Or another way to look at it is that with the same lens on, to get the same shot size you'll have to move back from say 10 feet when shooting full frame to 30 feet shooting windowed to 1080, again, more DoF.
Steve

So in other words, you ARE losing your 35mm DoF that a lot of people love and seem to be thinking they'll get from this camera.
Steve

So in other words, you ARE losing your 35mm DoF that a lot of people love and seem to be thinking they'll get from this camera.
Steve

I don't think so, I think that would have been mentioned by this time by Vincent or several other people that have had it hands on. Plus, the video shot is evidence enough that depth of field control is remarkable. Have you downloaded the raw straight from camera "Reverie" clips yet, Steve?

I'm assuming the camera is using the entire sensor area and downsizing the image much in the same way that it uses the entire sensor area even when you take a JPEG image at it's smallest size with the camera.

Also, again I don't think anyone on earth is thinking this $2700 DSLR camera is going to kill or even be equal to a F900 or Varicam!

"I don't think anyone on earth is thinking this $2700 DSLR camera is going to kill or even be equal to a F900 or Varicam" - why not, spec's as good plus you get 35mm DoF apparently, and much cheaper.

If it's reading the whole sensor area then down-scaling surely this has implications for the compression scheme, as it'll be working with vast amounts of data? This is why the RED can do 120 fps in windowed mode and only 30 in 4k mode, 'cos it's about the same amount of data (120 x 2000 x 1000 vs 30 x 4000 x 2000 roughly?)
Steve

"I don't think anyone on earth is thinking this $2700 DSLR camera is going to kill or even be equal to a F900 or Varicam" - why not, spec's as good plus you get 35mm DoF apparently, and much cheaper.

If it's reading the whole sensor area then down-scaling surely this has implications for the compression scheme, as it'll be working with vast amounts of data? This is why the RED can do 120 fps in windowed mode and only 30 in 4k mode, 'cos it's about the same amount of data (120 x 2000 x 1000 vs 30 x 4000 x 2000 roughly?)
Steve

Well, one reason I don't think it will kill the F900 and Varicam is because the pros that can afford to own an F900 or Varicam aren't going to be able to wrap their heads around this little thing. :)

Besides, I think the Red already killed them.

As for scaling down. Digital SLR cameras have been doing it just fine for.... well, ever. They have menu settings to change the size of the photos taken from RAW to large, medium or small JPG. They can even take a RAW + a medium or small JPG. So it shouldn't be an issue. I admit though, I don't know all the technicalities behind it.

The exiting moment will come when Canon throws the stuff in the 5D Mark II into a proper video camera body with proper video camera controls. NAB maybe... hopefully!

I believe that the 5D is ganging neighboring pixels together with simple addition. That keeps the full frame, and reduces the data to the processor/compressor. Unfortunately, it introduces aliasing artifacts. If you downscale to 720p or below, that will help deal with the aliasing to some degree.

Besides, I think the Red already killed them.

In a professional environment, specs and even price take a second seat to reliability, proven workflow and compatibility. Right now, if an organization hires a local shooter in a distant city to shoot on HDCAM or DVPROHD, they pop in the tape and start working as both are mature media. The RED workflow is constantly evolving and still showing growing pains.

I haven't yet heard of anyone shooting with a B4 mount on the their RED, but there are certainly many projects currently being shot on the 2/3" systems that would require the flexibility and speed of a 2/3" zoom lens--even a Super16 lens on the RED in windowed mode would not have the same range.

For feature and narrative style shooting, I would probably want to go with a RED over the other two formats at this point, although I'd want to make sure we had two bodies to insure against issues on set (which, frankly, there are still quite a few being reported with the RED, from overheating to mystery crashes).

Bottom line is that when time=money, the camera gear has to be as bulletproof as possible and also have the appropropriate form factor/ergonomics etc. to get the job done efficiently.

several) straight away, but if it's not then it can't be considered a top-notch video camera. If it's as good as an XL-H1 then it may be considered a pro-sumer quality video camera, but I'm sure it won't even be close to that. How could it be, and just think how many XL sales Canon would be robbing themselves of.
Steve

You sure are sure of a lot of things.

I'm a full time producer using both Canon and JVC HD and both Letus and P&S adapters.

I've seen the output of the 5DII in a pre-production model as have thousands of other photographers, and film makers and videographers and every one else who's seen the straight footage from Reverie (that I've heard from) was blown away by the quality and, in fact, several HD shooters compared it directly to top HD output.

You're really making a lot of noise with no visual reason to do so - just supposition.

There's very good reason to expect that the 5DII output could exceed HDV output.

THe DOF and FOV make it superior as a film making tool to any current HD/adapter system.

I think we understand that you doubt it - that really doesn't mean anything to people who've already seen it's early quality.

So what's your point?

I'm not making a lot of noise, just a little bit - putting forward my thoughts and opinions, kind of the point of a forum, no?
I've no axe to grind, it's not a camera that I'd ever buy, nor is shallow DoF, drama work etc., what I do, I'm just surprised at how much interest this seems to be generating, and really would be shocked to find out that it rivals even Z1, XL-H1 etc., let alone top-level cameras.
Steve

I'm just surprised at how much interest this seems to be generating, and really would be shocked to find out that it rivals even Z1, XL-H1 etc., let alone top-level cameras.

I guess it depends on what you mean by 'rivals' - in terms of sheer image quality you just need to watch Laforet's behind-the-scenes of Reverie to see back-to-back shots of the XHA1 and the 5DmkII under the same lighting conditions... based on that I wouldn't use the term 'rivals' as it tends to imply that they are close when in fact the gap between the two is huge (in the 5D's favor).

If you are referring to ergonomics and standard features on pro video cameras then it's a different story - the 5D doesn't really rival the cameras you mentioned as it's lacking several important features.

The question is really whether the gains in image quality are sufficient to outweigh the inconvenience of missing pro features, and that depends a lot on the type of video you shoot. I think the biggest market will be those who are currently using 35mm lens adapters on various cameras. For those using them on things like the HV20 it's a no-brainer, as you're already working under the same limitations as the 5D in terms of controls... if you're using an adapter on something else (HVX200, EX1, etc) losing some of those controls may be worth the tradeoff to eliminate the cumbersome adapter and gain significant low-light sensitivity.

if you look as his lenses used you will see that there is about $20,000 of lenses there....

"
EF Lenses used in the making of REVERIE:
FD 7.5mm f/5.6 (converted to EF mount)
EF 16-35mm f/2.8L II USM
EF 24-70mm f/2.8L USM
EF 50mm f/1.2L USM
EF 85mm f/1.2L II USM
EF 135mm f/2L USM
EF 200mm f/1.8L USM
EF 400mm f/2.8L IS USM
EF 500mm f/4L IS USM
TS-E 24mm f/3.5L
TS-E 45mm f/2.8
"

Expect to get caught up in the lens addiction with this thing.... I know when I upped to the 1ds3, I had to replace most of my glass with "L" glass to catch up to the 21MP.

Maybe at 1080 rez it won't matter so much.... ?? but fast glass ain't cheap.

I talked to a photographer who shoots with an early version 5D (not sure which one), and he said you really need "L" series lenses to get the full benefit of the 21.1 mpixel sensor on the Mark II.

Those aren't cheap. The difference in a 50mm L and regular 50mm is $1,200.00

Can anyone confirm whether you need that level of quality in glass.

L Series glass is amazingly nice, but in no way necessary to achieve great photos. The L Series is about more than just higher quality glass. They have better insides, better focusing, tougher build, ect. However, Canon makes great mid range glass that looks really nice.

I've read in more than a couple reviews that people have mentioned that Canon's 50mm f/1.8 lens that costs $79.95 tested as taking more detailed photographs than their 24-70mm f/2.8 L series lens set at 50mm. Granted it's a plastic lens and doesn't focus as fast. Reading that made me go out and get one. Here's a photo from a shoot I just did a couple weeks ago with it.

http://www.sybilludingtonmovie.com/images/cast/micah1.jpg

My point is, no, you in no way need to spend $20,000 on L glass to make great pictures. However, the nice thing to know is that when you do drop a couple grand on a lens, it's for life as you'll use it for years and years.

Can anyone confirm whether you need that level of quality in glass.

You do to impress other photographers David, but you most certainly don't to impress clients. For that you need skill, imagination, experience and the ability to get the job done, on time and within budget.

tom.

That's the answer I has hoping for.

Thanks

And you definitely don't need L-Series glass for video. 1080p is roughly 2MP.

While I am probably not going to buy the 5D, I am interested to see how the innovation filters down into Canon's video line. It would be great to see them improve upon the chips, work on some of the issues such as rolling shutter, and put it in a video camera body. That is what really got me excited when I heard about the new 5D (and the D90). It will be exciting to see what happens the next year or so.

Nicely put ;) Actually, the whole resolution is king argument is as old as Ansel Adams Vs The lyrical impressionist photogs of his time.
I've written about it ad naseum.

Based on some of the not so great art I've seen made with reeelly Hi Def Cams, I can beat many with my cellphone cam set on 'Video'.

The best gear just gives the best production an edge for the BIG screen.



You do to impress other photographers David, but you most certainly don't to impress clients. For that you need skill, imagination, experience and the ability to get the job done, on time and within budget.

tom.

Don't forget you can rent these lenses too. My local shop runs from $15-$100 a day for L series glass, with the vast majority of lenses running about $30/day. Friday afternoon to monday morning is a 1 day charge, and a week is a 3 day charge, so you're not looking at a lot of money to pick up 2 or 3 good lenses for a shoot.

I ordered the kit lens as a general purpose walking around lens, and I'll probably pick up the 50mm 1.4, but beyond that I expect to just rent whatever I need for each specific shoot.

You certainly won't have to use L lenses for good video on the 5D MKII, but they can help. L lenses typically have outstanding contrast, great color transmission, great bokeh, low distortion, and minimal CA...things that will show up in either HD or DSLR resolutions.

Pat

well i don't if they would show up I mean hd is only 2mp while those lenses are designed to do 20 something mp.

Does anybody know how the whole 24p thing is going with canon?

You certainly won't have to use L lenses for good video on the 5D MKII, but they can help. L lenses typically have outstanding contrast, great color transmission, great bokeh, low distortion, and minimal CA...things that will show up in either HD or DSLR resolutions.

Pat

Definitely not. In fact we did early tests comparing Zeiss Primes vs Nikon primes using the P&S mini35 setup on the Canon XL and found no difference in image quality. In fact the 85mm 1.4 for about $900 is built about as well and feels as good manually as the Zeiss 85mm we tested with.

I'm sure the same will be true of Canon glass.

Definitely not. In fact we did early tests comparing Zeiss Primes vs Nikon primes using the P&S mini35 setup on the Canon XL and found no difference in image quality. In fact the 85mm 1.4 for about $900 is built about as well and feels as good manually as the Zeiss 85mm we tested with.

I'm sure the same will be true of Canon glass.


In all fairness to Zeiss glass, that test is a bit moot when stepping to HD resolutions, even with the increase in sensor size.

In all fairness to Zeiss glass, that test is a bit moot when stepping to HD resolutions, even with the increase in sensor size.

Why is it moot? I Don't get your point.

GLASS
I believe that Jed was saying it was moot because if you were doing xl-1/s/2 then you were seeing the result in Standard Def so it was only 1/4 the resolution of HD resolution. If you look at it on an HD sensor the Nikon may not hold up as well as the zeiss lenses. The other thing to consider is that with the 5d2 you're not recording a piece of ground glass (that may hide the imperfections in a lens) you're recording straight to a sensor which may not be as forgiving.

Regardless though it seems silly to have a discussion over which lenses you "can" and "can't" use for a camera, most people I know who work will use whatever lenses they can get their hands on that will get the job done regardless of what anyone types (or hypes) on the internet. It's like the DP's who swear by Cooke lenses and won't use Zeiss master primes, they're all tools.

DOF
If it helps I could have sworn that I read that the sensor masks down to 16:9 resolution by greying out the area that won't be recorded to the card. I guess it serves as kind of a "safe zone" again, no one will know for sure until the camera come out. So it seems like you are using the full sensor from left to right but have some masking on the top and bottom, which I believe preserves the "35mm DOF"

ROLLING SHUTTER
Also I saw some of the rolling shutter "issues." Seriously? A guy shakes the camera like it's an earthquake and the image gets distorted? And 4000 people who will never shoot a movie in there life hop on the internet and complain about how that's unacceptable? What are you shooting where that kind of operating is expected? Earthquake 3? I mean come on, why would you even point out that if you're not very good at operating a camera that you shouldn't use this one? Seems like more of a "idiot resistant" feature then an issue. Show me the image getting screwed up during normal or even extreme (but not ridiculous) operation and I'll care a lot more. (or slow my pans and tilts down!)

F900/Varicam killer?
Seriously anyone that is proposing that this will Cannabalize the market of a varicam or f900 has not used one of these cameras, believe it or not when you buy a 100k camera you're getting more then a sensor, If they add multiple frame rates, an HD-SDI output, Y PR PB support, inter operability with a professional battery system, or a fiber optic system, then MAYBE MAYBE there's a small concern but come on, how many people do you know that bought an f900 to shoot something that needed 35mm DOF?

Glass
An XL2 with its tiny sensors would only use the center of the image circle. It actually needs much higher res glass than the mkII. When you guys talk about Zeiss, you mean their dslr lenses, right? The zeiss 85 is certainly not provably superior to the Nikon or Canon 85s. It has a somewhat different look. If you talking about the very expensive Zeiss cine lenses, I expect these would be somewhat sharper at the corners.

30p
The way Canon works, the software and hardware in the 5DII is capable of doing different frame rates. But the 5DII only does 30p so that a higher end dslr can be differentiated with more frame rates. They are not likely to make a firmware change for 24/25p.

Rolling Shutter
Looks good to me. Canon's done some work here.

High end competition
It's not about the 5DII, its about this technology going into high-end products. It seems to me Canon will be a new player in this area. The 5DII sensor isn't even a video specific design. What will a video specific cmos design look like? That would make me nervous if I was a high-end video equipment producer.
Remember that the 5DII sensor is designed for bayer processing. Video CMOS pulls all three color off the sensor - no interpolation. (Red: 12M/3 colors = 4M)

Don when they put the lens on a mini 35 adapter it's using the full lens (not just the center) when they just use like the ef adapter though you're absolutely right, but I don't think that's the test JIM and JED were talking about.

Hi, i was wondering if nikon's primes would work well (as well as L series zoom..) using the 5d m2 video mode ? i already have a 5d with 24-105 L and 70-200 L series but before that, i was an addict of my old unbreakable Nikon Fm2 + a set of old nikon primes (no autofocus ones) - Still have them, sleeping, but ready to rebirth.

I've seen a few adaptators (nikon to canon ring) that seems to keep the usuals electronics fonctions (light meter, auto focusing ?). For my use, focus will be manual
with redrock follow focus so i don't mind autofocus but i mind a little light meter (could be usefull even if i use an independant light meter) last and most important : Price...and old nikon's primes are really great and quite cheap. Would it work in vidéo 1080 P with the 5d mark 2 ? i'm not talking about "style" here but more about final image result.
I know no one have been doing test yet about this, but is there a particular reason why the video mode would make impossible the use of nikon lens ?

THKS - AG

...
Remember that the 5DII sensor is designed for bayer processing. Video CMOS pulls all three color off the sensor - no interpolation. (Red: 12M/3 colors = 4M)Don, I don't follow you here. The Red sensor is indeed a Bayer sensor. So it uses a color mask. The demosaicing/debayering process used on Red footage has to interpolate color from adjacent pixels.

Don, I don't follow you here. The Red sensor is indeed a Bayer sensor. So it uses a color mask. The demosaicing/debayering process used on Red footage has to interpolate color from adjacent pixels.

It's a mysterium to me.

It's a mysterium to me.

WINNER: DVinfo.net pun prize of the day!

So how does the math work on these cameras?

the 12mp Red sensor only needs about 8 mp for a 4K image. The 5DII apparently uses a full size 16:9 crop of a 21 mp sensor. That ~6 photosites per output pixel. Or perhaps they read alternate lines and it's ~3 photosites per output pixel.

Red has the future possibility of using all the phorosites for 2K, instead of a crop. This change could/should reduce noise.

I don't believe the lower noise DSLRs are using "classic" bayer algorithms anymore. I think when constructing a pixel the algorithms look at more surrounding sata to determine if a value is noisy.

You should see this: October 19, 2008 on Vimeo (http://www.vimeo.com/2053280)

In response to "You should see this: October 19, 2008 on Vimeo"

comment: That footage looks really beautiful!

question: This was filmed in the UK, and a few posters have mentioned that the 50hz lighting in PAL land would flicker when shot with 30P. I didn't see any of that in the praticals such as the street lights. I've never used NTSC in PAL land or vice versa so i really have no experience with light flickering.

So my question is where is the flicker???

If there's no flicker, this camera is fine for us living in PAL land! Soon i'll be shooting a lot of nighttime scenes for a TV show here in PAL land (china) and the thought of using HDV with its nighttime noise makes me cringe and could be a deal breaker when it comes to finding a distributor to sell the show!

A PS question: Film camera operates at 24 frames per sec and they use practicals all the time to supplement film lights, don't they? Say for example a household table lamp in a living room setup. I know they might replace the bulb in the fixtures with a higher wattage but still are they not running at a normal 60/50hz - wouldn't we see the flicker all the time in motion pictures? Do the gennys they use run at 24/48hz? AND if so, does the genny run all the power to a location, even to the practicals?

I really hope the answer is that it DOESN'T matter!!

perplexed
mike

I liked the images themselves but if I saw one more rack-focus I might have pitched the monitor off the desk. I'm not really sure why camera tests have to appear like music videos these days--the idea is to show off the imagery, so what's the point of quick cutting? Sorry to be negative here but I want to see what the camera can do and was frustrated and distracted by the way this was shot and edited.

The all-mighty power of the rack focus. At least they aren't showing you crash zooms.

At least it was mostly shot on a tripod. I have looked at so much HMC 150 footage recently that was hand held and of the family/pets I got motion sickness.

Charles, I think this link actually IS a music video the vimeo poster just forgot to encode the audio. (He mentions it further down the page) However I could only sit through about 3 minutes before I felt like I was looking at really well shot vacation footage, and scrubbed through to the end.

I think this thing is going to be the closest I get to what I ask for in my signature so I can't wait for it to be released to us plebeians!

Yeah, I saw the bit about the music. I'm all for making camera tests more watcheable--I made a short film to demo the EX1 and Redrock M2 that was hopefully as entertaining as it was demonstrative--but this became more about the cutting and the rack focuses and less about seeing what the camera could do. My feeling would have been to pick 1/3 of the better shots and just let them play.

You should see this: October 19, 2008 on Vimeo (http://www.vimeo.com/2053280)

It's hard to tell much about the camera with this video. Every shot is a rack focus and just as soon as the focus goes onto the main subject of the frame... the shot is cut. I would have loved to have seen longer, in focus shots as the low light once again looks great!

the main problem in PAL countries is not the 50hz flicker, which may or may not appear depending on the light conditions, well all we know about murphy's laws.

the real problem in PAL, is that 30p cannot be converted to 25p (or 24p) in a professional way. 30p footage cannot be used for PAL TV, HD or SD, and cannot be used for even create PAL DVD.

so in PAL contries , if canon won't relaease a 25p firmware forget to use the 5DII for anything else than home videos.

Is it just me, or is the latitude absolutely rubbish on the 5DII? All the test clips I've seen so far have crushed blacks and blown highlights that smack of a phone camera. Brilliant DOF, etc etc etc. Clarity is great. But I'm getting a bit nervous about what I've seen so far. (and I have recently sold my existing camera to replace it with a 5D2 or HVX, and now the $AUD has dropped so far the HVX has dissapeared from my budget constraints!)

Oliver I don't know exactly what you're talking about in the few clips that I've seen the exposure lattitude has been on par or better then most HD Cameras I've worked with. It may not reflect the true abilities of the still portion of the camera but it looks about right to me (especially with the web encoding thats been done) as far as the video clips.

I'm getting a bit nervous about what I've seen so far.

I do hope you're basing these conclusion on observation of the camera's output and not on footage crushed out of all recognition for web streaming Oliver.

Of course you are, you wouldn't have sold your current kit to buy a 30p camera for a 25p county unless you'd made careful analysis of the image and audio quality and deemed them up to scratch.

If at all possible, avoid being an early adopter.

tom.

the main problem in PAL countries is not the 50hz flicker, which may or may not appear depending on the light conditions, well all we know about murphy's laws.

the real problem in PAL, is that 30p cannot be converted to 25p (or 24p) in a professional way. 30p footage cannot be used for PAL TV, HD or SD, and cannot be used for even create PAL DVD.

so in PAL contries , if canon won't relaease a 25p firmware forget to use the 5DII for anything else than home videos.

I fully agree. 25p is (24p might be) mandatory.

You can convert 30p to PAL... Just keep it interlaced (50i). Looks perfectly smooth if you do...

Thomas, funnily that's the first time I've heard anyone say that. What software/hardware would you use to convert? I would have thought getting an interlaced 30p to 50i was pretty difficult.

Is it just me, or is the latitude absolutely rubbish on the 5DII?

It's not just you. I've noticed the same thing in the original camera .MOV files. I hope that it can be remedied with different in-camera processing settings. It has the same sensor as the 1Ds Mark III, and I know the dynamic range on that camera is at least 13 stops when downsampled to 1080p, so ISO 1600 should have at least 9 usable stops. But if the 5D Mark II doesn't sample every pixel, I can see how it would lose latitude fast.

What bothers me much more than the poor dynamic range, though, is the aliasing (jaggies, moire). My only hope there is that Gaussian blur and resampling to 720p will remove the artifacts (that will help with the compression artifacts and noise, too).

I prefer After Effects to convert 30p to 50i. Compressor can do this as well, but you must up the preset conversion settings to best (it defaults to good), which is slow unless you have a fast machine. After Effects is fast, and looks very smooth.

When shooting, I personally prefer 30p over 24p/25p. However, Canon should include all 3.

I can't imagine it not being able to do this, especially if sold in PAL countries.

Thomas, using AE or Compressor is supposed to give variable results for fast motion for 30p to 50i. Are you saying this is not true? have you got any samples online to see?

Also the 5DmkII is only 30p in all countries, this has been verified by Canon

Best

Dan

I recently converted one of my video projects from 30p to 50i using AE... It worked out great for me... Even fast motion looked identical to the 30p source, when viewed on a native PAL DVD player/PAL TV. I don't have any online samples. Be aware that any online samples would be 30p or 25p playback... You wouldn't see the 50i output, unless viewed on a PAL TV or Monitor. When I distribute for web playback, I always keep it 30p. All computers (worldwide) will playback 30p video files, as well as NTSC 30p DVDs (all regions). You only need to convert to 50i if distributing on PAL DVDs or for PAL Broadcast, where it will be viewed on a PAL TV.

The reason 30p converts to PAL 50i, is that you're actually adding frames (fields)...

30p (30 pictures/frames per second) - PAL 50i (50 pictures/fields per second).


*Canon sure made a mistake by not offer 25p PAL option...*