Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: Nikon D4 Passes but the D800 Fails the EBU/BBC Broadcast Quality Test  (Read 10802 times)
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« on: August 31, 2012, 12:35:42 AM »
ReplyReply

The EBU/BBC Test reports on the Nikon D4 and D800 are out, and the results are very surprising, at least to me.

The D4 barely makes the grade, while the D800 fails completely. The D800E wasn't tested. Major reason: Aliasing.

Read my notes on the report here: http://wolfcrow.com/blog/nikon-d4-passes-but-the-d800-fails-the-ebubbc-broadcast-quality-test/

I have also included links to the actual reports in PDF for download.

Thoughts, anyone?
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
RFPhotography
Guest
« Reply #1 on: August 31, 2012, 06:29:28 AM »
ReplyReply

The 5D Mk II also 'failed' the BBC test.  It's been used to shoot an episode of a major Hollywood TV series and Showtime used it to shoot a major production on the annual Army/Navy football game.  Those are two significant instances that come immediately to mind.  There are likely others. 
Logged
ftbt
Jr. Member
**
Offline Offline

Posts: 88


« Reply #2 on: August 31, 2012, 08:33:25 AM »
ReplyReply

The 5D Mk II also 'failed' the BBC test.  It's been used to shoot an episode of a major Hollywood TV series and Showtime used it to shoot a major production on the annual Army/Navy football game.  Those are two significant instances that come immediately to mind.  There are likely others.  

"Act of Valor" is another. (See: http://blog.planet5d.com/2012/02/act-of-valor-leap-of-faith/) And, with respect to the venerable BBC, according to Wikipedia:

"The BBC Two comedy series Shelfstackers, first broadcast on 4 September 2010, is the first BBC programme to use the camera. The corporation had initially refused its use due to "lack of quality" but were persuaded otherwise by the series' director, Dom Bridges. All six episodes of the series were shot on the camera for a total budget of 160,000."

George Lucas also used the camera for quite a few shots in his latest film that portrayed the Tuskegee Airmen, "Red Tails." And, an Oscar nominated documentary, "The Tsunami and the Cherry Blossom" was shot on the camera as well.
« Last Edit: August 31, 2012, 12:53:44 PM by ftbt » Logged
RFPhotography
Guest
« Reply #3 on: August 31, 2012, 08:51:26 AM »
ReplyReply

Interesting.  Wasn't aware of the big screen production. 

The Beeb kind of makes a mockery of its own 'standards'. 

Pretty much tells you all you need to know about 'the BBC Test'.
Logged
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« Reply #4 on: August 31, 2012, 09:36:26 AM »
ReplyReply

I wish the picture were that rosy.

Surely if the camera was that good it would be used for ALL shots of a major motion picture. So far, no major movie has used a DSLR exclusively, unless it's a crash cam or a compromise. Every single person I know who use a DSLR for video do so because they can't afford anything else.

And if the one episode of House was shot with a 5D and they were so happy, surely they should have ported all the other shows to it - the rental and operating costs of an Alexa are much higher than a 5D. One should never underestimate the marketing clout of a major camera manufacturer.

And if BBC was that convinced about the 5D, surely they would allow all producers to submit DSLR footage carte blanche - but try submitting something to BBC with DSLR material. I know people who shoot for the BBC, and sometimes submit material from cameras that are not up to broadcast standards.

With the right connections, anything is possible. But that's sort of like the black market, isn't it? The good guys trying to get in through the front door are screwed. The fact remains that most of the time, if you are submitting to a major network, one of the dumbest things you could do is shoot on a system that was officially rejected by them. This is also true of PBS or ESPN or any other network.

What is most surprising to me is their acceptance of the D4 - that is a first, and totally unexpected. Its codec is only about 24 Mbps - that means they have to accept the FS100, the AF100 and every other AVCHD camera out there! The real answer, no one outside the EBU will know.
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
RFPhotography
Guest
« Reply #5 on: August 31, 2012, 10:26:32 AM »
ReplyReply

When it comes to bitrate, you have to compare apples to apples.  AVCHD/MPEG4 is supposed to produce the quality of MPEG2 (what the standards are based on) at about half the bitrate. 

If you read the posted article on Act of Valor you'll see that the vast majority of it was shot with a 5D Mk II. 

As far as the Beeb, or any other network, allowing or disallowing DSLR video, it can take a long, long, long time to move stuck in the mud mindsets.

If everyone you know is using a DSLR for video because they can't afford anything else, then perhaps it's an indication that you need to broaden your range of acquaintances.  Grin
Logged
mac_paolo
Sr. Member
****
Offline Offline

Posts: 420


« Reply #6 on: August 31, 2012, 10:28:08 AM »
ReplyReply

What is most surprising to me is their acceptance of the D4 - that is a first, and totally unexpected. Its codec is only about 24 Mbps - that means they have to accept the FS100, the AF100 and every other AVCHD camera out there! The real answer, no one outside the EBU will know.
It can output uncompressed stream via HDMI, so the low bitrate is not "the issue".
Logged
RFPhotography
Guest
« Reply #7 on: August 31, 2012, 11:15:21 AM »
ReplyReply

It can output uncompressed stream via HDMI, so the low bitrate is not "the issue".

It doesn't seem that the uncompressed video is what was tested though.
Logged
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« Reply #8 on: August 31, 2012, 10:11:46 PM »
ReplyReply


If everyone you know is using a DSLR for video because they can't afford anything else, then perhaps it's an indication that you need to broaden your range of acquaintances.  Grin

Yes, by a factor of a billion!!

Anyway, please read David Heath's excellent and informative reply here: http://www.dvinfo.net/forum/digital-video-industry-news/510356-nikon-d4-passes-but-d800-fails-ebu-bbc-broadcast-quality-test.html
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7308


WWW
« Reply #9 on: September 01, 2012, 01:35:32 AM »
ReplyReply

Hi,

My guess is that the primary issue is that both the D4 and D800 are intended for stills, and thus high resolution sensors with correct antialiasing filters for native resolution. The sensor output from the D4/D800 needs to be downsampled to 1920x1080 this is quite possible with optimum processing but needs correct antialiasing methods in software. I guess that the ASICS presently used are not really optimized for video.

Regarding the DR issue, we need to keep in mind that video is not a 14-bit raw data stream but is preprocessed into 8-bit. Tonal range is compressed while converting to 8 bit.

Best regards
Erik
Logged

Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« Reply #10 on: September 01, 2012, 07:04:09 AM »
ReplyReply

Regarding the DR issue, we need to keep in mind that video is not a 14-bit raw data stream but is preprocessed into 8-bit. Tonal range is compressed while converting to 8 bit.

Best regards
Erik

You're right, Erik. Line skipping is the culprit.

I did a comparison of RAW vs Technicolor Cinestyle vs H.264 here - still camera RAW is miles ahead.
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
Bern Caughey
Full Member
***
Offline Offline

Posts: 179



WWW
« Reply #11 on: September 01, 2012, 12:01:22 PM »
ReplyReply

I couldn't open the EBU PDF just now, but assume it was done by Alan Roberts. I'm not smart enough to fully evaluate his tests, but do know he often misses widely on actual tech specs.

I do understand the need for these standards, & respect them. Capture is only a small part of the chain, & any flaws can be exaggerated as the footage makes it's way to broadcast.
Logged
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« Reply #12 on: September 01, 2012, 10:43:34 PM »
ReplyReply

Capture is only a small part of the chain, & any flaws can be exaggerated as the footage makes it's way to broadcast.

Good point, Bern. I feel the internet has a better delivery mechanism than broadcast today. With broadcast, a typical scenario is:

  • A master is transcoded to HDCAM SR for delivery(MPEG-4)
  • That is dumbed down by the broadcaster (usually to MPEG-2, if I'm not mistaken)
  • Cable operator or Satellite provider dumbs it down again into MPEG-4 or MPEG-2

It's very surprising it still looks okay after all that abuse. Compared to that, the HDMI feed off a 7D or a D4/D800 should look great!
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
bcooter
Guest
« Reply #13 on: September 03, 2012, 11:10:35 AM »
ReplyReply

I always find broadcast standards, BBC or anywhere, any country, any client to be funny.  Sure they need a baseline, but when it comes to news coverage or documentary production if it suits them or the footage is unique or newsworthy, they'll accept footage from anything, including a smart phone.

We all see it all the time, shaky busted up footage of some reporter covering an event and If it's advertising they all accept anything as long as the client is paying.

But all of this is the dumbed down version and none of us aspire to shoot dumbed down, we aspire to shoot professionally and one thing that is missing in all of these standards is focus and lighting.

Nothing will make footage look better than professional lighting and camera movement, regardless of the capture device.  That's why SMALL  cinema/tv production crews are twenty five to 50 people.

I agree with both Bern and Sareesh.

Quality at the time of capture makes a huge difference up and down the chain.   You can't really get rid of moire or artifacts properly with any program, though quality can be defined in a lot of ways.

But quality of the output is more than just the file size, bit depth, bit rate, or compression, it's a combination of all of them, with (IMO) more heavily weighted towards capture bit depth and sharpness.

If you want to do your own test, take a professional still of the same motion scene and put the still in your editorial system at around 2500 pixels across.  Then look at it on a broadcast monitor compared to the motion footage and the still will usually look much better regardless of how your compress and downsize later.

You can do you own faux broadcast test, put a a video in m4v, in any size from 480 to 720 on the short side and play it through your apple TV on your home broadcast monitor.  Compared to what the cable and satellite companies stream at 1080i usually the wireless apple TV footage looks as good, or better.

Regardless, as Bern says, the better your start the better you usually finish and really all of this usually comes down to budget.  Everyone will mention the House episode shot on a 5D, but the crew wasn't cut down and the process was just as expensive in post.

Same with Act Of Valor, http://www.hurlbutvisuals.com/blog/2012/05/act-of-valor-workflow/   

There might have been a savings in shooting with 5d's due to the multiple cameras, but just a few of those fluid heads or one panavision lens can costs more than all the cameras combined.

When you factor in the extra post work, much of which the workflow was ground breaking and make it up as you go, but they did a tremendous job and used those little cameras to the best of their abilities and don't think the BBC won't air this movie.

Anyway, this was a test I did early on with a 5d2 comparing the still capture vs. the video capture on the same exact scene and the 5d shoots soft compared to the stills.  Looks good in motion, but the still frame from the motion footage is soft and does some crazy stuff like the black hole in the blowout of the practical lights in the background, where the still image of the same scene is much sharper with less issues.

http://ishotit.com/rundsmc.jpg

IMO


BC

Logged
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« Reply #14 on: September 03, 2012, 10:01:52 PM »
ReplyReply


Anyway, this was a test I did early on with a 5d2 comparing the still capture vs. the video capture on the same exact scene and the 5d shoots soft compared to the stills.  Looks good in motion, but the still frame from the motion footage is soft and does some crazy stuff like the black hole in the blowout of the practical lights in the background, where the still image of the same scene is much sharper with less issues.

http://ishotit.com/rundsmc.jpg

IMO
BC

The black hole is weird! How did that happen?
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
bcooter
Guest
« Reply #15 on: September 05, 2012, 01:38:50 PM »
ReplyReply

The black hole is weird! How did that happen?

I'm not an engineer, though I saw this early on with 6mp still capture where there is a specular light in the background that is overexposed.

Maybe it's from the line skipping of the 5d2, but someone that makes this stuff should know.

When I did this test I was only trying to compare the output of the 1080 footage vs. the still footage.  Actually I was trying to see if I should go upstream and buy the REDs of stay with the Canons 5d2 (soon to be 3).

I went with RED's and use the Canon as a small car mount, tight area cam.

Actually, it seems all of the high end camera world is "on sale" today.  I know you can buy lightly used RED 1's MX at about $16,000 and Scarlets at about the same price, which with all my shooting and tests produces a much nicer file than any dslr or lower priced 2k cameras.

A used RED looks even more attractive by the time you completely kit a dslr out with matte boxes, rails, sound inputs/mixers, hdmi out recorders, etc.  I probably spent double on the 5d2 in accessories than I did on the camera and built up for professional work a dslr is lighter than any RED but not smaller.

Canon USA sent out an e-mail today where the C-300 is offered at 0% interest, so it may be the economy, or the dog days of summer, but high end cameras in still and digital video are priced to sell today.



IMO

BC
Logged
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« Reply #16 on: September 05, 2012, 10:23:48 PM »
ReplyReply


Actually, it seems all of the high end camera world is "on sale" today.  I know you can buy lightly used RED 1's MX at about $16,000 and Scarlets at about the same price, which with all my shooting and tests produces a much nicer file than any dslr or lower priced 2k cameras.

A used RED looks even more attractive by the time you completely kit a dslr out with matte boxes, rails, sound inputs/mixers, hdmi out recorders, etc.  I probably spent double on the 5d2 in accessories than I did on the camera and built up for professional work a dslr is lighter than any RED but not smaller.

Canon USA sent out an e-mail today where the C-300 is offered at 0% interest, so it may be the economy, or the dog days of summer, but high end cameras in still and digital video are priced to sell today.

IMO

BC

Agree 80% - I'd take a Red One over a C300 any day - only for its colors and tonality. I use the C300 and hate its look when 'strained'.

20% - Kitting up a Red One is hyper-expensive, even today. Very soon the accessories will get dearer too.

The worse thing about Red is its service. When my Canon DSLR had a scratch on its sensor, I went to the Canon repair center and got a new sensor in one week. I would factor in maintenance and post production costs with the Red - definitely worth it but no comparison with DSLR-budgeted projects.
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
bcooter
Guest
« Reply #17 on: September 06, 2012, 02:53:18 AM »
ReplyReply



20% - Kitting up a Red One is hyper-expensive, even today. Very soon the accessories will get dearer too.

The worse thing about Red is its service. When my Canon DSLR had a scratch on its sensor, I went to the Canon repair center and got a new sensor in one week. I would factor in maintenance and post production costs with the Red - definitely worth it but no comparison with DSLR-budgeted projects.

Red's service is kind of funky, but very detailed and yes some of their pricing is off the wall silly.  Those little magic arms on the RED one break.  It's just a screw that strips, but to get them to look at it cost like $75 and then they charge to fix it.  I can buy it from a third party for $150 so it doesn't make any sense, when it's just a screw and no they will not sell the screw.

As far as rods, matte boxes, mini xlr to full xlr adapters, batteries, etc. it's all available third party and cost no more or less than any other camera, also most of the used MX RED 1's are for sale with all the accessories.

I'm not pushing RED, I'm just saying if treated well an R1 will last a lot of years, (at least that's my plan) and it's a real 4k camera, with a real processing suite and more latitude than any camera still or motion, I've used.   

That stuff goes a long way to kind of future proofing a purchase and R1's are falling out of the trees for sale.

IMO

BC
Logged
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« Reply #18 on: September 06, 2012, 10:03:18 PM »
ReplyReply

This time I agree 100%. The Red One has definitely passed the 'reliability' stage.

Have you worked with the Epic? Unfortunately I haven't even seen one yet. I'm liking the skin tones and DR from it. Is it similar to R1?
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
bcooter
Guest
« Reply #19 on: September 07, 2012, 10:37:08 AM »
ReplyReply

This time I agree 100%. The Red One has definitely passed the 'reliability' stage.

Have you worked with the Epic? Unfortunately I haven't even seen one yet. I'm liking the skin tones and DR from it. Is it similar to R1?

No, I had an Epic on order, ready for delivery and I changed my mind and bought a second Red 1 and a Scarlet.

The R1's are reliable, the Scarlet then and somewhat now is a work in progress.

As far as the look, both the Scarlet and Red 1's have the MX sensor and should look the same, though the Scarlet has RED gamma 3 which is flatter and softer, (though now with a firmware update so do the R1's have red gamma 3).

Obviously you can change the gamma input settings in processing, though the R1's I usually process out in standard Red Gamma because to me, it looks like Cinema film where the Scarlet looks not just flat but somewhat video like, I can't really explain it until you see it.

The Scarlet is smaller, uses Canon lenses, doesn't autofocus well, (if that's important to you) and has to have a workaround to use even mini xlrs, as it has those awful Iphone type of mini sound plugs that pop and crackle if you look at them wrong.

I love the R1's, the Scarlet I almost sold, then think well, Red is good at finally getting there with firmware updates, but I use the R-1's most of the time.

The thing about the R1's is compared to most digital video cameras (except the Arri) they can be big so I have one outfitted for a solid mount with PL lenses and sticks and battery mount on camera etc., then I have the second R-1 stripped down with a still camera lens shade, small Zeiss still camera lenses battery, on a belt clip and a shoulder mount for hand holding which only is about 3 lbs heavier than an Epic or Scarlet, which is more than manageable.

The beauty of the zeiss zf still camera lenses is they have a short focus throw so tracking someone down the block to a few feet from the lens is much easier if your pulling focus yourself.  The PL mount works better if you have a dedicated person to pull focus remotely.

The thing I really dislike about the Scarlet is the glossy screen.  Working outside you track with someone and spin the camera and all of a sudden all you see is your face.  It's just amazingly annoying and wrecks the shot.

I also heard a lot of stories about the reliability of the R1's but I've used both of mine in incredibly extreme conditions without issue (knock on wood).

Bottom line is I don't own an Epic though the Scarlet is pretty close in form and function.  I just think the 50 grand cameras are getting kind of silly, for any production.

Without knowing the specs I even think the new 30 grand 4k Canon is overkill on price. IMO.

Second bottom line is I'll take two RED1's around the world (done that) and not worry.  I'll even take an R1 and a Scarlet around the world (done that) and not worry because I always have the R1.

IMO

BC
« Last Edit: September 07, 2012, 11:35:44 AM by bcooter » Logged
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad