Emmy award winner Richard Welnowski, started his evaluation of the Flash XDR yesterday. Over 380 shooting days with the Thomson Viper has made him a true expert with what many consider one of the finest digital motion picture cameras in the world.

Richard normally records full uncompressed using his Viper setup but was looking for a way to record very high quality images while achieving a much more manageable file size.

Richard is a very accomplished and well-respected cameraman.

After a brief setup of the Viper to work with the Flash XDR and a few minutes of learning how to use the Flash XDR, he started recording.

This was a spontaneous test, with no special lighting. The image is of a bicycle Richard designed. It was in his loft, lit by a small window. The Bicycle was 20' from his viper, and Richard was using a Canon Zoon lens, fully open, with +3db of gain in the Viper, which was set for 4:2:2 mode.

The Flash XDR was set to record at 100 Mbps, Long-GOP, mode. We call this our "Master Quality" mode, and we recommend it over all other modes when quality is priority one.

Richard recorded for a short time, then transferred the image into Final Cut Pro's latest version. The image below is a single frame. Richard recorded at 1080psf29.97.

The following is an untouched image; no color correction or adjustment of any kind was made (as I understand it, I was not there.) This is a 5 MB download.

http://convergent-design.fileburst.com/Thomson Viper/Bike-Viper-422.tiff

Recording full uncompressed is demanding in terms of cost of storage and processing power. The Flash XDR's 100 Mbps files are much easier to transfer and store while still providing excellent quality images.

Richard was evaluating the Flash XDR to see how our images compare.

He was very excited with the results and has given me permission to quote him.

"It's beautiful!"
"This is amazing, just amazing."
"I am really pleased".
"There are no artifacts!"

Richard checked each of the channels and found them clean. He was also very pleased how easy it was to upload the footage into Final Cut Pro.

We share Richard's excitement and thank him for his evaluation.

Richard will be filming the Houston Marathon tomorrow.
His website is: Compositing Expert (http://www.compositingexpert.com)

We welcome your comments.

A couple of years ago, Fincher concocted a strategic plan to R&D a tapeless workflow built around Thomson Viper FilmStream cameras recording to D.Mag Digital Film Magazines (from S.two of Reno, Nev.) on a series of five commercials. He has now applied lessons from those projects to a full-length feature film for the first time with Zodiac back in 2006.

Apart from S.2 provides an uncompressed Dual link 4:4:4 1080p up to 4K capturing solution, and I am strongly believed that Flash XDR can make an " Super Value " to film Directors as thier HD master viewing in HD Broadcast grade master format.

Check out the Zodiac's tapleless workflow at http://digitalcontentproducer.com/images/608MIL_zodiac_viperonsetacquisitionworkflow.pdf

for our reference, and we will have some ideas how we can use Flash XDRs to apply in the High-end Cameras based field recording workflow.

We can use XDRs ( 1920 x 1080 8 bit 4:2:2 at 50 -160 Mb/s ) as a better option than " HDCAM " quality ( 1440 x 1080 then applies to just 8 bit 3:1:1 ) for the well known Big boys cameras like Genesis / D20/21 / F35 & F23 / Viper / Varicam to gain HD broadcast market revenues.

And this is my one Cent.....

Best regards,
Stewart Chong
An Alchemist with the FlashXDR
Tapeless Workflow from HD to 4K ( Creative to Display )
Wow Holdings
Hong Kong

2009 / New Year's intention completely demolished on day 17 !

Used the F - word 12 x times watching the Thompson XDR Tiff.

Sorry !

Dear Herman,

May I assume that you were impressed with the image?

Dan,

Taking "new efforts" from Jpeg 2000 and AVC Intra 50/100 in Consideration, I think the XDR has proved the compression - mpeg "Sweet Spot" right.

I have been playing around with HDSI source material - compressed with Jpeg 2000 and AVC INTRA 100, taking a portable 'workflow" in consideration.

As is / The Flash XDR seems like the obvious choice.

Obviously your source input - in this case , the Thompson attributes a lot to the final result.

I think from a marketing perspective - Sample tiffs coming from a XLH1 as "mid range" source will emphasize your product more.There is thousands of "mid range" HDSDI outputs longing for the portable sweet spot.

Congratulations with the product !

Dear Herman,

We agree on all points. We do feel that we have found the "Sweet Spot" with 100 Mbps Long-GOP.

The Flash XDR and the upcoming nanoFlash dramatically improve the images from the under $10,000 cameras, such as the XL H1/XL H1s, EX1/EX3, and others with HD-SDI outputs.

Recording a 4:2:2 signal as opposed to the 4:2:0 signal when recorded in camera makes a big improvement in image quality.

This was our first opportunity to combine the Flash XDR with the Viper.

It was an instant success, Richard is very happy with the quality, and excited about the significant improvements in his workflow. 100 Mbps files are must easier to work with than uncompressed.

We will hear more from Richard after the Houston Marathon. This will be an interesting shoot, a real torture test of any codec with 25,000 runners; lots of detail and movement.

Richard will be examining the footage in detail after the event. Also, he will be performing other tests.

Using an ultra high-end camera, such as the Viper, with the Flash XDR could have shown up flaws in our Flash XDR, but we were able to meet Richard's high standards for quality.

Good to hear that XDR is a right choice for 8 or later 10 bit 4:2:2 Full HD recording device for those decent cameras.

I have been working on CF card based HD portable recorders for some Film Directors / DOPs here in HKG.

In coming days in HK, I will report comments from them on Genesis + XDR / F35 + XDR / RED one + XDR / Kinor + XDR / Varicam + XDR as high-end camera system as well.

And D21 + XDR at ARRI ASIA in HK as well.

Hope to see an uncompressed HD support soon on Flash XDR, and Dual Link + 2K option in future on " Turbo Flash XDR " in later 2009.

Do not forget Stereoscopic 3D will be a big market for CD for their 2 x HDSDI version of FlashXDR.

Sorry has SWR-1 not only can do 4:4:4 in SR mode but two HDSDI input of 2 cameras with TC Sync together for Stereoscopic field recording as well.

Sorry are so overpriced to me though !!!

Again, my 1 cent.

Cheers,

Stewart
Hong Kong - CHINA

Dear Stewart,

Just to be clear, Dual-Link HD-SDI is not in our present product plans for 2009.

Sorry Dan,

I am running too fast.
Never too late in over prediction in the fast developing technologies, and I have a great belief that CF based development will move as " fast as llightning " from 2009 ahead....

Yes, I know very clear from your today post for the current time table of present product plans at CD for 2009.

No offense to you to say that :
I had a Kona 3 card in almost 3 years ago to do dual link HD and 2K uncompressed capturing on my G5 anyway.

Anyway, it's a market driven economy to push us to the limit.

Respectfully,

Stewart
HKG - CHINA

Dear Stewart,

No problem at all.

I just did not want anyone to think that we were promising Dual-Link support.

Yes and Clear Dan.

Cheers,

Stew.

Hi Dan:
Respectfully, I don't consider this to be much of a test. This is more of a product "endorsement." You take what is considered to be the highest quality digital film production camera in the world operated by a recognized world expert and then plug that into a Flash XDR ? Of course the end result will be great. Why wouldn't the footage be great ? What I think is more of a good test is end result shown on the Convergent Website and other tests taken from the Canon XL H1 HD SDI signal recorded to the XDR.

Dear Mark,

The Flash XDR was requested by Mr. Richard Welnowski, he wanted to test the Flash XDR.

When one normally records 4:4:4, full uncompressed, one would need to test (or confirm) that 4:2:2, at the desired frame rates, at 100 Mbps, would actually meet his need for quality.

....What will become utterly fascinating will be any possible future tests between the Flash XDR running in full uncompressed mode 4:2:2 recording the 4:2:2 signal out of the Viper. Of course, I understand Mr. Welnowski's requirements for a manageable file size for editorial post. I bet the 50 Mbps setting on the XDR would look pretty darn good out of the Viper camera as well.

We now have more images courtesy of Mr. Richard Welnowski.

The first is from a still camera, not the viper. It shows the Thomson Viper and the Flash XDR in use in Kansas City, Missouri.

Richard was out testing the Flash XDR with his camera.

http://convergent-design.fileburst.com/Thomson%20Viper/XDR-Viper-KC_2341.jpg



This image is a frame capture from the Thomson Viper, as recorded to the Flash XDR at 100 Mbps, 4:2:2 1080p30.

http://convergent-design.fileburst.com/Thomson%20Viper/KCframe-30p.tif



This is also 100 Mbps, 4:2:2, but 1080p24. Notice the motion bluring due to the speed of the vehicles and the shutter speed used.


http://convergent-design.fileburst.com/Thomson%20Viper/KCframe-24p.tif


Richard is currently writing an article documenting his experiences with the Flash XDR.

We now have more images courtesy of Mr. Richard Welnowski.

The first is from a still camera, not the viper. It show the Thomson Viper and the Flash XDR in use in Kansas City, Missouri.

Richard was out testing the Flash XDR with his camera.

http://convergent-design.fileburst.com/Thomson%20Viper/XDR-Viper-KC_2341.jpg



This image is a frame capture from the Thomson Viper, as recorded to the Flash XDR at 100 Mbps, 4:2:2 1080p30.

http://convergent-design.fileburst.com/Thomson%20Viper/KCframe-30p.tif



This is also 100 Mbps, 4:2:2, but 1080p24. Notice the motion bluring due to the speed of the vehicles and the shutter speed used.


http://convergent-design.fileburst.com/Thomson%20Viper/KCframe-24p.tif


Richard is currently writing an article documenting his experiences with the Flash XDR......Uhhhh, well- I'm not so impressed this time :-( The first picture of a frame taken from the Flash XDR as recorded by Mr. Welnowski, has excellent, excellent resolution, but the contrast values left alot to be desired to my eye. The first picture from the Viper looked soft & grainy to me Dan. The 24 fps frame looks much sharper, but not as sharp as the still image to XDR. The contrast values are much better in the 24 fps capture as well. There is definitely more information in the shadows of the underpass and less grain in the sky. I bet you this image looks nearly 3D in projection.

Mark,
I don't think the first image was recorded on the Flash XDR. It's in the frame with the camera. This looks like a behind the scenes shot taken with a separate digital camera.

Dear Michael,

That is correct. This was a shot just to show the setup.

We have numerous mounting options available as well as powering options.

Dear Michael,

That is correct. This was a shot just to show the setup.

We have numerous mounting options available as well as powering options.....OK. This explains the first image, but not the remaining two. The 30 fps viper frame looks desaturated and slightly blurry. Look at the window detail on the red brick building on the left. However, the 24 fps frame looks much better with improved overall sharpness and much better contrast and details in the shadow area. I think the 24 fps frame looks really good but not the 30 fps.

Dear Friends,

Richard Welnowski asked me to post this latest image.

This is the the same frame that we posted earlier, but using a better technique to capture the image. This is a frame grap from the Viper's footage, as recorded by the Flash XDR at 100 Mbps.

http://convergent-design.fileburst.com/Thomson Viper/KC2-XDR-sf29-c.tif

This replaces the frame grab ending in ...-30p.tif

And this one replaces the -24p.tif

http://convergent-design.fileburst.com/Thomson Viper/KC2-XDR-sf24-c.tif

Dear Friends,

Richard Welnowski asked me to post this latest image.

This is the the same frame that we posted earlier, but using a better technique to capture the image. This is a frame grap from the Viper's footage, as recorded by the Flash XDR at 100 Mbps.

http://convergent-design.fileburst.com/Thomson Viper/KC2-XDR-sf29-c.tif

This replaces the frame grab ending in ...-30p.tif....Yeah, this image looks a little better, but still, look at the Marriot Hotel to your left. It's a little muddy. Where's the detail in the windows ? Look at the roof of the church. The red tiles seem to smudge slightly. I think the early 24 fps grab of that shot looked much better for some reason. (??) Even with 24 fps's natural motion blur it look really great on the XDR.

....Yeah, this image looks a little better, but still, look at the Marriot Hotel to your left. It's a little muddy. Where's the detail in the windows ? Look at the roof of the church. The red tiles seem to smudge slightly. I think the early 24 fps grab of that shot looked much better for some reason. (??) Even with 24 fps's natural motion blur it look really great on the XDR.

Look at where the shot is coming from. He's positioned on a live roadway *OUTDOORS* with traffic going by. I'm sure that tripod was moving all over the place. Frankly, I'm impressed it looks as good as it does. Especially that wide.

Does he have the RAW images that came off the thing? I'd imagine many shooters have not see RAW Viper images and have nothing to compare these images to.

Dear Perrone,

I believe that, for these Flash XDR test shots, all recording was done in the Flash XDR at 4:2:2 HL and 100 Mbps.

I do not believe any other HD recorder was used.

So, are we looking at the Viper in HDStream mode (don't think so) or in the YCrCb mode, which would show color corrections and so forth made in the camera?

I'm not sure what the Flash XDR can ingest from the Viper.

Dear Perrone,

I will get the answer to your question from Richard.

I do know that he put the Viper in 4:2:2 mode so that we could work with the HD-SDI output.

Dear Perrone,

I will get the answer to your question from Richard.

I do know that he put the Viper in 4:2:2 mode so that we could work with the HD-SDI output.

I am just asking out of curiosity. Not like I'll be shooting with the Viper any time soon. Apparently, you can take the RAW output off dual-link SDI, or take 4:2:2 unprocessed over SDI, or YUV over SDI with in-camera color corrections. Since this footage doesn't have the characteristic green tint, I am guessing that the last option was the one used. It WOULD be nice to know if the HDStream mode could be recorded with the XDR since it is also a 4:2:2 stream.

Dear Perrone,

Richard responded to my request and your questions:
(He responded very quickly, the delay has been on my part.)


I only record 4:2:2 YCrCb format to Flash XDR.

I could record HD Stream 4:2:2 also, but Viper is a 10bit and XDR is 8bit, therefore after I will have to grade image in 8 bit, which I do not think is good idea and will complicate workflow in post.

I setup the color balance, and all other parameters in the camera and used single HD-SDI processed signal to feed the Flash XDR.

If XDR becomes a 10-bit device I will record in 10-bit log output HD Stream and will grade the image later, this will give more flexibility in grading the final image in post.

Also with FilmStream and HD stream bypassing the processing, this increases the quality of image, especially noise in dark parts of image and you can gain couple of f-stops latitude.

Richard



So, he recorded in HD-SDI 4:2:2, with the image pre-processed in the camera, so that the Flash XDR directly recorded the desired image. This allows for a faster workflow in the Non-Linear Editor, exactly what Richard was wanting.

When we have Full Uncompressed with 10-Bit / 8-Bit available as an option, then Richard will have his choice of how to record out of the Viper.

The Flash XDR cannot handle 4:4:4, it has to be 4:2:2.

Yea, the HDStream is 4:2:2 but 10-bit. Wasn't sure if you guys could handle that. Keep up the good work on this unit. I've spent a bundle in the past 8 months moving to HD here at the office, and still have more to go. I'd love to get one of these units in the future though.

Dear Perrone,

Richard responded to my request and your questions:
(He responded very quickly, the delay has been on my part.)


I only record 4:2:2 YCrCb format to Flash XDR.

I could record HD Stream 4:2:2 also, but Viper is a 10bit and XDR is 8bit, therefore after I will have to grade image in 8 bit, which I do not think is good idea and will complicate workflow in post.

I setup the color balance, and all other parameters in the camera and used single HD-SDI processed signal to feed the Flash XDR.

If XDR becomes a 10-bit device I will record in 10-bit log output HD Stream and will grade the image later, this will give more flexibility in grading the final image in post.

Also with FilmStream and HD stream bypassing the processing, this increases the quality of image, especially noise in dark parts of image and you can gain couple of f-stops latitude.

Richard



So, he recorded in HD-SDI 4:2:2, with the image pre-processed in the camera, so that the Flash XDR directly recorded the desired image. This allows for a faster workflow in the Non-Linear Editor, exactly what Richard was wanting.

When we have Full Uncompressed with 10-Bit / 8-Bit available as an option, then Richard will have his choice of how to record out of the Viper.

The Flash XDR cannot handle 4:4:4, it has to be 4:2:2.....Huh ? Flash XDR is only 8 Bit ? Ugghhhhh ! I strongly dislike color grading in 8 bit. Shooting doesn't matter, but posting with an 8 bit file sure does ! One of the reasons we went up to Avid Media Composer from Xpress Pro, was to be able to import, capture edit and color grade in full 10 bit. I find the difference is especially noticeable when you have to deliver to a crippled video format like those for the Internet. Web video has improved enormously over the past two years, but I find it easier to keep image density and color renditions accurate if I encode from a 10 bit versus an 8 bit file. I have also observed color grading to be much more precise with those extra 2 bits. One wouldn't have thought there would be much of a difference, but there is. However, the 100 Mb 8 bit file quality of the Flash XDR is very, very clean.

Just because you record in 8 bit doesn't mean you have to post in it...

Just because you record in 8 bit doesn't mean you have to post in it......True. For some reason I just assumed Flash XDR was 10 bit. This is the reason I said shooting with an 8 bit camera doesn't matter, so much as recording your 8 space to 10 space and workflow stays 10 bit. However, it is what it is and importing an 8 bit file is but one hit against you. Dan was mentioning an eventual 10 bit uncompressed mode to come, so this would certainly improve things once more by allowing you to walk away with a 10 bit recorded file from an 8 bit camera :-)

Exactly. But I'd REALLY prefer to see 10 bit support in things OTHER than uncompressed. 10 bit uncompressed is REALLY going to burn through media. Would be great to get an hour's worth of 10 bit 4:2:2 for the ~$5k cost of the box.

Exactly. But I'd REALLY prefer to see 10 bit support in things OTHER than uncompressed. 10 bit uncompressed is REALLY going to burn through media. Would be great to get an hour's worth of 10 bit 4:2:2 for the ~$5k cost of the box....Hey Perrone. Yeah, imagine 50 Mb & 100 Mb long Gop in 10 bit color space @ 4:2:2 ? Wow - especially with that super clean MPEG 2 encoding chip Convergent's got in there. Can you explain what the quality difference is between I frame and Long GOP MPEG 2 ?

Just to be clear, all MPEG2 is 8-bit.

When we implement full uncompressed, we can do 8 bit or 10 bit. This is no longer using MPEG2 as it is uncompressed.

If we do 10 bit, then you have your choice of using an 8 bit or 10 bit camera.

An 8 bit camera, with 10-bit full uncompressed, is still 8 bit's effective.

With an 8-bit camera, in 10-bit full uncompressed, the least most two binary digits will always be 0.

Just to be clear, all MPEG2 is 8-bit.

When we implement full uncompressed, we can do 8 bit or 10 bit. This is no longer using MPEG2 as it is uncompressed.

If we do 10 bit, then you have your choice of using an 8 bit or 10 bit camera.

An 8 bit camera, with 10-bit full uncompressed, is still 8 bit's effective.

With an 8-bit camera, in 10-bit full uncompressed, the least most two binary digits will always be 0.......Hey Dan MPEG 2 is 8 bit ! You learn something new every day :-)

Just to be clear, all MPEG2 is 8-bit.

When we implement full uncompressed, we can do 8 bit or 10 bit. This is no longer using MPEG2 as it is uncompressed.

If we do 10 bit, then you have your choice of using an 8 bit or 10 bit camera.

An 8 bit camera, with 10-bit full uncompressed, is still 8 bit's effective.

With an 8-bit camera, in 10-bit full uncompressed, the least most two binary digits will always be 0.

Dan,

Do you happen to know whether the new Canon XL H1S is 8bit or 10bit out of its HD-SDI?

Dear Michael,

Sorry, but no.

With the XLH1 it's been said that the HD-SDI stream is 10 bit, but that 2 of those 10 bits are 0's.

I have always been confused by this.

Dear John,

With the original XL H1, the HD-SDI signal only contains 8 bits of actual data from the camera, since the XL H1 is an 8-bit camera.

The HD-SDI standard calls for 10 bits.

So the 8 bits of data are put into a 10 bit word, with the 8 bits being the most significant part, then the least significant part, the last two bits, are set to zero.

It all works.

If a camera has 12 bits of resolution, then only the most significant 10 bits can be sent over HD-SDI. The last two bits of resolution would be dropped.

If a camera has less than 10 bits of resolution, then the bits are put into the most significant part, and the remainder is set to zeros.

The important thing is to examine the image visually.

With high bit-rates, the image can still be good (or great, depending on your opinion), with only 8 bits.

Examine the images from the Viper, using only 8-bits. The images look good and it is hard to find "banding". In my personal opinion, "banding" can be caused by a limited number of bits, or a low-bit rate, high compression codec.

I have never been troubled by "Banding" with my original XL H1. I have shot in situations where I expected banding, but could not find it.

In working with the XL H1s, at the Canon Booth last year, I tried hard to get banding, but I could find any. I was shooting at a finely graduated poster which I was positive I would see banding, but I did not. I was using the HD-SDI output going into a monitor.

I do feel that it is best to work in 10 bit in post.

Dear John,

With the original XL H1, the HD-SDI signal only contains 8 bits of actual data from the camera, since the XL H1 is an 8-bit camera.

The HD-SDI standard calls for 10 bits.....This was one of the reasons why I thought the XDR & the Nanoflash were 10 bit, because they both possessed an HD-SDI interface. Furthermore, I thought the XL H1 was 10 bit when recording from the HD-SDI output.