Season Finale of "House" shot on Canon 5DII

[TMARK]

Graeme,

I haven't seen high iso M-X footage live up close.  I will see it first hand in a week or two.

I agree on the DSLR very high ISO clips looking ugly, but it is impressive.  Berry Lyndon looks much better, and John Alcot lit that with candles, and its just beautiful.  

You'd be surprised..... If you shoot 5219 500T with same lens / T-stop as a RED M-X and compare the outputs. (I'd assume that from Arri's numbers that their new camera would also look very good on such a comparison)

Although the 5D2 is "good" at high ISO, the M-X can look very good at high ISO ( http://www.reduser.net/forum/showthread.php?t=44171 ). The other issue with ISO on the DSLR's which showed up in a test we saw at the RED Studios, was that ISO on the DSLRs is done via analogue gain, which essentially means a stop of DR is lost off the highlights for each stop of analogue gain applied. So by the time you get up to a very high ISO, you might have acceptable noise levels, but you may not have enough DR, and if the highlight range drops enough to touch lit skintones, it can, as we saw, look very ugly.

Graeme

[bcooter]

And Zacuto's tests have shown that dSLRs are already very close to 35mm cine film in IQ, and surpass in some areas - the next generation will be even better. Dedicated motion cameras coming out in the near future will level the playing field further.

This is all just conjecture on my part, but I have some doubts that Canon and/or  Nikon are going to roll out a raw shooting full super cine motion camera in 6 months to crush RED or anything.

Canon has moved glacier slow in video for a long time.  After all it took them 3 years to come out with a solid state cards for video cameras and it's not like the indie and commercial guys have not been begging for a larger frame rate for the xl series for almost a decade and I mean a million voices screaming, not just a mention here or there.  There were video blogs like DV info Net and 2-pop talking about this stuff almost 10 years ago and in regards to Canon either they just don't read or just didn't pay much attention to it.

The fact that Canon was surprised of the Video popularity of the 5d2 IS surprising considering how long a larger format motion camera has been requested.

In the still area Nikon took forever to make a full frame 35mm camera, later still on a high megapixel one, later still on new tilt shifts, so though Nikon has no territory to protect in video, it took them 7 years to make an equal to the original 1ds.   When I've talked to Nikon they seem a lot more interested in Sports photography than anything else.

I'd love to see it, a $7,000 RED equal with true autofocus and precise manual focus lenses, but I have my doubts it will be anytime real soon, though I hope I'm wrong.

But price controlling a project.  Price controls every project, everywhere in the world, even a 200m blockbuster.  You don't have to spend much time in LA to hear the stories of who got paid what, so the idea that Hollywood will look at any way to cut costs is not going to be new news.

In fact if you knew how some Hollywood producers/directors tipped on a meal you'd laugh, unless you were the service staff so if anyone thinks the big boys in the movie business don't question every expense, whew, they make Blue Cross look like spendthrifts.

Panasonic and Sony, there in their own world and if the Sony response to red was the CineAlta then somebody got something wrong when they figured out the price point, or locked their calender to 2006  I mean Sony makes a still camera with no live view, no video and if anyone should be able to figure that one out you would think it would be them.

But as far as costs, in the cinema world the RED is cheap, well at least cheap compared to anything that shoots film or has a PL mount.

Canon/Nikon don't concern me because they always seem to do what they want, when they want to and though they may be listening, I do wonder why it took them so long to even get where the 5d/7d are today, which is not the best form factor for motion.

The only thing that concerns me about RED, other than the slowness of delivery, is that they've already changed form factors, from the RED One to the Epic and I don't read every word on the RED site, I have to admit it's somewhat confusing.  How much more an Epic costs, how many parts are interchangeable I don't know, but I do know none of us want even modest investments in equipment to be phased out not in today's world where an artist really has to be dead sure of where they spend every dollar.

Actually what really concerns me about RED or Arri, or even Panasonic is while they are in the kitchen cooking, the Canon 5d2's are out in the dining room partying like a bunch of drunk sailors.

Regardless of all this Camera talk, I have this belief that if an image is beautiful, or a story is compelling almost any camera will do and though I own way too many cameras and I don't believe the camera makes the shoot, everyone in my studio that's looked at the RED footage compared to the 5d2 is blown away and not blown away with a hmmm, the RED looks a little better than the Canon, they're blown away with a "Holy Shit,  I didn't know you could do that to any file, much less video".

BC

[Graeme Nattress]

Yes, smaller and lighter form factor and new modules, but gear like EVF and LCD still work if you have them from the RED One, and a simple adaptor will get the batteries running too.

The issue with others doing as we have done basically all comes down to (and it's more nuanced than this, but this is the key) sensor speed. You need a very fast (in electronic readout terms) sensor to do high resolution and low skew, and that incidentally leads to the ability to do high fps if the rest of the electronics can keep up. That is why we're advertising 5k resolution at 120fps. The best DSLR does around 10fps full sensor RAW at the moment - that is why they use line-skipping tricks and only record "1080p", and they don't have the sheer processing power needed to treat the raw image with complete respect while realtime downsampling to 1080p. So, to do this "properly" means significant issues in sensor design and the speed / power of accompanying electronics.

Graeme

[TMARK]

I think that Red's major advantage is vision and a small, forward thinking organization.  Of course Sony, Canon, panasonic, Nikon et al could do what you guys are doing, but they are compartmentalized burocracies that compete with other departments within their own organization and with their traditional external competitors.  These companies do not make markets, unless its accidental like the 5d2.  

Yes, smaller and lighter form factor and new modules, but gear like EVF and LCD still work if you have them from the RED One, and a simple adaptor will get the batteries running too.

The issue with others doing as we have done basically all comes down to (and it's more nuanced than this, but this is the key) sensor speed. You need a very fast (in electronic readout terms) sensor to do high resolution and low skew, and that incidentally leads to the ability to do high fps if the rest of the electronics can keep up. That is why we're advertising 5k resolution at 120fps. The best DSLR does around 10fps full sensor RAW at the moment - that is why they use line-skipping tricks and only record "1080p", and they don't have the sheer processing power needed to treat the raw image with complete respect while realtime downsampling to 1080p. So, to do this "properly" means significant issues in sensor design and the speed / power of accompanying electronics.

Graeme

[feppe]

Your second para is msleading.  The Zacuto tests were interesting, and played to the dslr strengths of high iso.  And that is awesome, but there is no way that the dslr files (or Red or panasonic or Arri digital) are as good as a color graded digital intermediary from Pacific Film, when shot at normal, say up to 800, iso.

Did you read the second paragraph? I didn't say the things you imply I said.

It was revealing to see seasoned pros being totally blown away by the dSLR footage. There are obvious and well-known shortcomings in IQ, but as I pointed out, take into account this is the first generation of dSLRs. RAW video and cameras which don't rely on decimation or other jury rigging for motion are right around the corner.

Film will be dead soon for motion, just like it already is in stills.

(No flaming on the last sentence, please; I still shoot film, also)

[Graeme Nattress]

You make a very good point. It's very expensive to do things right, but you get a lovely end result. The business model for that is not the sub $1000 consumer market where C/S/N live. However, we believe that there is a market for those who want exceptionally well specified gear, so that's what we're making.

Graeme

[bcooter]

However, we believe that there is a market for those who want exceptionally well specified gear, so that's what we're making.

Graeme

Graeme,

Hold that thought.

Good or bad, I try to think like client's think.  

Part of it is how much they pay, do they notice the difference, do I notice the difference and can I deliver something that is hopefully more than they anticipated.

I don't care about the $1,000 market cause that's not where the professional world lives (or shouldn't), but at 50 large  I start to care, at 50 large that I gotta wait on or change out in a year, I care a lot.

(not implying anything here, just my thoughts).

Anyway,

I did download the sample files and yanked them around thinking about real world clients and how they react.

You know where they say can you make the car snap and the background more subtle.  (though I don't shoot cars).

The mercedes shot in the Santa Monica Mountains, is the one I messed with most, because it was shot it deep shade and honestly I was amazed at how well it held up pushing it to 1200 iso.

I then took it out in prorezz 422 and stuck it in CS4 extended, did adjustment layers, burned in the edges, contrast etc. and though that shot wasn't typical 35 person film crews and obviously my post abilities were limited using cs4, it looked amazing to me, especially compared to any footage I've ever worked.

Without a doubt you guys have the market today, I just hope you get stuff out quick,  quick and easy for the buyer and I'm serious about the buying quick and easy part cause the world isn't gonna stand still waiting for anything.

I can't go anywhere there is not a 5d2 on set and you can't walk past any still rental stage in NY or LA without seeing some hand written sign that says, quiet please, video in progress.  I have no doubt the the photographers who are now shooting motion wouldn't like a better file, or something that gives them an edge, but I also know that since 95% of them barely own anything in the studio other than an I phone and a 5d2, they're gonna hold still (pun intended) and see what shakes out.

I did a gig a few weeks ago that we shot the 5d2 and the agency brought in their in house guy with a RED to shoot one dialog segment.  The RED worked fine (old sensor version), but he only ran it for about 10 minutes.  The rest of the day he walked around with a 5d2 and a zaguto finder shooting everything but catering.

I didn't talk to him much, other than to ask why he didn't shoot everything with the RED and since he had it kitted out with 400lbs of stuff and was using window light for all his "B" roll, he said the 5d2 was just easier.

So if your going to sell the Epic to movie guys that are use to renting lenses that cost $20,000 or video guys that were buying $90,000 engs, then your fine, but if your going for the still world I would strongly suggest to dip your toes in the still studios and listen to  a still photographer talk about the business as it is today.

BC

[TMARK]

Did you read the second paragraph? I didn't say the things you imply I said.

It was revealing to see seasoned pros being totally blown away by the dSLR footage. There are obvious and well-known shortcomings in IQ, but as I pointed out, take into account this is the first generation of dSLRs. RAW video and cameras which don't rely on decimation or other jury rigging for motion are right around the corner.

Film will be dead soon for motion, just like it already is in stills.

(No flaming on the last sentence, please; I still shoot film, also)

I was reacting to where you said the IQ of the dslrs is very close to 35cine.  Its not.  I also liked the Zacuto test, and my take away was that for doc projects, or really, really low light, they are the way to go.  

And I agree, the next gen of video dslr should be really nice.

Film won't soon be dead for motion.  This is from someone who shoots almost all motion projects on a Red.  For commercials, videos, sure, digital is fine for the same reasons digital is better for clients in a commercial context, but why shoot digital on a feature film?  There is no client on set.  The director is the closest thing to a client, and the director usually works very closely with the DP on the look of a film.  So the main benefit of digital, instant playback, is not necessary or even desired.  Except in low light, 35cine looks better, so where is the push for digital in features?

[TMARK]

BC,

The client's guy with the Red shooting a 5d2 is where the Scarlet should come in, right?  Small and spontanious, easy, less obstacles.  The Red and 5d2 really shouldn't be compared in some ways, because the 5D2 is really a documentary 35mm slr, which is great, while the Red is a cinema camera.  You can break the Red down into a smaller unit, but never as handy as a DSLR.

[feppe]

Except in low light, 35cine looks better, so where is the push for digital in features?

Right here. And here. And here. And here.

Seriously.

I've seen several of those movies, and haven't heard anyone complain about the IQ. Granted, they are all laymen, by they are the ones paying the bills. In fairness, I felt cheated after seeing Miami Vice with its distinctly video look. I don't care it was based on a TV series, it it still looked cheap - and I'm a huge Michael Mann fan.

And I thought the main benefit of digital is low cost. And that's always in demand.

[BJL]

Dedicated motion cameras coming out in the near future will level the playing field further.

My guess is that there is great potential for combining the cost efficiencies of using a relatively high volume still camera sensor instead of a dedicated video sensor, in a body designed for motion photography. Like the $6,000 4/3" format Panasonic camcorder using a modified GH1 sensor. Modifications could be made to the sensor "toppings" to improve performance over a still/digital combo-cam:
1. Adapt the optical low pass filter to video resolution (so too much blurring for still images), to reduce aliasing.
2. If line skipping is needed to get adequate read rates, modify the Bayer CFA so that every second line can be read instead of every third:
GRGR
XXXX
BGBG
XXXX
with "X" for lines not read. This improves resolution, light gathering, and aliasing compared to reading every third line, as all DSLRs seem to do so far.
3. Elongate the micro-lenses to have them spread out over part of the unread lines above and below each read line, to divert light from the unread lines into the read lines and so improve sensitivity.

Panasonic seems to be doing (1), but I would be pleasantly surprised if it does either of the others.

[BJL]

Right here. And here. And here. And here.

What surprises me is how much of that is not just digital, but also 2/3" format rather than any of the various cine-35mm format digital options that have been around for several years. Even Avatar apparently, where the budget would seem enough to have used a larger format.

[TMARK]

Feppe, I know films get shot digitaly. I'm saying, where is the push coming from? I see a push coming from ad agencies and small production companies like mine, shooting commercials, music videos, in store POS displays, etc. There is no push to shoot digitaly in features, unless the director and DP want to.

That's cool if you disagree, and I know you cited to Wikipedia and all, but from where I sit, there is no burning desire by anyone making features to give up 35 film.

[mtomalty]

Gotta imagine these guys have some industry cred ?


http://www.youtube.com/watch?v=H4xqrXcuTU0

http://www.youtube.com/watch?v=xJqs1-dFEXE


Mark

[feppe]

Feppe, I know films get shot digitaly. I'm saying, where is the push coming from? I see a push coming from ad agencies and small production companies like mine, shooting commercials, music videos, in store POS displays, etc. There is no push to shoot digitaly in features, unless the director and DP want to.

That's cool if you disagree, and I know you cited to Wikipedia and all, but from where I sit, there is no burning desire by anyone making features to give up 35 film.

Oh noes, I had a wikipedia link to a list of movies shot digitally, that really killed the credibility. How will I ever get over the shame and ridicule?! Why does almost every disagreement on this forum have to spiral into a mudslinging contest?

Look, shooting features is a business, and although I'm not in that business as a bean counter anymore, the push is most certainly coming from the bean counters because it's cheaper and faster, and the audience doesn't care as long as they're entertained. As Mark's links and the House interview prove there are creative pros who push as well due to the unique capabilities of digital.

You can disagree all you want, but digital is here to stay, and its market share will grow faster and faster in all applications, including features.

[Craig Lamson]

..... but from where I sit, there is no burning desire by anyone making features to give up 35 film.

Was there a real burning desire for still shooters to go digital a few years back?  As much as I embraced it back then, I really felt forced by market conditions and trying really hard to be future proof.  Does the same apply now to motion?

[TMARK]

Was there a real burning desire for still shooters to go digital a few years back?  As much as I embraced it back then, I really felt forced by market conditions and trying really hard to be future proof.  Does the same apply now to motion?

I felt pushed by clients and competition to shoot digital. For personal and editorial I still shoot film. But as you know, in a commercial context film is just gone for 99.9% of projects.

[TMARK]

I didn't mean to insult you. Seriously. I ribbed you on wikipedia because, you know, it's wikipedia but I didn't mean anything by it.

Oh noes, I had a wikipedia link to a list of movies shot digitally, that really killed the credibility. How will I ever get over the shame and ridicule?! Why does almost every disagreement on this forum have to spiral into a mudslinging contest?

Look, shooting features is a business, and although I'm not in that business as a bean counter anymore, the push is most certainly coming from the bean counters because it's cheaper and faster, and the audience doesn't care as long as they're entertained. As Mark's links and the House interview prove there are creative pros who push as well due to the unique capabilities of digital.

You can disagree all you want, but digital is here to stay, and its market share will grow faster and faster in all applications, including features.

[feppe]

I didn't mean to insult you. Seriously. I ribbed you on wikipedia because, you know, it's wikipedia but I didn't mean anything by it.

Fair enough. Ridiculing wikipedia is sooo 2009  Besides, it's just a list of movies so it should be a bit more valid than some wildly editorialized commentary on the auteur theory.

[Graeme Nattress]

My guess is that there is great potential for combining the cost efficiencies of using a relatively high volume still camera sensor instead of a dedicated video sensor, in a body designed for motion photography. Like the $6,000 4/3" format Panasonic camcorder using a modified GH1 sensor. Modifications could be made to the sensor "toppings" to improve performance over a still/digital combo-cam:
1. Adapt the optical low pass filter to video resolution (so too much blurring for still images), to reduce aliasing.
2. If line skipping is needed to get adequate read rates, modify the Bayer CFA so that every second line can be read instead of every third:
GRGR
XXXX
BGBG
XXXX
with "X" for lines not read. This improves resolution, light gathering, and aliasing compared to reading every third line, as all DSLRs seem to do so far.
3. Elongate the micro-lenses to have them spread out over part of the unread lines above and below each read line, to divert light from the unread lines into the read lines and so improve sensitivity.

Panasonic seems to be doing (1), but I would be pleasantly surprised if it does either of the others.

When you line skip, it's like having a very low fill factor in that direction, and that actually increases the sensor part of the MTF, making aliasing stronger, meaning you don't just need an OLPF set to the new pixel pitch, but a stronger one even still. That means you're either going to be very blurry, or still have significant aliasing. Instead of line skipping, the solution is larger pixels. However, because of the losses of proper optical low pass filtering, you're going to get a soft 1080p if that's all the pixels you have, even though they're now larger. Arguably, the result is better than the line skipping, but it's still hardly optimum.

Who really know what Panasonic is trying - they did say in their presentation that it was the ND filter that removed the aliasing, but Jan from Panasonic corrected that gaff on the DVXUser forums.

In the end, there is no easy shortcut to the best of image quality.

Graeme