Well, we have some of the long awaited full motion video comparisons ready for your download and analysis. First I need to describe how these clips were created. These are blue-screen shots using the HD-SDI (1080i59.94) output from a Sony EX1 camera. They were first captured into a FCP system 8-bit uncompressed. This clip was placed onto the timeline and output as a uncompressed Quicktime movie (file = Blue Screen uncompressed).

Then we output the uncompressed stream out HD-SDI (from the MAC) and captured into the Flash XDR at 100Mbps 4:2:2 Long-GOP full-raster (1920x1080) onto a Compact Flash card. We then played back the clip from the XDR and recaptured as uncompressed 8-bit into FCP and generated a QT movie (file = Blue Screen 100 Mbps).

So the only difference in the two files is the 100 Mbps compression from the Flash XDR box.

We also grabbed a single frame and did an overlay in Photoshop and set one layer to difference. You will see the difference as a black image on the right side of this BMP file (file name = UNCvsMPEG2_100Mbit). You'll need to turn off the lights and close the shades, because the differences (shown as white dots) are really difficult to find.

Here's the download links:

http://convergent-design.fileburst.com/Blue Screen Uncompressed.zip (Blue Screen Uncompressed QT .mov 128 MB)

http://convergent-design.fileburst.com/Blue Screen 100 Mbps.zip (Blue Screen 100 Mbps QT .mov 128 MB)

http://convergent-design.fileburst.com/UNCvsMPEG2_100Mbit.zip (Blue Screen Comparison, Single Frame + Photoshop Difference, 8MB)

Special Thanks to Jim Arthurs of Image Shoppe (Colo Spgs, CO) for providing the footage and allowing us to borrow his Sony EX1.

We anticipate more requests for all sorts of different footage, but we need to concentrate on finishing the Flash XDR and nanoFlash now. We believe that these movies should lay to rest most concerns regarding the quality of the Long-GOP compression in areas of high motion. Kudos to Sony for an outstanding design on their MPEG2 CODEC.

Please feel free to do some compositing and keying with these clips.

Thanks Mike.

I look forward to seeing them.

One request or constructive critique, sorry!.

In future tests, I would suggest using a progressive mode as a lot of people will be looking at frame grabs and interlaced just does not function well for screen grabs.

It might make comparisons a little cleaner imho.

Please disregard if you used interlaced on purpose.

Also, a comparison of the camera's native recording bitrate (35mb for the EX) to see what we as the consumers are gaining by using the XDR over the original cameras output would be great.

Thanks!

I downloaded the 100Mbps .mov and slowed it down in FC 6.0.3 to 10% then dropped the chroma key filter on and it keyed very nice. I can imagine what better keying solutions will do. Thanks Mike for the call today and I look forward to more test footage.


Michael Palmer

Hi all... just a note that the footage actually is 23.976 progressive stored in an interlaced stream with 3:2 to pad it out to 29.97/59.94. In your compositing package, simply choose to interpret or remove pulldown to regain the original progressive source.

If anyone has specific questions about the original footage, I'm glad to answer them. Like you all, I look forward to testing the 100Mb/sec in comparison to the original footage!

Regards,

Jim Arthurs

Hi Jim and thanks for your part in getting these clips out on the board here. My questions are related to lighting, how far away from the blue back drop was the subject? Can you tell us what you used and where the lights were. It looks as there was one stronger half light from the camera left about seven feet high or so and a much higher up half light from the right and my guess is some fill near camera. Oh and did you light the blue separately?

After playing in Motion 3 I found it even easier to key the 100 Mbps .mov.

thanks again
Michael Palmer

My questions are related to lighting, how far away from the blue back drop was the subject? Can you tell us what you used and where the lights were. It looks as there was one stronger half light from the camera left about seven feet high or so and a much higher up half light from the right and my guess is some fill near camera. Oh and did you light the blue separately?

Hi Michael, the footage is shot under a mixture of daylight balanced flo's and full CTB gelled tungsten units, the talent was about 8' in front of the cyc, in the same lighting setup as the background. I had two trusty 4 tube biax 55 case lights for the main key and for bounce fill, while the 1K tungsten with CTB gave some directional light from the right.

I just downloaded the 100 Mb/sec clip, which was in the third party BlackMagic 8bit codec and the colors are out of whack on the PC... my recommendation is to replace that existing file with a plain Jane native QT for better compatibility from Mac to PC... then more people can play :)

Regards,

Jim A.

I just downloaded the 100 Mb/sec clip, which was in the third party BlackMagic 8bit codec and the colors are out of whack on the PC... my recommendation is to replace that existing file with a plain Jane native QT for better compatibility from Mac to PC... then more people can play :)

For what it's worth, it looks correct when viewed inside After Effects and Digital Fusion, just wonky when viewed straight from the QT player on windows... shrugs... of course you'll need the free codec download from Black Magic Design for this codec...

Jim A.

mike,

can't we cat the real life samples. and also comparison with hdv, xdcam codecs with H1 and sony ? I really want to see what we get over these (hdv, xdcam) codecs, and as you stated in sony codec comparison chart, did we get high end results like sony hdcam sr practically ?

thanks,

alkim.

mike,

can't we cat the real life samples. and also comparison with hdv, xdcam codecs with H1 and sony ? I really want to see what we get over these (hdv, xdcam) codecs, and as you stated in sony codec comparison chart, did we get high end results like sony hdcam sr practically ?

thanks,

alkim.

Hi Alkim-
When we have more time we can do a lot more video comparisons. In the short term, we have to concentrate our efforts on finishing the Flash XDR and nanoFlash. With that said, I will be doing a series of still image resolution chart tests today using the Sony EX1, comparing uncompressed to 100 Mbps, DVCProHD and HDV. I think the results will speak for themselves.

I also think we need to give a great deal of credence to the internal tests conducted by Sony. This chart, created by Sony, compares the overall video quality of the MPEG2 Long-GOP CODEC to HDCAM and HDCAM SR, all of which are Sony CODECs. Remember, the Sony MPEG2 4:2:2 CODEC module is the heart of our Flash XDR and nanoFlash products. It ultimately determines our quality. We're using the same CODEC as the PDW-700, we have just increased the bit-rate and write the datastream to low-cost CompactFlash cards.

As I mentioned in other posts, this is a world-class 6th or 7th generation MPEG2 CODEC with all the advanced features such as full-raster processing (1920x1080), 4:2:2 color, variable bit-rate coding and 100 Mbps Long-GOP processing. So according to this chart, the 100 Mbps 4:2:2 Long-GOP is above HDCAM quality and just below HDCAM SR.

The real kicker is that you get this quality level in a portable recorder for under US $5K (under US $3.5K for nanoFlash) and can record 35 minutes of 100 Mbps footage on low-cost (US $135) CompactFlash cards.

hi mike,

this means flsah xdr resize the 1440x1080 image of xlh1 and record as 1920x1080 full raster at 422 color ?

thanks,

alkim.

hi mike,

this means flsah xdr resize the 1440x1080 image of xlh1 and record as 1920x1080 full raster at 422 color ?

thanks,

alkim.

Hi Alkim-
Actually the XL-H1 does the resizing from 1440 to 1920 before sending out the HD-SDI. 1080i/p HD-SDI is, by definition, always 1920x1080 4:2:2 10-bit. It's up to the transmitting device (the camera in this case) to make the necessary changes to the video to meet this requirement (resizing, 4:2:0 to 4:2:2 conversion, etc).

Mike,

This may have been addressed, so forgive me if I am redundant. I know there are issues with incremental under/over crank but will the XDR/Nano support good old 720/60p from a Sony EX3?

Mike,

This may have been addressed, so forgive me if I am redundant. I know there are issues with incremental under/over crank but will the XDR/Nano support good old 720/60p from a Sony EX3?

Hi Jerry-
Absolutely, 720p60 and 720p50 are definitely supported.

Yes, we currently do not support overcrank or overcrank as these are "features" that occur downstream from the HD-SDI output. We do not know if these capabilities can be implemented in Flash XDR / nanoFlash.

Hi Alkim-
Actually the XL-H1 does the resizing from 1440 to 1920 before sending out the HD-SDI. 1080i/p HD-SDI is, by definition, always 1920x1080 4:2:2 10-bit. It's up to the transmitting device (the camera in this case) to make the necessary changes to the video to meet this requirement (resizing, 4:2:0 to 4:2:2 conversion, etc).

Hi Mike, I am very close to sending the deposit for the Flash XDR - just want to make sure what is the situation with the Sony HVR Z7 (PAL) and the Flash XDR as the Z7 doesn't have the HD-SDI. I know I can get an adapter to convert the HDMI of the Z7 to HD-SDI from you. With this adapter - will I get 1920x1080, 4:2:2 10 Bit and the rest of the upgrading in signal quality? Will I get exactly the same quality as using a camera with the HD-SDI ?
I don't want to be in a position of ordering the device just to realise I can't really get its full potential with my HDV Sony HVR Z7E.

Thanks,

Ofer Levy
Sydney

Hi Mike, I am very close to sending the deposit for the Flash XDR - just want to make sure what is the situation with the Sony HVR Z7 (PAL) and the Flash XDR as the Z7 doesn't have the HD-SDI. I know I can get an adapter to convert the HDMI of the Z7 to HD-SDI from you. With this adapter - will I get 1920x1080, 4:2:2 10 Bit and the rest of the upgrading in signal quality? Will I get exactly the same quality as using a camera with the HD-SDI ?
I don't want to be in a position of ordering the device just to realise I can't really get its full potential with my HDV Sony HVR Z7E.

Thanks,

Ofer Levy
Sydney

Hi Ofer-
You will definitely get the full 1920x1080 4:2:2 via the conversion from HDMI to HD-SDI (which is a totally lossless step). But, just to be clear this is an 8-bit path not 10-bit. I have yet to see a 10-bit signal out the HDMI port from any camera.

Otherwise, you should get beautiful video quality. We actually use a Canon HV20 connected to a nanoConnect as our video feed for most of our Flash XDR tests in the lab. The video quality is astounding from this little $800 camera, especially when we crank up the bit-rate to 50/100 Mbps level.

Hi Ofer-
You will definitely get the full 1920x1080 4:2:2 via the conversion from HDMI to HD-SDI (which is a totally lossless step). But, just to be clear this is an 8-bit path not 10-bit. I have yet to see a 10-bit signal out the HDMI port from any camera.

Otherwise, you should get beautiful video quality. We actually use a Canon HV20 connected to a nanoConnect as our video feed for most of our Flash XDR tests in the lab. The video quality is astounding from this little $800 camera, especially when we crank up the bit-rate to 50/100 Mbps level.

Thanks for your reply Mike !
I am going to place the pre-order and the deposit in a few days. Can't wait to put my hands on what looks like wonderful device !
Cheers,
Ofer Levy
Sydney - Australia
http://www.oferlevyphotography.com

With this adapter - will I get 1920x1080, 4:2:2 10 Bit and the rest of the upgrading in signal quality?
It's worth noting that whilst the EX1 chips are about 2 Megapixel each(1920x1080) those of the Z7 are half that - about 1 Megapixel - hence the camera won't (can't) give full 1920x1080 resolution, even if it outputs a 1920x1080 signal. I wouldn't expect to see a resolution improvement when using the XDR over the cameras native abilities.

What XDR recording WILL undoubtably give is improved colour space, and much lower compression.

I see it being very worthwhile (with a Z7) for those factors alone.

It's worth noting that whilst the EX1 chips are about 2 Megapixel each(1920x1080) those of the Z7 are half that - about 1 Megapixel - hence the camera won't (can't) give full 1920x1080 resolution, even if it outputs a 1920x1080 signal. I wouldn't expect to see a resolution improvement when using the XDR over the cameras native abilities.

What XDR recording WILL undoubtably give is improved colour space, and much lower compression.

I see it being very worthwhile (with a Z7) for those factors alone.

Hi David-
Thanks for this additional clarification. You are absolutely correct, you will not see any improvement in the horizontal resolution as the camera will upsample from 1440 to 1920 before sending out the HDMI port.

It is worth noting that we do see this improved picture quality and color space when capturing video from the Canon XL-H1 which is also limited to 1440x1080 sensors. The 50/100 Mbps 4:2:2 video is head and shoulders above the native HDV.

It is worth noting that we do see this improved picture quality and color space when capturing video from the Canon XL-H1 which is also limited to 1440x1080 sensors. The 50/100 Mbps 4:2:2 video is head and shoulders above the native HDV.That's what I like to hear Mike! My biggest headache is how to get rid of the bad degration in motion that I get when using the HDV-compression. Hopefully this unit will be in a reliable quality for tough wildlife use!?

That's what I like to hear Mike! My biggest headache is how to get rid of the bad degration in motion that I get when using the HDV-compression. Hopefully this unit will be in a reliable quality for tough wildlife use!?

Hi Per Johan-
I have been absolutely amazed at the quality from this Sony CODEC. This is their 7th generation part and they truly did a stunning job designing this part. It has very low-power (around 3W in 4:2:2 mode), small size and superb video quality. It's the same CODEC as inside the PDW-700 camera, we just cranked up the bit-rate from 50 to 100 Mbps.

It's the same CODEC as inside the PDW-700 camera, we just cranked up the bit-rate from 50 to 100 Mbps.

Folks, one thing that's very exciting about the Flash XDR and its upcoming little brother is that Convergent Design is basically leap-frogging the slow incremental progression of Sony and giving us the "end game" of what these MPEG chips can do.

Think about it... Sony, in the PDW-700, is basically selling the smallest fraction upgrade from their 35 Mb/sec current standard that they can get away with at 50 Mb/sec.

How many generations of new cameras will Sony take to give us the full 100 Mb/sec or 160 Mb/sec I frame only that the Flash XDR instantly delivers? My bet is that they try to pass off a new 75 Mb/sec camera before unleasing 100 Mb/sec. Their history with camera bandwidth upgrades supports this (25 Mb/sec to 35 Mb/sec and now 50Mb/sec)

With the XDR/nano you can basically see into Sony's future at least two generations and take advantage of that potential today with the EX1/EX3.

It's not just the prosumer cameras that can profit. Yesterday I was chatting with a fellow who works with an F900 CineAlta with the 4:2:2 back option. Running that output into the XDR is better quality by far than the native recorded 3:1:1 HDCam signal on tape. Previously, the only way to utilize that full 4:2:2 1920 by 1080 from his camera was either to digitize into a land-locked NLE system, the AJA IO box converting to ProRez, or using one channel of the HDCam SR deck... nanoFlash will be a far less expensive option than any of those previous choices.

Interesting times...

Regards,

Jim Arthurs

Folks, one thing that's very exciting about the Flash XDR and its upcoming little brother is that Convergent Design is basically leap-frogging the slow incremental progression of Sony and giving us the "end game" of what these MPEG chips can do.

.............How many generations of new cameras will Sony take to give us the full 100 Mb/sec or 160 Mb/sec I frame only that the Flash XDR instantly delivers? My bet is that they try to pass off a new 75 Mb/sec camera before unleasing 100 Mb/sec.
You raise a very interesting and valid point, Jim. If the chips can deliver 100, 160Mbs MPEG2, why aren't Sony enabling it themselves?

I think it's worth remembering that the PDW700 is primarily an optical disc camera, though an addon SxS unit was announced for it. As such, the data rate is limited by the write speed to the disc, rather than the coder chip, so any implementation of higher bitrate mode would have to be to SxS only.

Sony have stated for a long time that they believe solid state will become more important as time goes by, but the consumable nature of XDCAM disc is more appealing to many users at the moment. When the solid state/disc scales finally come down in overall favour of the former, that's when I think we'll see a "solid state PDW700" and likely with the option of higher than 50Mbs bitrates.

In the meantime, the 50Mbs mode seems to be getting some pretty good reports, and broadcasters seem to be approving it for unrestricted HD acquisition.

You raise a very interesting and valid point, Jim. If the chips can deliver 100, 160Mbs MPEG2, why aren't Sony enabling it themselves?

I think it's worth remembering that the PDW700 is primarily an optical disc camera, though an addon SxS unit was announced for it. As such, the data rate is limited by the write speed to the disc, rather than the coder chip, so any implementation of higher bitrate mode would have to be to SxS only.

Hi David, just jumping around all over Sony's product line, and I don't mean to flame them specifically for this. All the big guys milk existing tech and provide features at a slower than possible rate accessed by new "B" models or basically new models that are just firmware upgrades and/or a few extra user buttons. Think about how much mileage Panasonic is getting out of pixel shifting SD chips and limiting access to their top of the line full raster codec, etc.

I'm just excited to have access to the maximum that the latest generation of mpeg chips can do, and have it today with my existing EX1!

Reframing my point to just the top of the line prosumer models, the EX1/EX3. and what will follow.. I imagine the next model introduced at NAB will provide 50 Mb/sec 4:2:2 data rate (basically the same level of compression as the 35Mb/sec EX1, just at 4:2:2 instead of 4:2:0), but I'd bet a good chunk of money that it won't be higher than that even though the chips could do it. I think we're two iterations out from the EX1 before we see Sony take full advantage of their already existing option of 100 Mb/sec in the top of the line prosumer camera.

Of course the "head end" sensor and lens package will be more advanced on this "future" cameras, but I'm just focusing on the recorded data rate and compression because that's the area that Convergent Design is addressing...

Just my 2 cent's worth...

Jim Arthurs

50 mbits/s is not a bad format at all. In fact I could see a lot of people using that format even on the XDR. One of the main reasons why using 50 mbits/s is important is record time. You can get a lot more video onto a card which means less card swapping and less money invested in cards. 50 mbits/s 4:2:2 is a very capable format and 100 mbits/s is only a few percentage points better in quality. SONY picked 50 mbits/s because it was the perfect middle ground between high quality and storage space. The way mpeg2 works the higher you go in bitrate the less important it really is. 40mbits/s to 50 mbits/s might be a noticeable jump in quality but going from 50 mbits to 60 mbits wouldn't be as noticeable. The higher you get the less you will actually notice a change in quality. 100 mbits/s is great because it is virtually perfect but for 95% of shooting situations it would be total overkill. People want to be able to record lots of video per dollar and one of Sony's claims to fame is that you can get better then DVCPROHD quality at half the bitrate which equals much more record time. If Sony did use 100 mbits then they would no longer have that advantage over Panasonic. Trust me you will find 50 mbits/s to be a very solid format.

Dear Thomas,

I agree completely.

Our testing has shown the 4:2:2 50 Mb format to be very good and very close to 4:2:2 100 Mb.

I also think that for many applications the 50 Mb format will be widely used for the reasons you mentioned.

For the most demanding of shoots, I would consider the 100 Mb or the full uncompressed in very special cases. It is much easier to work with 50 Mb or 100 Mb data, as you are well aware, but for high dollar productions, some will want full uncompressed.

Personnally, I feel that it will be very interesing to see how many adopt the 50 Mb and 100 MB options over the full uncompressed.

I wonder if some full uncompressed users will test the 50 Mb and 100 and find these options very desirable. Time will tell.

I like the fact that we can now choose the bit rate, in one piece of gear, for the task at hand. Full uncompressed may be just what is needed for one shoot, but the next may be better suited by 50 Mb.

Nothing wrong with 50 Mb/sec, but the compression (macroblocking) is visually the same as the 35 Mb/sec flavor, that extra 15 Mb/sec just gives you the advantage of the better 4:2:2 color space. Which gives you an overall better picture, no argument there.

However, I would disagree about the marginal gains going from 50 Mb/sec to 100 Mb/sec. When you kick it up to 100 Mb/sec the macroblocking all but vanishes compared to 50 Mb/sec . The 160 Mb/sec I frame version has about the same compression quality as 100 Mb/sec, the value for your data rate is that each frame is unique.

I tend to always look at this stuff from the end game of "how good will it key?". My apologies, as I'm always coming from that point of view and there are always equally valid concerns for different scenerios.

The great thing is that these products will give you a choice, based on your scene requirements; day to day shooting at 50 Mb/sec, 100 Mb/sec long GOP for critical work and 160 Mb/sec if you're concerned about motion artifacts on a particular subject. And, down the road, uncompressed on the XDR for those situations when none of the above will work for you.

Hi David, just jumping around all over Sony's product line, and I don't mean to flame them specifically for this. All the big guys milk existing tech and provide features at a slower than possible rate ..........
I didn't read it as a flame, Jim, but just wanted to remind everybody that there is a technical reason for not going above 50Mbs on the 700 - the speed of optical disc. My take is that Sony think that consumable media is more important in this market sector at the moment than absolute quality (and they will sell you top end HDCAM SR, of course) and I'm inclined to think they're right.

Of course, there's no reason (??) why higher than 35Mbs couldn't be provided on the EX series, and I guess that is where marketing does come in - they couldn't be seen to have any higher spec than more upmarket disc models.

I fully agree with all that has been said about bitrates, though the move to full raster recording (1920x1080) I think as important as 4:2:2. 50Mbs is likely to be more than adequate the vast majority of the time, but as Jim says, the great thing is to have the choice.

I certainly agree that the 50 Mbps 4:2:2 looks great and I can say that the majority of our customers who plan to use XDR / nano in TV production work are planning to use this data-rate. But we still do see some improvement as you go up to 100 Mbps, but is is substantially less than the jump from 35 to 50.

Keep in ming that this is a constantly changing equation as the price of solid-state NAND Flash memory will continue to drop. I get a bimonthly report on DRAM and Flash memory chip prices. With rare exception, the prices are always down 1-2% over the previous report. I also know that Samsung is in full production on their 64 Gb Flash chips, which means that 64GB CF card can't be too far behind.

Looking out further, NAND Flash density and a prices are expected to make further drops with the advent of SSD (Solid State Drives) and ever increasing sales of iPhones, iPods, and cell phones in general. The good news is that we will ride this wave with low-cost Compact Flash memory, already 1/10 the price of other solid-state media.

So, in the not too distant future, we may not worry about bit-rate issues for acquisition as the memory capacity and cost will be so low, that we will always shoot at the highest compressed data-rate.

I also know that Samsung is in full production on their 64 Gb Flash chips, which means that 64GB CF card can't be too far behind.
Different people have different needs, but what some are waiting for are cheap, smaller, single cards, able to contain a reasonable amount of video. I'd say 30 minutes was a reasonable figure.

At 50Mbs that equates to 16GB cards, and I agree that in the Compact Flash form they should be comparable in cost to a 30 minute tape in not too many years. When that happens, solid state won't mean harddrives, laptops, and a transfer process at the end of a shoot - the cards will just be given to client in the way tapes are today.

Dear David,

I just did a rough calculation.

1080i60 at 50Mb yields a little over 30 minutes of footage using one 16GB CompactFlash card. (I calculate 35 minutes, but I am trying to be conservative.)

A Transcend 133x 16GB CompactFlash card is now available, including shipping for $72.00.

A 8 GB card costs $34.00 (over 15 minutes of footage).

It will be interesting how this plays out in a year or two.