camera makers may finally focus on something besides more and more resolution

[combatentropy]

http://www.dpreview.com/reviews/nikond1/

According to this Gizmodo article, the tide is finally turning to light sensitivity instead of more and more megapixels --- having hit what it calls "megapixel fatigue."

Considering that your average theater print is, like, 1K, I've thought that going to 4K in digital cinema is a waste of pixel area when you still lag so far behind film in dynamic range.

On a side note, the article mentions that in 1999 Nikon made the D1, which sold for $5,000 and it had a 2000 x 1312 Super-35-sized CCD. Granted, it did not shoot 24 frames per second. But are you really going to stand up and tell me that eleven years later, we still cannot make that CCD shoot at video rates for under $5,000?

 

[trez]

(quote): "In 1975, the first digital camera took 23 seconds to record a 100-line black-and-white photo onto cassette tape. Today, a Nikon D3s takes photos with 12 million pixels at 1/8000 of a second" - wow, do they mean it can shoot 8000 images per second? what a bullshit...

While I generally agree light sensitivity is important, I wouldn't say 4K is a waste even if you end up with 1K print - oversampling has significant advantages (mtf/sharpness, easier optical lpf design, not to mention debayering issues in case of single chip camera)

 

[Simon Höfer]

wow, do they mean it can shoot 8000 images per second? what a bullshi*...

Maybe you should read it again.

 

[PerroneFord]

But are you really going to stand up and tell me that eleven years later, we still cannot make that CCD shoot at video rates for under $5,000?

Making it shoot 24fps isn't the problem. Making it shoot 24fps in something smaller than a Genesis, an F900, etc., could be.

Physics is darn hard to overcome, as is thermodynamics. Likely the reason the Scarlet has been so delayed.

 

[brian hanson]

i was at red day yesterday and got to see the 4k projection of their reel, it has to be seen to be understood just how good it looks.
jannard said that if the scarlet isnt released by the end of summer he will kill himself.

 

[Ian-T]

Uh oh...I guess it's about time to write out a eulogy.

Jim was a good man. He had very high hopes for his girl....Scarlet...

 

[trez]

Maybe you should read it again.

not really. maybe you should...
They compared the old digital camera which RECORDS one frame to the media (tape) in 23 seconds to the new Nikon which shoots at 1/8000 EXPOSURE TIME - two entirely different things. I don't think that old camera had 23 seconds long exposure time (maybe I'm wrong, though)
If Nikon was capable of processing and RECORDING single frame to its storage media (card) in 1/8000 seconds it would be (potentialy) capable of shooting 8000 fps - hence my strange comment

 

[Samm_Parnell]

to me it is alot about the consumer market, like cars the with more horsepower (useless on anything but a racetrack) its all about the big numbers, i remember in film school that everyone was going mad about 2k res and had no idea how to light, frame or what not, just thought the higher the number the better......right, seen lots shot on the red at 4k that is crap

 

[Otis Grapsas]

http://www.dpreview.com/reviews/nikond1/

But are you really going to stand up and tell me that eleven years later, we still cannot make that CCD shoot at video rates for under $5,000?

Based on the size of the market, a product with such a sensor would cost more than 80,000 euro, and that's conservative.

It can be done cheaply for a still camera if one of the big companies use it. They have to market size to put it out at less than 1,000 euro. But there is no real reason to use CCD today in a still camera. The low data rate CMOS sensors are as good and much more cost effective. There are many slow readout large pixel CCD sensors in the scientific imaging market.

But not many people would buy a camera with such a large pixel size. The market is based on megapixel count and that will never change. Look at the shrinking of pixel size and dynamic range on a consumer camcorder. We went from 0.4mp on 1/3" sensors to 4megapixels on sensors that are 5 times smaller. 1/4" with 2 or 3megapixels is considered professional now.

 

[Otis Grapsas]

If Nikon was capable of processing and RECORDING single frame to its storage media (card) in 1/8000 seconds it would be (potentialy) capable of shooting 8000 fps - hence my strange comment

That's just a shutter speed. It doesn't mean the sensor is already read, or that it's ready for another shot and that the data is output. None of all this actually. It can take a sensor 1/50th to read a frame and the shot would still be 1/1000 sec or anything else because that's just the mechanical or electronic shutter time.

This is the problematic part:

"In 1975, the first digital camera took 23 seconds to record a 100-line black-and-white photo onto cassette tape. Today, a Nikon D3s takes photos with 12 million pixels at 1/8000 of a second. And it can see in the dark."

The modern readout time could be anything but it's never the shutter speed. The best low light cameras available today for scientific imaging also have very long readout times for technical reasons.

 

[trez]

That's just a shutter speed. It doesn't mean the sensor is already read, or that it's ready for another shot and that the data is output. None of all this actually. It can take a sensor 1/50th to read a frame and the shot would still be 1/1000 sec or anything else because that's just the mechanical or electronic shutter time.

Good point, Otis. I'm aware of this distinction. I deliberately identified two different parameters (shutter speed and readout time) to expose the serious mistake in the original text which is based on similar incorrect identification - comparing the 23sec readout/write time of old CCD camera with 1/8000th shutter speed of new Nikon doesn't make any sense, because these two are very different things. If they were comparable Nikon would be capable of shooting 8000 fps - this is what I meant.

 

[Otis Grapsas]

Thanks for the clarification trez. I was also trying to clarify.

To be fair to the small pixel supporters, the pixel size can be smaller if the technology allows a lower noise. 40,000 (full) /40 (noise) is the same with 10,000/10 in dynamic range. Both 10 stops. And if the sensors are the same size, the quantum efficiency is the same, and the full bucket figure manages to scale linearly to the reduction of total pixels, the sensitivity will also be the same. 1/4th of the photons reach the small pixels but the pixel is happy because its bucket is filled with that 1/4th, it achieves the same signal level (100%). The 4 times lower noise can also preserve the dynamic range as mentioned above, so all is well.

The lens quality in also not a limitation. Getting 4 times the pixels on the same lens and sensor size simply oversamples the image, which is good if the final output will have the same projection size (screen) as the large pixel sensor.

This is entirely theoretical though and something I have never seen in practice, so there must be lots of technical limitations when using small pixels. For video I would prefer everything to be 1:1, to use a sensor of same resolution to the required output format for the application. Oversampling has a chroma advantage but with the present HD delivery formats being 4:2:0, even 4:2:0 recordings of a bayer sensor cover the requirements.

 

[combatentropy]

For video I would prefer everything to be 1:1, to use a sensor of same resolution to the required output format for the application. Oversampling has a chroma advantage but with the present HD delivery formats being 4:2:0, even 4:2:0 recordings of a bayer sensor cover the requirements.

I agree. Especially when you are recording uncompressed, as you are, Otis, extra megapixels consume media real fast.

I will gladly trade oversampling for more dynamic range, until sensor advancement reaches around 16 stops. Then, if you want to spend dynamic range on extra resolution, you can.

I would really like to see a 2K Super 35 chip camcorder, with a global shutter (CCD or global-shutter CMOS, I don't care). I believe that the reason we haven't seen it for a price that we independents can afford is because the people who make them can still sell them for $100,000. Although I think they are making an error. They may sell 100 or 1000 of them. But if they made something like what I want, they might sell 100,000 of them. I'm not expert on the market, and it's their right to make whatever they want and set whatever price they want.

But I have yet to hear why, for technical not market reasons, we can't have an affordable 2K S35 global-shutter sensor that can do at least 60 frames per second.

 

[Otis Grapsas]

I would really like to see a 2K Super 35 chip camcorder, with a global shutter (CCD or global-shutter CMOS, I don't care). I believe that the reason we haven't seen it for a price that we independents can afford is because the people who make them can still sell them for $100,000. Although I think they are making an error. They may sell 100 or 1000 of them. But if they made something like what I want, they might sell 100,000 of them. I'm not expert on the market, and it's their right to make whatever they want and set whatever price they want.

But I have yet to hear why, for technical not market reasons, we can't have an affordable 2K S35 global-shutter sensor that can do at least 60 frames per second.

There is no doubt that an affordbale camera can sell more. The new affordable digital cinema cameras could actually get 15 to 20% share of their price range (which is filled with inferior but much more practical camcorders). The 80% that is left does not care for this type of camera.

But not all hardware systems allow the high units low cost paradigm. Most big cameras already cost a reasonable price for their manufacturing cost. If the camera is easy to source in materials and can be affordably supported and serviced because the components allow that, you can be sure it will be cheaper. Everybody wants a larger market share.

It looks like an important market to us, those who are involved in video, but I doubt it is more than a drop on the ocean for the large companies who do business in other segments. A few dozens cameras on rental can serve a few 100s of films every year, probably all the cinematography that would care for digital. I don't see these cameras becoming cheaper and I wouldn't be surprised if some of the large companies were actually losing money in the digital cinema sector. If they did have 1000 preorders at 50% of its present price they could probably achieve that retail price. With the present interest this market gets, it's not probable.

There is another parameter also. Rentals. This is a small market and an owned cinema camera does not get many hours per year. Rentals are effectively shrinking the market and increasing the cost for those who actually care to own the camera. Some rental houses rent for less than would be reasonable only to keep the rest of the stock that comes with the package busy. The very availability of the camera in rental hurts the product and the camera owners. A latest trend is owners that see the camera as a business and get into rental. I believe the best paradigm for rentals is to have them controlled by the actual company. Setting greed aside it can bring the camera to more users and owners at far lower prices. But some rental houses would oppose of course, and eventually try to make your life harder by not renting support equipment without a camera package. Anyway, this is way out of topic.

For CCD solutions, limiting the frame rate makes sense because it reduces noise a lot. You want the lowest frame rate. Some scientific cameras come in different editions like 15fps, 20fps, 30fps, and most people prefer the lowest speed for the superior image quality. The high end cameras are extremely low in frame rate and they achieve even better image quality. All have the same sensor of course, only support electronics change.

 

[combatentropy]

There is no doubt that an affordbale camera can sell more. The new affordable digital cinema cameras could actually get 15 to 20% share of their price range (which is filled with inferior but much more practical camcorders). The 80% that is left does not care for this type of camera.

But not all hardware systems allow the high units low cost paradigm. Most big cameras already cost a reasonable price for their manufacturing cost. If the camera is easy to source in materials and can be affordably supported and serviced because the components allow that, you can be sure it will be cheaper. Everybody wants a larger market share.

It looks like an important market to us, those who are involved in video, but I doubt it is more than a drop on the ocean for the large companies who do business in other segments. A few dozens cameras on rental can serve a few 100s of films every year, probably all the cinematography that would care for digital. I don't see these cameras becoming cheaper and I wouldn't be surprised if some of the large companies were actually losing money in the digital cinema sector. If they did have 1000 preorders at 50% of its present price they could probably achieve that retail price. With the present interest this market gets, it's not probable.

There is another parameter also. Rentals. This is a small market and an owned cinema camera does not get many hours per year. Rentals are effectively shrinking the market and increasing the cost for those who actually care to own the camera. Some rental houses rent for less than would be reasonable only to keep the rest of the stock that comes with the package busy. The very availability of the camera in rental hurts the product and the camera owners. A latest trend is owners that see the camera as a business and get into rental. I believe the best paradigm for rentals is to have them controlled by the actual company. Setting greed aside it can bring the camera to more users and owners at far lower prices. But some rental houses would oppose of course, and eventually try to make your life harder by not renting support equipment without a camera package. Anyway, this is way out of topic.

For CCD solutions, limiting the frame rate makes sense because it reduces noise a lot. You want the lowest frame rate. Some scientific cameras come in different editions like 15fps, 20fps, 30fps, and most people prefer the lowest speed for the superior image quality. The high end cameras are extremely low in frame rate and they achieve even better image quality. All have the same sensor of course, only support electronics change.

Well, you bring up some points that I would enjoy discussing, but maybe not all this particular thread, because it would overwhelm it. I know that you are eagerly working on your camera, and I agree with several of your design decisions (larger sensor, uncompressed, open standards, etc.)

In my involvement in videomaking since 1990, I've witnessed what I think is frustratingly slow progress. It's sort of like the plight of the electric car. Shall I make a documentary called "Who Killed the Reasonably Priced Uncrippled Digital Motion Picture Camera?"

I think it is a big part greed, but also a great deal of fumbling around by manufacturers who keep doing what has worked for them, somewhat. It sort of reminds me of used car commercials. Nobody likes watching used car commercials. They are loud, garish, positively assaulting --- my family would always mute the TV when they came on. I mean, it's a long-time American joke. Everybody knows that they are awful. So, why do they keep making them that way? Because that's what they're competitors are doing. And they're car salesmen, not filmmakers. So they tell the commercial company to "Make it look just like that one by Jack Cheapskate's down the road, only with my name on it, not his."

My point is, if we (the buyers) could consolidate our voices more about what we want, they would maybe progress faster. I read a lot of excuses by us on their behalf about what is possible. How do we know what chipmakers are capable of? Is it because you have hands-on knowledge making image sensors? Or do you think if you hear buzzwords enough (quantum efficiency, fill factor) that you understand this very specialized field? Or are you going by what the camera makers tell you? I'm sure they're not biased.

Another one I love: No event videographers or documentarians like shallow depth of field. Only those film students want shallow DOF. Event videographers MUST have 1/3" sensors. Certainly anything more than 2/3" will turn their videos in an unfocused mush. Oh really? You mean like these gorgeous wedding videos that were shot with a Canon 5D?

http://vimeo.com/user439397

Plus, this myth flies in the face of all those photojournalists who shoot sports and events and all sorts of crazy stuff on the mammoth full-frame 35mm (that's 36x24mm, compared to the Super 35 that Hollywood movies are shot on is MUCH SMALLER at about 25 x 19mm). I have even heard a rumor that there were these photojournalists at LIFE magazine, and they used to use MANUAL LENSES. And they didn't even have focus pullers!!!

 

[studio1972]

But I have yet to hear why, for technical not market reasons, we can't have an affordable 2K S35 global-shutter sensor that can do at least 60 frames per second.

1. You would need to design a sensor specifically for this camera, a large (but relatively low resolution compared to a stills camera) sensor that did not overheat.

2. You would need to design an extensive set of video lenses, primes and zooms, which can be attached to the camera, this would be very expensive and involve a large engineering effort.

3. The resulting camera must not cost too much more than a DSLR

If you don't think the camera you require needs these things, then you would be pretty happy with a DSLR really, which is available already.

 

[ajcourtney]

I have even heard a rumor that there were these photojournalists at LIFE magazine, and they used to use MANUAL LENSES. And they didn't even have focus pullers!!!

LOL!:laugh:

 

[combatentropy]

1. You would need to design a sensor specifically for this camera, a large (but relatively low resolution compared to a stills camera) sensor that did not overheat.

Why would this be more expensive than the sensors in DSLRs that have video mode?

2. You would need to design an extensive set of video lenses, primes and zooms, which can be attached to the camera, this would be very expensive and involve a large engineering effort.

Why can't I use the lenses already in existence for this sensor size, which is basically APS-C?

 

[Barry_Green]

Why would this be more expensive than the sensors in DSLRs that have video mode?

If it had a low-pixel-count large sensor, that may not have to be any more expensive at all. Only problem at that point might be one of economy of scale. Selling millions of cameras each year probably makes the sensor cost very little per-camera, but if you're talking about making a dedicated large-sensor video camera, now you're talking about selling maybe 10,000 per year and so the cost per sensor might be much more.

Why can't I use the lenses already in existence for this sensor size, which is basically APS-C?

Which of them are dedicated video camera lenses? Which operate like a video camera's lens does?

 

[studio1972]

If it had a low-pixel-count large sensor, that may not have to be any more expensive at all. Only problem at that point might be one of economy of scale. Selling millions of cameras each year probably makes the sensor cost very little per-camera, but if you're talking about making a dedicated large-sensor video camera, now you're talking about selling maybe 10,000 per year and so the cost per sensor might be much more.




Which of them are dedicated video camera lenses? Which operate like a video camera's lens does?

Thx Barry, I was going to respond but you just said what I was going to say.