David Fincher's The Social Network

[John Sprung]

2. Large sensors - beyond the obvious reasons (cost, size, lenses, DoF) the production process of sensors becomes the limit. I think 20x30mm is the max. size given by the production equipment ....

 

Chips are made from large wafers. They put as many as they can on a wafer, which can now be as much as 460mm in diameter. The downside to big chips is that the percentage of good ones per wafer goes down as the individual chips get larger. Each speck of dust ruins one chip, typically. So, for much larger chips, the economics become unfavorable.

 

 

 

-- J.S.

[georg lamshöft]

Wafers have very large diameters, that's correct. It's mainly an economic thing for very high quantities, many high-end-fabs still use smaller 150mm-wafers suitable for smaller production numbers of high-quality sensors - it's not limiting the quality of the sensor itself.

But the "stepping process" (I'm not sure about the correct English vocabulary) is made with smaller entities "projected" onto the wafer and creating the structure of the chip/sensor. This size is limited (especially by the optics) to about 20x30mm - larger sensors are made from one wafer but stitched from multiple "steps".

The problems caused by stitching (I've seen images from MFDBs with visible "stitches") are accepted in the photographic world (just like dead photosites are mapped out in firmware), but professional digital cinematography is another story.

[Marc Roessler]

Cost rises exponentially as sensor size increases. That's why large sensors are so ridiculously expensive (assuming that you don't stitch). And that's also why I'm curiously awaiting RED's 617 sensor, and it's price tag... Or have the plans been dropped already? Haven't found anything about this on the red webpage any more.

 

To my eye, film still looks better. It is not only about the resolution. For example for lighting a film Robert Richardson style, film is really the easiest (and maybe only) way to go. I think that film is far from dead.

 

Other than that, choice is always good. And the Arri Alexa series will be an excellent choice for those who prefer to shoot digitally. Having been aboard the Alexa development team from day one (and D21, before that) I think I can safely state that.

 

Greetings,

Marc

Edited March 22, 2010 by Marc Roessler

[georg lamshöft]

That's true, but as I said, every larger sensor is stitched. I don't think you can physically "fuse" two sensors from entirely different places within a wafer (or even different wafers) that's why stitching doesn't reduce the problem of exponentionally rising costs.

RED is buying the sensors from an unknown source, stitching is a very specific technology which are only a few fabs capable of (DALSA, Kodak, Sony, Canon and maybe Cypress? CMOS-based sensors could be manufactured by any halfway advanced fab around the globe) - let's see if their supplier is capable of delivering 6x17-sensors.

 

But to my eyes, sensor size isn't the issue with RED...

 

This film vs. digital-discussion might be boring and exhausting but when I see great artists affected by the digital hype, I cannot help to start it all over again because the very same arguments remain valid. :rolleyes:

[John Sprung]

Does anybody have a good reference on this stitching technology?

 

 

 

 

 

-- J.S.

[Marc Roessler]

That's right, stiching does not mean to "fuse" different parts of the wafer together. It's still one piece that's cut from the wafer. As Georg wrote, stitching is referring to the optical masks used, not the wafer itself. It thus doesn't bring down the exponential cost factor, it just lowers the total cost/effort. The way my post was written was a bit misleading in this regard.

 

Stitching is critical because you always have some "interesting" effects where the different sensor parts meet which are somewhat hard to get rid of.

 

Let's say, "the majority of the larger sensors are stitched"... it is not a must. You can see whether a sensor is stitched with your bare eye by examining it closely, by the way.

But we're straying too far from the intial topic I suppose.

 

Greetings,

Marc

[Guest Roland Zagonyi]

Here's a low light test shoot with the Mysterium X that Fincher did with DiCaprio - might explain things better for those who haven't seen it:

 

http://www.youtube.com/watch?v=jJpLgdD7r3c

 

The match is the only key light (and then the cigar).

[Vincent Sweeney]

I just saw the trailer for the Social Network in a commercial theater, equipped with Sony 4K projectors, and was a little surprised at how underexposed, soft and video-like it appeared.

 

However that may not have been as obvious to me if the trailer that had played before it wasn't Wall Street 2, which was obviously 35mm. The image quality and sharpness really stuck out on it, more than usual even.

 

This post helps explain to me why I took note of the differences.