When will sheet film be obsolete?

[Cor]

It goes on way too long to be a troll. My fingers are crossed that it's a brilliant creative piece, and that it just happens to be over my head.

Pheww, I feel better now, I could not understand it, and blamed it on my lack of understanding english..

Best,

Cor

[Greg Miller]

The Hasselblad H4D-60 has a sensor area of 40.2×53.7mm, ISO 50 - 800, and 1.4 sec/capture per frame. So this one is as close to 6x4.5 as you're going to get for now, all for a paltry $40,000. So extrapolating that technology, a 4x5 back would be 16,353 x 20,430 = 334,091,790 pixels, and cost $501,137,685,000. So call it at 334Mp for $500 billion.

Sure, in some future tomorrow, the InVisage QuantumFilm product may blow that through the roof. First it's for the cell phones, and then some time down the road after that, there will be a product for professional photographers. The marketing blurb says 4x, so the Hasselblad sensor would be 240Mp. But I can't see it ever resulting in a full 4x5 back.

That math doesn't quite work out. 4x5 = 102mm x 127mm. It would take about 6.25 40.2mm x 53.7mm chips to fill the same area. So the cost of a 4x5 chip (assuming the process scales up linearly, which it does not), would be closer to $250,000. Still not very affordable but nowhere near $500 billion.

[jonathan_lipkin]

Don't know if they have any plans to develop this commercially, but Canon has developed a huge 120MP sensor:

Canon has announced it has developed a 120 megapixel 29.2 x 20.2mm APS-H CMOS sensor - the same size used in its EOS-1D series of professional DSLRs. The sensor, for which Canon has announced no production plans, has a pixel count nearly 7.5 times larger than the company's highest pixel count commercially available sensor. It offers full HD recording (using 1/60th of its surface area) and can deliver 9.5fps continuous shooting. This follows a 50 million pixel sensor of similar format the company developed in 2007.

http://dpreview.com/news/1008/100824...20mpsensor.asp

[harrykauf]

smaller than 35mm is not "huge"

[paulr]

Is it the price of silicon or something else that makes it more attractive to make pixels smaller vs. sensors larger?

[Oren Grad]

Is it the price of silicon or something else that makes it more attractive to make pixels smaller vs. sensors larger?

Perhaps the number of non-fixable defects scales more rapidly with surface area than with pixel density, so the cost per usable sensor increases more rapidly with sensor size than you'd expect just from the cost of the wafer. To take an extreme hypothetical case, if a sensor takes up the entire wafer and you have one irremediable defect, that unit is lost, but if the same wafer can provide 20 small sensors, you might still have 19 usable pieces to sell.

[Arne Croell]

Perhaps the number of non-fixable defects scales more rapidly with surface area than with pixel density, so the cost per usable sensor increases more rapidly with sensor size than you'd expect just from the cost of the wafer. To take an extreme hypothetical case, if a sensor takes up the entire wafer and you have one irremediable defect, that unit is lost, but if the same wafer can provide 20 small sensors, you might still have 19 usable pieces to sell.

Right, and its not just the price of the polished or epitaxied silicon wafer as it comes from a manufacturer such as MEMC, Siltronic, or Shin-Etsu, but the additional cost of all or most the manufacturing steps to make the sensor also counts even if it has to be thrown away.
This would be an 8" wafer for a single 4x5 CCD or CMOS chip, as 6" is too close, and 7" is not a standard size. Alternatively, 2 or 3 4x5 chips would fit on a 12" wafer, where the price of the starting wafer is probably somewhere beween $50-100.

[John NYC]

Don't know if they have any plans to develop this commercially, but Canon has developed a huge 120MP sensor:

Canon has announced it has developed a 120 megapixel 29.2 x 20.2mm APS-H CMOS sensor - the same size used in its EOS-1D series of professional DSLRs. The sensor, for which Canon has announced no production plans, has a pixel count nearly 7.5 times larger than the company's highest pixel count commercially available sensor. It offers full HD recording (using 1/60th of its surface area) and can deliver 9.5fps continuous shooting. This follows a 50 million pixel sensor of similar format the company developed in 2007.

http://dpreview.com/news/1008/100824...20mpsensor.asp

You would think if the images were completely superior to anything today, they would be showing them everywhere as PR... if they were smart. Even if the company had no plans to put the sensor into production, Canon zealots would be using the image samples to validate to themselves and others that they went with the best company with the most assured future.

[dh003i]

smaller than 35mm is not "huge"

No, it isn't, but this is. Canon has just developed an 8 x 8 inch CMOS sensor.

[dh003i]

Perhaps the number of non-fixable defects scales more rapidly with surface area than with pixel density, so the cost per usable sensor increases more rapidly with sensor size than you'd expect just from the cost of the wafer. To take an extreme hypothetical case, if a sensor takes up the entire wafer and you have one irremediable defect, that unit is lost, but if the same wafer can provide 20 small sensors, you might still have 19 usable pieces to sell.

Why is the entire sensor lost? What do you mean by one irredeemable defect? One photosite not working, or a few photosites not working?

Why does the entire sensor need to be perfect? Does one non-working photosite cause the whole thing to not work? Or does one non-working photosite merely mean there is one non-working photosite (so what? why does everything need to be perfect?)

I'd gladly take a 4x5 digital sensor with 20 pixels that were always screwed up on every exposure.

Or put in more reasonable financial terms, if someone offered me a 36x24mm DSLR for the price of a normal APS-C or 4/3rds DSLR, around $800-$1200, I'd gladly snatch it up even if it had 20 non-working pixels.