Jump to content

Star Wars 9 to be shot on 65mm film


Recommended Posts

  • Premium Member

I think you're somewhat confused.

No confusion...

 

Single CCD's aren't even on the radar/even worth discussing.

 

Three CCD's are impossible to use with industry standard cinema glass that isn't specifically made for that design.

 

With the amount of "new" lenses on the market today, for multiple mount types, there is zero reason a large-imager 3 CCD camera couldn't have been made and specialty glass that goes along with it. Panavision did it, but even their attempt didn't work well enough. The whole concept that lenses dictated the imaging technology we use today in my eyes is semi-bogus.

 

We use CMOS for dozens of reasons as pointed out above, but also because it's A LOT CHEAPER AND SMALLER then a 3 CCD block. My "argument" above is that CCD's as good as they are, unfortunately aren't as good as CMOS in many ways and even CMOS isn't really that great.

 

To this day, we still haven't figured out a decent digital imaging capture and presentation system. What we have is good, but it's not great and requires far to much manipulation with both the origination files and in the presentation.

Link to comment
Share on other sites

I get the feeling that a couple of you guys would rather 100% of movies were still shot on film (correct me if i'm wrong).

If a DoP choses to shoot on video for some reason (I can't imagine anything other than light sensitivity though) - no problem. But much more often it's a producers decision. And it sucks, because a) it shows hagglers are more and more involved in technica/artistic things which inevitably impacts quality and film is at risk of disappearing because some ignorant hucksters who've been only laundering money before somehow consider film to be obsolete.

 

You're confusing yourself with the notion of "different tools with different looks". Digital cinema isn't something really progressive, it's, well, television tech trying to imitate the negative-positive process. Other than high sensitivity, there are no advantages.

Latitude? It's comparable to film in numbers, but you can't use the extremes of video characteristic curve - you need to add a "roll off" in processing (in-camera or post) and elevate the exposure enough above the noise floor for image to look natural, all while sacrificing potential stops of latitude. Film has a linear response plus smooth toe&shoulder "dy default".

Resolution? You don't care for it if it's sharp, i.e. has a good contrast at needed frequencies - which negative, both 16 and 35, does. See angular resolution of the eye and MTF.

Color? The most complicated thing. Modern color neg has purer color dyes than any single-sensor camera. Look at Fuji Eterna spectral graphs. It has no problem reproducing any color video "see", plus hues that video blends and doesn't distinguish. Color saturates naturally of film - on video, lighter = more saturated, which's unnatural and just looks wrong.

 

So if 100% of productions would be on film, DoPs will be still able to create images at least as subjectively good, and no movie will lose any production value. Quite the opposite. If it will be 100% digital though...

Edited by Michael Rodin
Link to comment
Share on other sites

With the amount of "new" lenses on the market today, for multiple mount types, there is zero reason a large-imager 3 CCD camera couldn't have been made and specialty glass that goes along with it.

What FFD would such a camera have? I mean equivalent light path, taking refraction index of prism into account. 120, 150mm? Now imagine, say, a 14mm T1,3 wide angle with a 150mm backfocus. Plus sharp and apochromatic. Then, a T3 10x zoom. It must be possible, but really, I mean, really expensive.

So you end up with $2-3-4 million camera package that your customers (rentals) can sell only for its colorimetery. Which is the single most important thing for a motion picture camera, but hagglers (sorry, production companies) don't understand it.

They won't do it. Too expensive in R&D and not marketable.

Link to comment
Share on other sites

You're confusing yourself with the notion of "different tools with different looks". Digital cinema isn't something really progressive, it's, well, television tech trying to imitate the negative-positive process. Other than high sensitivity, there are no advantages.

 

Well I think it's more subjective than that. Collateral looks good to me not only because of the fact that it required digital cameras to shoot the way they did at night, but because (i'm a layman) what I assume to be a high ISO creates heavy digital noise which, in Collateral at least, looks beautiful, and the appeal for me is similar to the appeal of heavy film grain in movies like Man of Steel or Minority Report. Similar appeal, but a distinct look.

Edited by Jesse Straub
Link to comment
Share on other sites

  • Premium Member

No confusion...

 

Single CCD's aren't even on the radar/even worth discussing.

 

Three CCD's are impossible to use with industry standard cinema glass that isn't specifically made for that design.

 

With the amount of "new" lenses on the market today, for multiple mount types, there is zero reason a large-imager 3 CCD camera couldn't have been made and specialty glass that goes along with it. Panavision did it, but even their attempt didn't work well enough. The whole concept that lenses dictated the imaging technology we use today in my eyes is semi-bogus.

 

We use CMOS for dozens of reasons as pointed out above, but also because it's A LOT CHEAPER AND SMALLER then a 3 CCD block. My "argument" above is that CCD's as good as they are, unfortunately aren't as good as CMOS in many ways and even CMOS isn't really that great.

 

To this day, we still haven't figured out a decent digital imaging capture and presentation system. What we have is good, but it's not great and requires far to much manipulation with both the origination files and in the presentation.

You really haven't got a bloody clue, have you?

Link to comment
Share on other sites

I really don't think you can design a stock that is "better" for 16mm yet isn't better for 35mm, because if the 16mm is better, then 35mm shooters are going to want to use it. Better as in sharper and less grainy? There isn't a market for that in 35mm?

Well, they haven't given resolution much priority because there's so much real estate on 35mm. The main priority is has been reducing grain, which is basically done by making the emulsion extremely thick. That allows for more light-sensitive volume so you can get away with smaller grains for a given ISO at the cost of more light scattering through said emulsion. Optimizing for 16mm would basically entail thinner emulsions to get a sharper image while sticking to lower ISOs to minimize grain. Like, 5203 is 50D but the MTF is 50% at 70L/mm. In 16mm with optimal lenses, the contrast is 50% at 700 lines. Now if you had a thinner 16D emulsion, you could get more like 160L/mm but with less grain at the same time. Plus, the lower speed would make it easier to get in the sweet spot of the lens without an ND filter.

When Vision2 was current, 50D wasn't sharp enough, 200T was too grainy, so I used 7212 rated at 50 ISO for everything as a compromise. Even then, emulsions were getting fairly thick. Now I'm forced to use 7213, which has about the same grain as 7212 but lower res.

 

 

Single CCD's aren't even on the radar/even worth discussing...We use CMOS for dozens of reasons as pointed out above, but also because it's A LOT CHEAPER AND SMALLER

Cost is the sole reason CMOS won that war. CMOS was vastly inferior but kept getting better that's where all the R & D has been going. A basic S16mm-sized CCD costs about $400. Enlarge that to S35-sized and it's $2,200, not including the external amplifiers, ADC etc. Now if you have 3x of them (not to mention the prism)... Consider a 2/3" (regular 16mm sized) 3-CCD camera costs at least $23,000 without a lens. Compare that to CMOS, which is about $80-$100 for S16 and $600 (though some are more) or so in S35. The amplifiers and ADCs are built into the CMOS chip as well as a lot of the image processing, so it not only make cameras manufacturing cheaper, but engineering is easier as well. A 3-CCD S35 camera would be unbelievably large and expensive, larger than any 35mm film camera. So, single CMOS it is.

 

The Digital Bolex, when fitted with a good quality lens (which nobody posting online examples seemed to use), was by far the most film-like video camera I've ever used and it was CCD-based. Great color, global shutter, 11-12 stop latitude when handled properly. Water under the bridge at this point. Hopefully, when the current technology matures, cameras will go back to having global shutters again and color/dynamic range will take precedence over cheated ISOs and pixel count.

 

 

If a DoP choses to shoot on video for some reason (I can't imagine anything other than light sensitivity though)

In my experience, most video cameras are natively 160-400 ISO, Alexa being an exception due to its enormous pixels.

 

 

 

You're confusing yourself with the notion of "different tools with different looks". Digital cinema isn't something really progressive, it's, well, television tech trying to imitate the negative-positive process.

Looking back at some of my earlier digital narratives, I did myself and my clients an enormous disservice by chasing a "film-like" look on video equipment. By just letting video be video, like I did later, I got much better images that ultimately look more professional even now.

 

 

 

Color saturates naturally of film - on video, lighter = more saturated, which's unnatural and just looks wrong.

YES, and the most disturbing artifact, which seems to have only appeared in the last ten years or so, is when light sources of certain colors are within a scene, they make the areas around them DARKER. Police car and dance club strobes are a great example... a blue light flashes and everything reflective around it turns dark blue rather than bright blue.

 

 

So you end up with $2-3-4 million camera package that your customers (rentals) can sell only for its colorimetery. Which is the single most important thing for a motion picture camera

A major problem with single-chip design is that the green channel only has 1/2 the sensor resolution and red/blue are 1/4. You can interpolate the values on the fly at low quality, like most cameras do, or you can do it as an offline process with much greater quality, but you can't restore information that was never captured. That is the sole reason the Alexa has a 2.8K area; so it can capture more color information and then be resized to 1920. It still isn't a perfect solution. All single-chip cameras allow at least some alias distortion in order to get the highest resolution possible. They merely hope you won't notice, particularly with cheap cameras, which tend to lack OLPFs (why they look sharper than pro cameras).

 

In order to avoid all artifacts, the optical resolution must be restricted to 1/4 the sensor's resolution. That means it would take an 8K sensor, optically limited to 2K to avoid alias distortion. Of course, 8K on an S35 sensor means .003uM pixels, or roughly 1/7th the native sensitivity and dynamic range of Alexa as well as 49x the data (more processing, more storage, worse rolling shutter). A 3-chip design can capture almost its full resolution without alias distortion and no interpolation/resizing is necessary.

While this conversation is purely academic, there is a lot of merit to 3-chip design, which is why it's so important for live production. The Thompson Viper was a raw 3-CCD camera and I was amazed by its quality, though still not a substitute for 35mm IMO.

  • Upvote 1
Link to comment
Share on other sites

  • Premium Member

Cost is the sole reason CMOS won that war. CMOS was vastly inferior but kept getting better that's where all the R & D has been going. A basic S16mm-sized CCD costs about $400. Enlarge that to S35-sized and it's $2,200, not including the external amplifiers, ADC etc. Now if you have 3x of them (not to mention the prism)... Consider a 2/3" (regular 16mm sized) 3-CCD camera costs at least $23,000 without a lens. Compare that to CMOS, which is about $80-$100 for S16 and $600 (though some are more) or so in S35. The amplifiers and ADCs are built into the CMOS chip as well as a lot of the image processing, so it not only make cameras manufacturing cheaper, but engineering is easier as well. A 3-CCD S35 camera would be unbelievably large and expensive, larger than any 35mm film camera. So, single CMOS it is.

Agreed, but if CCD was that much better, Panavision would have continued with Sony to develop a solution that kicked everyone's ass. However, they moved onto CMOS and they ain't looking back. Panavision is the only company who could have made a specialized solution work and even they gave up on the idea. I don't think it had anything to do with money.

 

The Digital Bolex, when fitted with a good quality lens (which nobody posting online examples seemed to use), was by far the most film-like video camera I've ever used and it was CCD-based. Great color, global shutter, 11-12 stop latitude when handled properly. Water under the bridge at this point. Hopefully, when the current technology matures, cameras will go back to having global shutters again and color/dynamic range will take precedence over cheated ISOs and pixel count.

I'm fortunate enough to have graded raw Digital Bolex footage and I thought it looked like crap. It has a forced look that can't really be manipulated in post that is trying to hard to look like film, it forgets what it is. Plus, it was super noisy, latitude was like 8 stops at most and it had harsh highlight clipping. I'm not a fan and that's part of the reason they went out of business, where Blackmagic still sells Pocket cameras like hot cakes. No, I can't make the pocket look like the Bolex Digital camera without manipulation in post, but honestly in most cases, I don't WANT it to look that way either. The benefits of shooting digitally go out the window if you're stuck with a very particular look from your digital camera.

 

While this conversation is purely academic, there is a lot of merit to 3-chip design, which is why it's so important for live production. The Thompson Viper was a raw 3-CCD camera and I was amazed by its quality, though still not a substitute for 35mm IMO.

I never liked the Viper.

Link to comment
Share on other sites

  • Premium Member

At risk of diving face-first into a fire ant nest while smeared liberally with honey,

 

 

 

that video cameras are somehow vastly easier and cheaper to use than film cameras

 

To be fair, it is very difficult to deny that without being really, really creative with the facts.

 

P

Link to comment
Share on other sites

Agreed, but if CCD was that much better, Panavision would have continued with Sony to develop a solution that kicked everyone's ass. However, they moved onto CMOS and they ain't looking back. Panavision is the only company who could have made a specialized solution work and even they gave up on the idea. I don't think it had anything to do with money.

You have to remember that while new CCDs were still being developed, efforts to improve the technology itself pretty much stopped 20 years ago to give CMOS tech more effort. By the time CMOS chips started appearing in things like the Alexa (which initially had a mechanical shutter), the differences in performance for large pixel sensors was minimal. However, CCDs can still generally beat CMOS in small pixel sensors but price matters more in that category. Case in point, I hacked a CCD camera with 1.4uM pixels to capture raw data and got a 10-stop range out of it (8 without the hack). You'd be lucky to get an 8-stop range out of a CMOS camera of similar pixel size without HDR or DNR trickery. Did I mention almost all CMOS chips have DNR built into them (and often more in the camera processor)? It may hide noise, but it introduces other artifacts.

 

 

I'm fortunate enough to have graded raw Digital Bolex footage and I thought it looked like crap. It has a forced look that can't really be manipulated in post that is trying to hard to look like film,

I'm sorry to hear that. The camera itself does almost no processing. It pretty much just converts the 16-bit linear data to 12-bit Gamma-1 and stores it. The rest depends on who's using/grading it. It's no coincidence that the Kodak sensor in the D16 has similar performance to Kodak 7217. The gamut is way beyond what most software/equipment handles by default, so it needs to be converted to Rec.709 for video release, to XYZ G2.6 for DCP or back to 16-bit linear for film-out. It seems most D16 shooters are self-taught video shooters using bad lenses, bad lighting/exposure technique and don't understand the raw workflow. It took me very little effort to get a natural look using freeware. I know another guy (professional video technician) who independently developed a similar work-flow and he also got great results.

 

 

Plus, it was super noisy, latitude was like 8 stops at most and it had harsh highlight clipping.

Not in my experience, though it doesn't have DNR built into it like CMOS cameras. If you want to crank up the gain, you have to apply the DNR yourself. As for the highlight clipping and latitude, that's an exposure/workflow issue. I found it has a 12-stop range and holds up to heavy grading quite well, though I really hate when people try to make the look of a project in the edit regardless of what camera they use.

 

 

I'm not a fan and that's part of the reason they went out of business, where Blackmagic still sells Pocket cameras like hot cakes.

$$$. The BMPCC was designed for wanna' be film shooters while the D16 was designed for actual film shooters. The Digital Bolex team said up front that they were making a camera for a very small niche market and actually got better response than expected. They mentioned the color space up front and that the viewfinder (even an external one) should not be used to judge exposure or white balance, just like on a film camera. They can't be held responsible for people using the technology incorrectly. They went out of business not from a lack of sales but because manufacturing costs more than doubled since they started, had issues with one of the manufacturers and knew they wouldn't survive a price hike. They could either become like everybody else, which they didn't want to do, or they could quit while they were ahead. I also suspect they hurt themselves with the hipster/retro marketing.

 

I was actually seriously considering getting a BMPCC, despite my gripes, just because of the price. After considering the rig I'd have to make to handle it, adding a proper optical block etc. it would have cost over $1,500 but still have worse color and rolling shutter than my $800 Canon. I can't stand rolling shutter. Even in multi-million dollar productions, that artifact just screams "cheap" to me.

Link to comment
Share on other sites

To be fair, it is very difficult to deny that without being really, really creative with the facts.

I have to agree with that. The down side is now everybody with $1,000 claims to be a professional "film" maker when most don't even deserve to even call themselves video makers.

Though if I were to do a run & gun documentary, I still don't think you can beat a 16mm camera in some respects. There's no boot/load time, no waiting for auto-focus etc. You just push the button the moment you see some action and know you have SOMETHING. I was on a motocross doc in the mountains on 2x 400' loads of 7201 with a crew of three (director, camera op, me on audio) and a CP-16. We never missed a thing; get a shot, run through the woods to the other side of the track, get them coming the other way, go back to the beginning for an interview, catch an accident as it happens, all without leaving the camera running (makes editing easy).

Edited by Stephen Baldassarre
Link to comment
Share on other sites

  • Premium Member

$$$. The BMPCC was designed for wanna' be film shooters while the D16 was designed for actual film shooters. The Digital Bolex team said up front that they were making a camera for a very small niche market and actually got better response than expected. They mentioned the color space up front and that the viewfinder (even an external one) should not be used to judge exposure or white balance, just like on a film camera. They can't be held responsible for people using the technology incorrectly. They went out of business not from a lack of sales but because manufacturing costs more than doubled since they started, had issues with one of the manufacturers and knew they wouldn't survive a price hike. They could either become like everybody else, which they didn't want to do, or they could quit while they were ahead. I also suspect they hurt themselves with the hipster/retro marketing.

 

I was actually seriously considering getting a BMPCC, despite my gripes, just because of the price. After considering the rig I'd have to make to handle it, adding a proper optical block etc. it would have cost over $1,500 but still have worse color and rolling shutter than my $800 Canon. I can't stand rolling shutter. Even in multi-million dollar productions, that artifact just screams "cheap" to me.

 

This statement may be true, but it doesn't explain why "multi-million dollar productions" use them as expendable crash cameras. Where before they used Canon, now GoPro and the BM Pocket is used. I am continually impressed by the pocket camera from BlackMagic. I also hate rolling shutter, but I haven't seen better color from a Canon. I have seen great Rec.709 Video color. The Panasonic GH5 also makes a very nice pretty picture, but I wouldn't want to grade it.

 

I can push Blackmagic footage as far as I want, cut it with Alexa footage, grade it, and it looks great. There are many shows who deploy PL mount Blackmagic cameras along side their Alexa cameras with no problems.

 

I find it hard to believe that a cheap Canon camera without the capability of RAW output can exceed the color depth of RAW 16bit data stored in a 12bit log file.

Link to comment
Share on other sites

  • Premium Member

The BMPCC was designed for wanna' be film shooters while the D16 was designed for actual film shooters.

The BMPCC was designed for people who wanted a cheap cinematic look out of a small camera.

 

The D16 was a complete and utter toy that had zero purpose for existence. If it was literally a digital Bolex, as in the chassis/feature set were identical to a Bolex, only it had a digital imager, that would have been quite cool. However, what they designed was atrocious, not just in physical form, but also in what it was capable of doing. If it were sold at your local camera shop for $999, it would be worth the risk, but the prices they wanted were outrageous. The pocket camera blows the doors off the D16 any day of the week in pretty much every area.

They went out of business not from a lack of sales but because manufacturing costs more than doubled since they started, had issues with one of the manufacturers and knew they wouldn't survive a price hike. They could either become like everybody else, which they didn't want to do, or they could quit while they were ahead. I also suspect they hurt themselves with the hipster/retro marketing.

They could have kept going and changed vendors and/or manufactured somewhere else. If they were selling like hot cakes, they would have made it work, but when the quantities are so low, it's hard to justify. The design/marketing was atrocious.

 

I was actually seriously considering getting a BMPCC, despite my gripes, just because of the price. After considering the rig I'd have to make to handle it, adding a proper optical block etc. it would have cost over $1,500 but still have worse color and rolling shutter than my $800 Canon. I can't stand rolling shutter. Even in multi-million dollar productions, that artifact just screams "cheap" to me.

If you're very careful, the pocket can deliver great images without much rolling shutter effect. It's just, people aren't willing to be careful with how they shoot. All of these low-budget cameras have similar limitations, it's all about how you adapt. As I said earlier, I've been shooting with the pocket for going on 4 years now and it's been a frustrating experience sometimes, mostly because the thing is so damn sensitive, you've gotta run ND's to make a good image unless it's interior/dark.

Link to comment
Share on other sites

This statement may be true, but it doesn't explain why "multi-million dollar productions" use them as expendable crash cameras. Where before they used Canon, now GoPro and the BM Pocket is used.

I can't explain that either, especially GoPros. Their image looks like standard def and the lens distortion is nauseating.

 

 

I have seen great Rec.709 Video color.

Sure, but that only defines digital output ranges. What actually gets captured can vary extremely and you can't go directly from something with color/contrast ranges approaching Kodak 7217 to a Rec.709 output without doing something to reign in said image, or you wind up with a very fake looking 8-stop pallet like Mr. Purcell got (clipped Y and Cb/Cr channels, no knee).

 

 

The D16 was a complete and utter toy that had zero purpose for existence. If it was literally a digital Bolex, as in the chassis/feature set were identical to a Bolex, only it had a digital imager, that would have been quite cool. However, what they designed was atrocious, not just in physical form, but also in what it was capable of doing. If it were sold at your local camera shop for $999, it would be worth the risk, but the prices they wanted were outrageous.

How is it a toy?

Just to clarify, if it DIDN'T have professional audio capabilities, no HDMI or other options, it would have been a more serious camera? That's exactly what Joe Rubenstein (the creater) wanted. I pushed for that as well. It was the Kickstarter backers that insisted on adding in a bunch of other stuff, which delayed production and greatly increased the cost of the camera. They did not pass that added cost onto the customers, they just wound up selling them for close to their cost.

 

 

The pocket camera blows the doors off the D16 any day of the week in pretty much every area.

It beats the D16 in precisely one area, price. Having used both, I might accept a BMPCC if somebody gave it to me, but I wouldn't spend money on one.

I'd like to see some of your D16 DNGs. I can only imagine whoever was shooting didn't know what they were doing and/or you have an issue with your work-flow.

 

 

They could have kept going and changed vendors and/or manufactured somewhere else. If they were selling like hot cakes, they would have made it work, but when the quantities are so low, it's hard to justify.

True. While most of the camera was made from off the shelf parts, assembled in Canada, the FPGA and chasis were custom built in China, which I believe is what was giving them grief. Switching manufacturers would mean almost starting from scratch. I know the prototype cost about $1,500, but that was before they added in all that other stuff. Seriously, just the HDMI port complicated things SO much! They originally weren't going to have one, but the backers griped. So they added HDMI, which required another daughter card and a bigger processor, which added power consumption and heat, required a re-designed FPGA, which all together required a fan, which added more power consumption. They weren't going to have audio capture originally, but the backers insisted. Then they were going to put in crap audio capabilities like all cheap cameras do but I (along withe a few others) said "I'd rather it not have audio because I'll never use that." They eventually decided to add pro quality, with good op-amps, phantom power and analogue gain. It has 24-bit 96KHz capable ADCs as well, near top of the line. Does ANY sub-$10,000 have that? The audio also added to the FPGA design and another daughter card. By the time they were done implementing all the features people wanted (they held firm on no compression, thankfully), the chasis went from a single board with a sensor daughter card to being jam packed with stuff. Then they had to write the firmware, which also added delays and cost, but they never budged on the price.

 

 

The design/marketing was atrocious.

Got that right! I don't know WHAT they were thinking.

 

 

If you're very careful, the pocket can deliver great images without much rolling shutter effect.

See, I don't think somebody should have to change their project to suit the camera. I don't want to be forced to avoid all but subtle movement because the camera won't take it. I don't want to be forced to avoid fine details, to use lighting that doesn't emit IR etc. because the manufacturer wanted to keep the price under $1,000. A proper optical block costs almost $400 in that size, so they just skipped it.

Link to comment
Share on other sites

  • Premium Member

The Blackmagic pocket camera really is extremely good.

 

As to the ergonomics, load time, and other issues on many modern cameras: no serious camera operator of any stripe would deny it.

 

I've complained endlessly that many modern cameras have absurdly awful ergonomics, especially given that the best layout is (to my mind) pretty clearly that of an ENG camera, and that's very reliant on a specific design of lens as a crucial part of the package. The FS700, for instance, has a layout that's absolutely laughable. It's essentially impossible to operate handheld with any degree of effectiveness without significant rigging.

 

One particularly crucial problem that bedevils a lot of modern cameras and particularly raises its head when rigging them for off-the-shoulder use is lag in the viewfinding and/or signal outputs. It's very rarely talked about, even in reviews of third-party viewfinders, but it's a huge issue. The Ursa Mini is a brilliant thing, but the viewfinder lag is at the very least four frames or it was last time I measured it a few months ago. That's way too long, and it's not something that can be fixed with bolt-on accessories.

 

You can also complain about rolling shutter. It needs to go away.

 

Now, the Blackmagic Pocket obviously wasn't designed to do that work and I don't know what the lag is like, but you could in theory put it on a Tilta BS-T03 baseplate with a V-lock mount, a viewfinder, and one of MTF's adaptors to a B4 lens, and have something pretty usable.

 

And cheap.

 

P

Link to comment
Share on other sites

  • Premium Member

I can't explain that either, especially GoPros. Their image looks like standard def and the lens distortion is nauseating.

 

Now I will argue with you on this point, for the sake of argument. Please show me, in context, a frame from a film released in the past 3 years that has an image that looks standard def or with nauseating lens distortion. The GoPro footage used on The Martian look fine, especially in context. The BM Pocket cameras used on Transformers: Age of Ultron surely look better than SD, or they would have chosen a different camera.

Link to comment
Share on other sites

 

Now I will argue with you on this point, for the sake of argument. Please show me, in context, a frame from a film released in the past 3 years that has an image that looks standard def or with nauseating lens distortion. The GoPro footage used on The Martian look fine, especially in context. The BM Pocket cameras used on Transformers: Age of Ultron surely look better than SD, or they would have chosen a different camera.

 

An example of bad GoPro footage for me would be in the second Hobbit movie during the barrel chase.

Link to comment
Share on other sites

The Blackmagic pocket camera really is extremely good.

Maybe I'm more sensitive to its issues than most, but I just don't get why people think it's a good camera. They just cut SO many corners.

 

 

given that the best layout is (to my mind) pretty clearly that of an ENG camera

No argument here, though many serious ENG cameras are a bit heavy for long shoots. There's a good number of 16mm cams with similar form factor too, like some of the Aatons. You can shoulder one of those for hours... and putting the take-up right by the supply is genius, so the weight doesn't shift as you shoot.

 

 

One particularly crucial problem that bedevils a lot of modern cameras and particularly raises its head when rigging them for off-the-shoulder use is lag in the viewfinding and/or signal outputs.

That's a great point. I have been in the habit of keeping both eyes open while shooting so that doesn't particularly bother me. I see the action through my left eye and the framing with my right eye (opposite with my K3) if that makes sense. I can imagine a 4-frame delay would be useless for catching fast action, but I wouldn't want to use a CMOS camera for fast action any way.

 

 

You can also complain about rolling shutter. It needs to go away.

A lot of CMOS sensors have global shutter modes (including the BMPCC) but are disabled because you typically lose something like 4-5 stops in the shadows. I'm REALLY wishing somebody will do something with those new Sony CMOS sensors that store/dump charges like CCDs. There's a $500 USB 3.0 industrial camera that uses them, so how hard could it be to make a raw camcorder with it? It has a 12mm frame and very accurate color too. So, stick in a proper OLPF, rig up some kind of portable SSD capture system and you're in business. I would TOTALLY buy that!

 

 

Now, the Blackmagic Pocket obviously wasn't designed to do that work and I don't know what the lag is like, but you could in theory put it on a Tilta BS-T03 baseplate with a V-lock mount, a viewfinder, and one of MTF's adaptors to a B4 lens, and have something pretty usable.

Don't forget the replacement optical block and something to add some weight (there is such thing as TOO light). All in all, you're talking at least 2,800 to get a usable camera out of it.

Maybe parallax viewfinders could come back!? JK :p

 

Now I will argue with you on this point, for the sake of argument. Please show me, in context, a frame from a film released in the past 3 years that has an image that looks standard def or with nauseating lens distortion. The GoPro footage used on The Martian look fine, especially in context. The BM Pocket cameras used on Transformers: Age of Ultron surely look better than SD, or they would have chosen a different camera.

Did they replace the lenses? I see GoPro video inter-cut with professional cameras on TV often and it's like going from "Blu-Ray" to "S-VHS delivered via You Tube".

 

They used cameras on the Transformers movies? From the previews, I thought it was all obvious CG. I've only seen the 1986 movie (when it was new) but that was 35mm.

Link to comment
Share on other sites

  • Premium Member

Maybe I'm more sensitive to its issues than most, but I just don't get why people think it's a good camera. They just cut SO many corners.

And Sony and Panasonic's cameras which are 2x the money, didn't? I think the A7SMKII looks like utter crap. Even though the GH4 and now GH5 look "ok" neither one looks marketably better, OR has a global shutter.

 

Again, compare apples to apples... what other $998.00 camera is BETTER then the pocket camera? Also, what's the lowest cost camera that shoots RAW and 10 or 12 bit Pro Res?

 

Honestly, I refuse to have an extra piece of hardware connected to my camera JUST to get a decent recording out of it. If a camera manufacturer can't build-in a decent codec that's usable, I refuse to even discuss their product's existence.

 

A lot of CMOS sensors have global shutter modes (including the BMPCC) but are disabled because you typically lose something like 4-5 stops in the shadows.

Yep... and remember, the Alexa and Red cameras have rolling shutter as well. So the whole debate about rolling shutter is kind of moot if ALL the cameras have the problem.

 

Don't forget the replacement optical block and something to add some weight (there is such thing as TOO light). All in all, you're talking at least 2,800 to get a usable camera out of it.

I use the built-in LCD display with a viewfinder adaptor that sticks on the back, so you can put your eye up against it. Again, been shooting with it for going on 4 years, dozens of shows per year and never once had an issue. Even though the display isn't high quality, with the viewfinder adaptor, it does work fine.

Link to comment
Share on other sites

  • 2 weeks later...

This is a discussion for another thread cuz we could go on about CCD's vs CMOS all night long! :)

 

Suffice to say, nobody has made a CCD look as cinematic as CMOS yet, artificial or not, the high resolution cinema camera version doesn't exist. So until it does, it's all theory and discussions.

 

I will have to strongly disagree here. The F35 looks much better and more cinematic than the F65 or F55. It looks much more cinematic than anything from Blackmagic, Red or Panasonic too. I think the reason is mainly the CCD. The other is that it has better color because of the sensor design. Then it is a global shutter which helps with the cinematic motion. It's a pity that the whole 4K hype brought on by the release of the Red One made many overlook it. Today when you compare Red One footage to F35 footage it's so obvious how superior the F35 is. But back then the 4K koolaid clouded the minds.

  • Upvote 1
Link to comment
Share on other sites

Agreed, but if CCD was that much better, Panavision would have continued with Sony to develop a solution that kicked everyone's ass. However, they moved onto CMOS and they ain't looking back. Panavision is the only company who could have made a specialized solution work and even they gave up on the idea. I don't think it had anything to do with money.

 

 

I'm fortunate enough to have graded raw Digital Bolex footage and I thought it looked like crap. It has a forced look that can't really be manipulated in post that is trying to hard to look like film, it forgets what it is. Plus, it was super noisy, latitude was like 8 stops at most and it had harsh highlight clipping. I'm not a fan and that's part of the reason they went out of business, where Blackmagic still sells Pocket cameras like hot cakes. No, I can't make the pocket look like the Bolex Digital camera without manipulation in post, but honestly in most cases, I don't WANT it to look that way either. The benefits of shooting digitally go out the window if you're stuck with a very particular look from your digital camera.

 

 

I never liked the Viper.

 

The main reason CCD was replaced by CMOS is cost. It doesn't matter that it was better. A 4K CCD presents very expensive technical challenges. CMOS was good enough and much cheaper.

 

  • Upvote 1
Link to comment
Share on other sites

 

A major problem with single-chip design is that the green channel only has 1/2 the sensor resolution and red/blue are 1/4. You can interpolate the values on the fly at low quality, like most cameras do, or you can do it as an offline process with much greater quality, but you can't restore information that was never captured.

 

Yes this is a problem with single sensor. But I think Sony's solution for the F35 and Genesis is the most elegant and works the best. It definitely produces better color than CMOS cameras and bayer. It gets you "full RGB."

Link to comment
Share on other sites

  • Premium Member

There were some downsides to the RGB striped approach -- true, you had roughly 2.2MP equally of red, green, and blue photosites so the color was nice, but the striped approach this required created color moire problems due to the two-row gap between, let's say, a green row and the next green row. And for whatever reason, maybe related to the striped sensor, the OLPF was fairly weak so you had a lot of sharp but false edges -- I had moire problems with almost everything in the frame that had a pattern: rough fabric on a corduroy couch, horizontal rows of shingles on a roof, etc. Every piece of wardrobe had to be tested for moire issues.

 

And it was not the cleanest of signals for whatever reason, and that noise combined with the compression of HDCAM-SR, though mild, gave the image a unique texture in underexposed areas -- some people thought it gave it a film-like grain texture but sometimes to me it just looked unpleasant, just depended on the lighting. When you had plenty of light, the image looked pretty good with rich colors. But in warm, underexposed light like in a low-lit interior at night, there could be a noisy muddiness. Probably some of that was pattern noise.

 

To create a 4K version of that sensor, it would be something close to 8MP per color for a 2:1 sensor, 12MP for a 4:3 sensor, so either a 24MP or 36MP sensor in a Super-35 form (unless you wanted to go larger with a full-frame sensor.) Something like the 8K S35 Helium sensor except dedicated to outputting 4K per channel.

 

But I also think that one can oversell the idea of "full RGB" -- what does that exactly mean in real-world scenarios? After all, color negative can be scanned at full RGB... but that doesn't mean that each color layer resolves detail equally. The old 3-strip Technicolor camera recorded three separate b&w negatives but the red record was always the softest and grainiest. So the fact that on a Bayer sensor, there is "better" green information than red or blue doesn't mean you can't resolve color information well. And as sensors get higher in resolution, the color channel limitations of a Bayer sensor matter less because you are oversampling in general.

Link to comment
Share on other sites

There were some downsides to the RGB striped approach -- true, you had roughly 2.2MP equally of red, green, and blue photosites so the color was nice, but the striped approach this required created color moire problems due to the two-row gap between, let's say, a green row and the next green row. And for whatever reason, maybe related to the striped sensor, the OLPF was fairly weak so you had a lot of sharp but false edges -- I had moire problems with almost everything in the frame that had a pattern: rough fabric on a corduroy couch, horizontal rows of shingles on a roof, etc. Every piece of wardrobe had to be tested for moire issues.

 

And it was not the cleanest of signals for whatever reason, and that noise combined with the compression of HDCAM-SR, though mild, gave the image a unique texture in underexposed areas -- some people thought it gave it a film-like grain texture but sometimes to me it just looked unpleasant, just depended on the lighting. When you had plenty of light, the image looked pretty good with rich colors. But in warm, underexposed light like in a low-lit interior at night, there could be a noisy muddiness. Probably some of that was pattern noise.

 

To create a 4K version of that sensor, it would be something close to 8MP per color for a 2:1 sensor, 12MP for a 4:3 sensor, so either a 24MP or 36MP sensor in a Super-35 form (unless you wanted to go larger with a full-frame sensor.) Something like the 8K S35 Helium sensor except dedicated to outputting 4K per channel.

 

But I also think that one can oversell the idea of "full RGB" -- what does that exactly mean in real-world scenarios? After all, color negative can be scanned at full RGB... but that doesn't mean that each color layer resolves detail equally. The old 3-strip Technicolor camera recorded three separate b&w negatives but the red record was always the softest and grainiest. So the fact that on a Bayer sensor, there is "better" green information than red or blue doesn't mean you can't resolve color information well. And as sensors get higher in resolution, the color channel limitations of a Bayer sensor matter less because you are oversampling in general.

 

 

Naturally there were compromises. Even a 3 chip set up has compromises as anything when it comes to engineering. But what I was trying to say is from all the solutions the F35 sensor was the best approach IMO. The least of the evils if you will.

 

The particular problems you mentioned were greatly reduced with updates. Especially the very last update which also increased the dynamic range and opened up 12-bit 444 recording. Unfortunately when that update came was a little too late already.The F65 was already out, or at least announced. But it greatly reduced if not completely illuminated the problems you point out. Not to mention you can add filters to help it too.

 

You are right though that it was not the cleanest signal. But Sony and Panavision's intention was to imitate film. So I think it was never meant to be super clean. The grain is a part of the look there. I'm pretty sure this is the reason it is there. I have actually seen it mentioned when promoting the camera. But there is a noise reduction function which works really well if you need it. It was again further improved with the very latest firmware. But I particularly like the grain and think it looks less like video noise than the other cameras.

 

Your complain about the noise in combination with HDCAM-SR compression is another sign which tells me you are basing this on the early release F35. This is the problem since most have only used the camera with the tape recorder and limited HDCAM-SR tape format. Once Sony opened up 12-bit 444 recording, which you can record either at 880mbs which is about 2:1 compression or fully uncompressed , any compression problems disappeared. The noise is very fine and free of artifacts. S-log 12-bit 444 at 880mbs and 2:1 compression for HD is very high quality. Uncompressed even better of course.

 

Yes, a 4K 35mm sensor using the same technology would be around 36MP. As it is the F35 sensor is 12MP. That sensor would also be extremely expensive to produce if a CCD.

 

In real world scenarios full RGB means better overall color. When grading something shot on the F35 compared to grading something shot on the F5 for example the F5 image will break first. I also think the color in the F35 looks so nice because of the full RGB. Especially the skins.

 

The fact that most movies shot with the F35 never got to use the 12-bit 444 format shows how it was never really used at its best. Most movies or people who tried it did so with the HDCAM tape recorder. A wasted chance for Sony really. They should have updated the camera earlier. If I remember right the update came in 2011 or 2012.

Edited by Adam Paul
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...