I have to admit the science is a bit beyond me. I've always been told that on 35mm you don't want to go beyond about f8, f5.6 on 2/3" etc. I was just responding to the idea that getting big DoF on DSLRs is no problem as you can just shut the aperture well down - didn't seem to be as good an idea as the poster suggested.
Cheers,
Steve

I've always been told that on 35mm you don't want to go beyond about f8, f5.6 on 2/3" etc.


That's probably because the advice is geared towards a resolution far higher than 1080p (which is just 2 MP).


I was just responding to the idea that getting big DoF on DSLRs is no problem as you can just shut the aperture well down - didn't seem to be as good an idea as the poster suggested.

I think it's a great idea!

Unfortunately, I haven't seen anything with an impressive dynamic range from the video on a DSLR. The colour depth (space?) from the video on a DSLR, you're the first person I've heard make that claim, usually it's described using terms like fragile. I can understand people saying positive things about a RED in this regard.

I was referring to a constant aperture zoom for the Canon would be physically larger than the current 20x lens, which does have a pretty large aperture ramp. Not always an issue, but as it can cause an issue if this noticeablely kicks in earlier than say half way through the zoom range.

Unfortunately, I haven't seen anything with an impressive dynamic range from the video on a DSLR.


I'll have to shoot a comparison between the XH-A1 and 5D2 sometime.


The colour depth (space?) from the video on a DSLR, you're the first person I've heard make that claim, usually it's described using terms like fragile. I can understand people saying positive things about a RED in this regard.


I did not mean color space. I wrote "improved color depth (i.e. SNR over the used dynamic range)". What it means is that SNR (Signal to Noise ratio) over the part of the dynamic range that is used in the image is higher. For example, if you take a tone that is five stops below saturation on the XH-A1, the SNR is very poor. The low SNR negatively effects in the image in a variety of ways, including the reduction of the possible number of tonal levels (even in 8-bit recording). By comparison, the 5D2 color depth is much higher (at the same exposure, not the same DOF).


I was referring to a constant aperture zoom for the Canon would be physically larger than the current 20x lens, which does have a pretty large aperture ramp. Not always an issue, but as it can cause an issue if this noticeablely kicks in earlier than say half way through the zoom range.

I see. Thanks for the clarification.

They already invented that -- just use the f-number. On FF35, you get the exact same effect by switching between f/2.8 and f/24. f/24 (and high ISO) gives you the same (poor) low light performance and deep DOF as 1/3", while f/2.8 (and low ISO) gives you the super thin DOF and amazingly good low light performance of super large sensors.

While that is true, one would have to add a fair amount of ISO (gain) to make f24 = f2.8 Which would lead to noisy and trending to flat images.

I just tested this on my 5DMkII and to get f4 @100 iso to equal f22 I had to go to iso 3200. Not exactly the cleanest setting.

Ideally one would want clean images for whatever they shot...

Daniel, one way of looking at what you're saying is that f28 on 35mm is the same as f4 on 1/3" so it's OK. The other way of looking at it is that f28 on 35mm is majorly diffraction affected and therefore so is f4 on the 1/3" chip - that's the way I see it. I thought that was one of the issues broadcasters had with 1/3" chips - the lenses were diffraction limited at every aperture!
Steve

While that is true, one would have to add a fair amount of ISO (gain) to make f24 = f2.8 Which would lead to noisy and trending to flat images.


Yes, it does lead to noisy images, but only in the exact same proportion as going to smaller sensors leads to noisy images.


I just tested this on my 5DMkII and to get f4 @100 iso to equal f22 I had to go to iso 3200. Not exactly the cleanest setting.

Ideally one would want clean images for whatever they shot...

Yes, you're right. Compared to itself, the 5D2 at ISO 3200 is not that clean. But have you compared it to a 1/3" single-chip camcorder in the same low light and DOF?

In my experience, the 5D2 is noisier, but not as much as you might expect from a 5-stop difference in f-number. Ideally, the 5D2 at f/22 ISO 3200 would match even the 3-chip camcorders in the same low light situation (e.g. XH-A1 at f/3.1). The reason it doesn't right now is what I explained in the big post above, and it's the reason why 3-chip will always be around for low light work: they have three times the amount of light for any given DOF. If you compared the 5D2 with a single-chip camera, it comes out more favorable, because then they both get the same total amount of light.

Part of the reason why the current DSLRS can't match the low noise level of camcorders (at the same DOF) is because they have additional handicaps on top of only a single senor: row skipping, which reduces the total amount of light by 1+2/3 stops, and suboptimal resampling algorithms which alias the noise power from higher spatial frequencies into the 1080p Nyquist. But they also have advantages, such as better read noise performance per area. After the other problems are addressed in the future, this performance advantage may be enough to offet the fact that it only has one sensor instead of three.

In any case, even if we compare just the cameras that are on the market now (like the 5D2), there are some situations where the extremely deep DOF of the 1/3" camcorders isn't needed. So instead of stopping down all the way to f/22 ISO 3200 to get an exact match, it may be acceptable (sometimes) to use only f/11 ISO 800 (or whatever the exact setting is that matches the noise level). The DOF will be thinner, but still deep enough for some applications. And at f/11 ISO 800, the noise level will be a much closer match to f/3.1 on the XH-A1. (Of course, these DSLRs are unsuitable for most applications for other reasons -- I'm just addressing the DOF/noise issue.)

The other way of looking at it is that f28 on 35mm is majorly diffraction affected and therefore so is f4 on the 1/3" chip - that's the way I see it.

Yes, either way works for me.

Wait till you see what CANON will be offering. After NAB, they will introduce a solid state camera that is niether HDV or AVCHD. Later this year or early next year look for a camera that will give SONY a 'run for the money.'


Probably the Canon 5d mark 11 chips?

High ISO/gain with minimal grain......shoot in ambient light w/o a light

Only time will tell.

BTW: The JVC XH A1S will still be in the lineup due to good sales.

BTW: The JVC XH A1S will still be in the lineup due to good sales.

Do you mean Canon?

JVC already makes a solid state cam that is neither HDV or AVCHD (actually a couple).

JVC also doesn't make the XH-A1s, but it has occurred to me that JVC isn't all that shy about coloring outside the lines, and if any of the majors took the guts of a DSLR and made a video came with it, it wouldn't surprise me if JVC was the one to do it.

Daniel, one way of looking at what you're saying is that f28 on 35mm is the same as f4 on 1/3" so it's OK.
What is of significance regarding diffraction is the actual physical dimension of the aperture, not the f stop. That's because nothing can be done about the wavelength of light. ("Ye canna change the laws of physics!") What f no that corresponds to will depend on focal length.

Hence a given diameter of aperture may correspond to f4 for a short focal length lens, but f28 or whatever for a lens of longer focal length - which is Daniels fundamental point.
I thought that was one of the issues broadcasters had with 1/3" chips - the lenses were diffraction limited at every aperture!
Steve
Not quite. Diffraction will limit the smallest acceptable aperture, and engineering considerations will limit the largest aperture. (Together with cost.) The problem with 1/3" chips is that the limits effectively close together to leave a usable adjustment window of only a stop or two - not much use if you're following a subject from a sunny area to a shaded one, for example.

That doesn't apply with bigger sensors as it's normally possible to make lenses with max apertures much greater than the diffraction limited minimum - provided the cost, size and weight issues are tolerable. Cost and weight aside, you SHOULD be able to get around f2 for any chip size, so the bigger the chip, the bigger the usable f stop range possible. It will just cost more - a lot more.

The other (separate but related) issue is depth of field. Even wide open, the lens on a 1/3" chipped camera normally gives too much depth of field for any real control.

Hence the desire to use 2/3" cameras as a sensible compromise, and if they are just too big, then 1/2" is far better than 1/3".

Broadly, I agree with Daniels reasoning, but the only factor that seems to be left out is the relative efficiencies of single chip v 3 chip. For the same resolution and overall imaging size, 3 chip must always be more efficient than single chip as no light gets lost in the filtration.

What is of significance regarding diffraction is the actual physical dimension of the aperture, not the f stop.


Great post, David; I agree.


Broadly, I agree with Daniels reasoning, but the only factor that seems to be left out is the relative efficiencies of single chip v 3 chip. For the same resolution and overall imaging size, 3 chip must always be more efficient than single chip as no light gets lost in the filtration.

Agreed. It gives 3-chip an advantage of 1+2/3 stops in low light. I think 3-chip is going to stick around for the low light deep DOF applications.

Sorry to interrupt this interesting discussion, but my concerns are more basic. After having read Chris' article about the new Canon, I'm a bit disappointed.

It's quite likely, that it won’t comply with the EBU specification for HDTV not to mention certain national specifications – and that’s the major problem for me as a videojournalist. I am looking out for an affordable, compact cam that will meet German specifications (2/3” chips, 50 MBit data rate, XD CAM HD 422 or P2HD codec, 720 p 50 recording) on the long run but as far as I can see, there’s currently nothing on the horizon coming close to that. And yes, broadcasters sometimes make exceptions from specs – regarding their own equipment!!! As a freelancer / small service producer you better don’t rely on their mercy (“He’s got outdated gear, but he’s such a nice guy…”). Also, there’s a defined ‘transition time’ to get ready for the specs (in Germany until end of 2011) – but what then?

So, what do you think: Will ‘they’ bring out a prosumer cam meeting these specs? And if so: When will it probably happen?

Currently your best bet is the EX series with a Nanoflash, later in the year (assuming RED keep to their current time scale) the 2/3" Scarlet looks like a likely candidate for broadcast work - assuming you can live with the rendering times etc.

Personally, I feel the 2/3" is the best compromise format for many broadcast productions. No size fits all, so there will be a need for the 1/3" and the larger sensors depending on the particular needs of a production and its market.

The manufacturers may be waiting to see what the demand for the 35mm Scarlet actually is, although I've a feeling such a camera from other manufactures is likely to be in the EX3 type price ball park plus.

Quote: "What is of significance regarding diffraction is the actual physical dimension of the aperture, not the f stop."

That's semantics, for our purposes it amounts to much the same and I'm sure you know that's what I meant.

Quote: "Not quite. Diffraction will limit the smallest acceptable aperture, and engineering considerations will limit the largest aperture."

Yes, and these limits might even be all the way from the smallest to largest apertures, that's what I said.

Steve

That's semantics, for our purposes it amounts to much the same and I'm sure you know that's what I meant.
Yes, I knew what you meant, but I suspect this whole subject is confusing to anyone who doesn't already understand. Why on earth should the diffraction problem start at f4 for one camera, and not until f45 for another?

My post was intended to give a bit of physics as to WHY. The wavelength of light is fixed, diffraction becomes an issue when dimensions come down to a given multiple of it, hence the problems set in when the iris is equal or less than X millimetres. For different chip sizes, different focal lengths are needed for equivalent angles of view, and since f no=size of aperture/focal length, it follows as to why the same physical size of aperture can be referred to as many different f stops.

I'm one of those who loses interest (or I suppose understanding) when the physics get too complex, but yes it is nice to have it there as reference material - good to have you here to give it!
Steve

Currently your best bet is the EX series with a Nanoflash, later in the year (assuming RED keep to their current time scale) the 2/3" Scarlet looks like a likely candidate for broadcast work - assuming you can live with the rendering times etc.

Personally, I feel the 2/3" is the best compromise format for many broadcast productions. No size fits all, so there will be a need for the 1/3" and the larger sensors depending on the particular needs of a production and its market.

The manufacturers may be waiting to see what the demand for the 35mm Scarlet actually is, although I've a feeling such a camera from other manufactures is likely to be in the EX3 type price ball park plus.

Thanks Brian. The current EX 3 does not comply with these specs, but may be a new model would. I'm not too optimistic regarding 'Scarlett' - the release was postponed... how often? For the next months, I'll probably stay with my XH A1, keeping my eyes open.

The Nanoflash overcomes the problem with the EX3 codec.

That's the problem with telling everyone about your products during development (something that usually tends to slip), perhaps the Apple total secrecy is better in that regard.

Sorry to interrupt this interesting discussion, but my concerns are more basic. After having read Chris' article about the new Canon, I'm a bit disappointed.

It's quite likely, that it won’t comply with the EBU specification for HDTV not to mention certain national specifications – and that’s the major problem for me as a videojournalist. I am looking out for an affordable, compact cam that will meet German specifications (2/3” chips, 50 MBit data rate, XD CAM HD 422 or P2HD codec, 720 p 50 recording) on the long run but as far as I can see, there’s currently nothing on the horizon coming close to that. And yes, broadcasters sometimes make exceptions from specs – regarding their own equipment!!! As a freelancer / small service producer you better don’t rely on their mercy (“He’s got outdated gear, but he’s such a nice guy…”). Also, there’s a defined ‘transition time’ to get ready for the specs (in Germany until end of 2011) – but what then?

So, what do you think: Will ‘they’ bring out a prosumer cam meeting these specs? And if so: When will it probably happen?

Jan,

Pardon my assertion here, but I think you are waiting for the impossible. What you are looking for already does exist (PDW-700, PMX-350, HPX2100...), you just need to spend more and get a "professional" camera. I put quotes around the word because any camera with 2/3" chips and 4:2:2 50mbps recording will never be labelled "prosumer" and cost $5,000 USD.

Canon has a history of making still cameras, lenses and consumer/prosumer video cameras. These are their stated markets and they do not want to produce high end video cameras to compete with their provideo lens customers.

I am dissapointed that we will have another 1/3" chip camera on the market as well, but the possibility of this new camera being 2/3" with 4:2:2 is just not on the radar.

:)

WOOPS! Yes.... CANON. I own too many cameras :-)

Do you mean Canon?

After shooting with the 7D for several months and having owned a XHA1, I would love a camera with great low light capabilities, more generous DOF (read easier to focus), a big, beautiful monitor, and 20X lens. The XHA1 updated with a much lower noise sensor and perhaps a little shallower DOF (like a 2/3" sensor in conventional thought) would be great. Would like to use a less demanding compression scheme then H.264 or .mov if possible. I'd rather use larger cards and easier-on-my-pc compression. At 100Mbps, that would be around 50GB per hour with sound which means somewhere around 85 minutes on a 64GB card.

Oh, and give me more latitude in post (4-2-2?) and great audio!

I am looking out for an affordable, compact cam that will meet German specifications (2/3” chips, 50 MBit data rate, XD CAM HD 422 or P2HD codec, 720 p 50 recording) on the long run but as far as I can see, there’s currently nothing on the horizon coming close to that.
I'd seen the EBU remarks, and the BBC position in the UK, but it's interesting to hear what is happening in practice in the rest of Europe. Just to clarify, don't you mean at least 1/2" chips? That's the EBU recommendation, and I don't see any likelihood of 2/3" in a compact camera.

You're right. Nothing on sale now or on the horizon meets the spec, and in principle I agree fully with Brian - currently, the EX cameras are the closest by far, and by adding an external recorder can be made to meet it fully. Yes, the nanoFlash is the obvious one, but if you're an FCP user it's also possible to record directly to ProRes, albeit at the expense of portability.

The position amongst broadcasters does seem to be hardening on minimum spec. Latest I've seen is via TVBE - TVB Europe : January 2010 (http://cde.cerosmedia.com/1Y4b4ee89cce59b253.cde) - when the BBC's position for HD was underlined at a seminar. (See p10)
But despite the growing acceptance of HD, Quested showed no sign of relaxing quality standards, reeling off a list of no-go specifications. These included footage from up-res SD, Super 16, HDV or any camera with sub half-inch sensors, compression at less than 100Mbs (intraframe) or 50Mbs (interframe). NLE codecs operating below 160Mbs and live MPEG-2 contributions at less than 60Mbs, plus all 720p equipment – with the honourable exception of the Panasonic Varicam, as it’s variable framerate, so essential to landmark knowledge series such as "Life" was not available on 1080-line kit.

1/2" might be o.k. if they can get them to provide great low light.

Tim: Thanks, I do know, that there are already cams in the pro segment meeting the specifications. But considering the given salaries for Veejays in Germany, it is just not economic to buy them. Apart from that, most of them are too bulky for a 'one-man-show'. May be, switching to rental services is a solution for a 'transition period', until a) industry will offer suitable and affordable gear or b) the broadcasters will finally lower their demands a little bit. But considering Davids quote of the BBC position I doubt, that they'll do so.

David: No, I am quite sure regarding 2/3" Chips. My source is the 'Handbuch HD-Produktion',published end of Nov. 2009 (Peter Vogel, Andreas Effenberg (HG), edition: Schiele&Schön, Berlin) that is putting together these guidelines.

Quote (p. 157, referring to the public broadcaters ARD, ZDF, ORF, translated by me): "Camera sensors should have native 1920 x 1080 pixels and should at least be as large as 2/3".

Additionally, some exceptions are defined (p 155):

"HD Low-Cost and Consumer formats are solely acquisition formats and will only be used for special purposes like video journalism, newscasting from crisis areas, etc." Sounds moderate at the first look, but: "Therefore, a further processing of these formats has to be avoided. For example, XD CAM EX, HDV and AVCHD are not professional HDTV formats."

And that means, as far as I do understand it, that you may -occasionally- shoot on these formats. But for the post, you have to transfer it to professional plattforms and formats.

While we're spec'ing out the new A1 prime, can we add mechanical focus and zoom instead of pure servo controls?

I have actually just seen an article that states the BBC will accept EX1 footage as HD, criteria has changed due to the increasing costs in excess baggage etc. Stipulations are that it has to come from a cam with minimum :- 1/2" chips and capture on a nano flash or equiv at 100 mbps.

And at present the EX1 is the only cam at this price point that fulfills that, and probably 90% of people who shoot with the EX1 will never put forward footage for broadcast it sure is a big draw card.

I know of a HD production for Nat. Geographic being shot with a EX3 fitted with a Nanoflash. The BBC's spec allows 1/2" sensors, the EX1 codec was just a tad low, the Nanoflash is a way round this.

The BBC mostly commissions programmes or buys in programmes from other major broadcasters (Series like "The Wire") or co-productions with these broadcasters. Spec programmes are unlikely to get taken up unless there's something unique and compelling about about them.

The BBC has a lot of budget pressures at present, so for the average general program they'll be trying to reduce costs. Although one suggestion would be to cut out some of the huge BBC administrative tail.

Quote "While we're spec'ing out the new A1 prime, can we add mechanical focus and zoom instead of pure servo controls?"

Roger, I'd vote against that. Putting in mechanical focus rather than servos is very expensive and would drive up the cost of the lens - at the moment the low price of kit lenses like those with the EX3 and PMW350 is a big draw, about £1500 vs £8000.
The servo focus seems to work pretty well too, certainly 100% useable - and I'm very fussy!
Steve

There is *some* kind of change involved, as the lens barrel on the new camera
sports some kind of windowed gauge, similar to the type found on a still photo
lens barrel... and on a semi-related note, I've updated the article:

http://www.dvinfo.net/news/canon-adopts-mpeg-2-full-hd-422-file-based-recording-codec.html

From Chris' update:

"Formats are not always tied to specific media; for instance, it’s possible to record an HDV data stream to a Compact Flash memory card. If the card slots are indeed SDHC, then I would say it’s most likely that the format is AVCHD… but if the card slots are something else, such as Compact Flash or (not as likely) Express Media, then the recording format could very easily be something other than AVCHD."


All I can say is that if Canon tries recording 24F HDV to a CF card, I'm going to kill a kitten*.







* I love kittens and would never hurt one, but...

I don't think it'll be HDV.

Frame mode (24F and 30F) might actually carry over to the new camera, but that's a good thing since it's identical to native progressive scan (for all practical purposes, it is the same thing as native progressive scan).

I don't care how they get the progressive image, only how they record them. 24F looks great even though it's tecnnically not progessively originated.

But 24F is a RPITA for Avid users b/c there is something funky about the timecode. Other NLE users report having none to serious problems as well.

I find the window on the lens very curious, on a still lens it has feet/meters but I can't think of the last time I used it to verify focus. In the old days there were different f-stops to show what was in focus, but I doubt they would be on there.

On a video camera I can't think of anyone that wouldn't use the viewfinder/LCD to check focus over those so close together little numbers and pretty much anything you want can be shown right on the LCD.

Things that make you go hmmm...

folks, I can confirm it IS NOT going to be AVCHD.... =)

Here is the UK press release - Enjoy and speculate

Not from Admin: Thanks for posting this, Vincent -- I have changed it to the worldwide press release, linked from DV Info Net:

http://www.dvinfo.net/news/canon-adopts-mpeg-2-full-hd-422-file-based-recording-codec.html

Looks like Peter is going to have to upgrade to MC 4.0.5... :) AMA is the future Peter!

With those specs, all that time waiting will be worth it. I just wish it had a 720 60p mode.

Exciting news to say the least. Canon make good video cameras. The assumption is that this is going into the handycam that they have been showing mock-ups of, but could this be a different, possibly bigger camera?? There is no mention of form factor or sensors etc which is a little odd. You would think they would be saying things like first time in a small camcorder, also they talk about file based but not solid state, could they have licensed the optical disc technology from Sony? I doubt this very much, I expect it will be the small EX style camera that we have all seen mock-ups of, but interesting what isn't revealed in the press release.

Wait until after the 8th and maybe all will be revealed ;-)

A 1/2" sensor with HD 4:2:2 that met the minimum HD broadcasters specs would be of interest to the people currently out shooting broadcast SD on the current HDV cameras.

Canon is working in cooperation with Adobe Systems Incorporated, Apple Inc., Avid Technology, Inc. and Grass Valley to ensure compatibility with major editing and processing software programs widely used within the video imaging industry.

Wow - If this new Canon Codec can be edited like DVCPRO-HD I'd sell my HVX-200 and HPX-170.

Lets hope its not too over-designed...

-C

Carry on at the discussion thread for the new pro video codec:

http://www.dvinfo.net/forum/digital-video-industry-news/472115-canons-new-50mbps-mpeg-2-full-hd-4-2-2-codec.html

See you there,