do say "max data rates" - is it therefore continuously selectable from 50-100Mbs in this profile, or just 50 and 100?

[Mike Schell]

Originally Posted by Barlow Elton

Interesting take. I think you're also correct about Canon's original intentions with the H1's HD-SDI output. As inconvenient as it is to not have embedded audio and timecode like the G1 or JVC 250, IMO it's still worth jumping through a few extra hoops with regards to audio in order to be able to exploit the uncompressed video output.

HDV audio: I've never had a problem with it. It of course depends so much more on the quality of the pre-amps, converters etc, onboard the camera than the fact that it's moderatley compressed by HDV. I've even rescued some recordings that required heavy background noise reduction processing in Soundtrack Pro, and got very acceptable results. The compressed MPEG2 audio was a non-factor.

Regarding the missing embedded audio on the XL H1, I strongly suspect that Canon simply did not have enough time to finish the code development to meet the product launch date. HD-SDI embedded audio is actually quite difficult to develop (about 5X the comlexity of SD-SDI embedded audio), so Canon was probably forced to forego this feature to meet delivery schedule.

We are continuing to study the possible options to capture the audio from the XL H1. The analog approach would eliminate the MPEG1 Layer 2 compression, but could introduce other issues (A/D and D/A conversion losses) depending on the exact implementation (ie do we take the unbalanced audio out of the camera or create a complete audio subsytem with microphone inputs). Capturing the audio out of the HDV stream also requires some engineering work to decompress the audio and sync with the HD-SDI stream. We'll have to way off the pros and cons of each approach.

Mike Schell

[Mike Schell]

Originally Posted by David Heath

What about 50Mbs, as used in the PDW 700?

(Though again, I see you do say "max data rates" - is it therefore continuously selectable from 50-100Mbs in this profile, or just 50 and 100?

We will probably support a couple of different rates. While the MPEG2 CODEC can technically be set to a wide range of values, we want to limit the available number of data-rates so these can be properly tested and verified.

I can say that we will have presets for both 50 and 100 Mbps 4:2:2 full-raster Long-GOP. We'll set the I-Frame rates in the near future (160Mbps will be one of the presets).

Mike Schell

[David Heath]

It all sounds good! But does bring one further question to my mind - will these be compatible with XDCAM-HD files (at least the 50Mbs version) as far as an NLE is concerned? I believe those are wrapped as MXF files - will your product do the same?

[Mike Schell]

Originally Posted by David Heath

It all sounds good! But does bring one further question to my mind - will these be compatible with XDCAM-HD files (at least the 50Mbs version) as far as an NLE is concerned? I believe those are wrapped as MXF files - will your product do the same?

MXF will be the first format we support. We plan to implement the OP-1A method where video and audio are interleaved into one file, just like XDCAM-HD. So, you should be able to import these files into most NLEs (may require the aid of some re-wrapping software).

Mike Schell