How does the FujiFilm S3 work?

[peter_lamotte]

The new Fujifilm S3 looks like it may be perfect for what I need. However I am confused as

to how the "R-Pixels" technology works. According the dpreview.com "The SR sensor's

unique double photodiode design will produce images with a dynamic range four times

greater than that of a sensor working on single-pixel technology." Could someone break

this down into laymen�s terms for me. How do the two sets of 6MP work? Is this really

comparable to 12MP, or is it maybe even better? I realize this is a new camera, but I didn't

know if this was new technology.

[carl smith]

Basically every photo site is split up in to two pixels. One larger pixel which is more sensitive and helps accumulate shadow detail and then the extended range pixel which is much smaller and should help prevent highlights from being blown out.

 

The combination of the different pixel sizes supposedly will allow the camera to better reproduce a larger range of tones without excessive noise or the highlights losing all detail.

 

If I remember correctly they also tried to pull off the stunt saying that since you now have twice the pixels you can get a 12mp image. While this is technically true it doesn't necessarily work that way because for every two pixels on this sensor the S2 had one, so the question ends up becoming does the lens have the resolution to provide that tiny extended range pixel with any new information? Also, under normal shooting conditions that secondary pixel is so small that it wont generally have a strong enough signal to noise ratio to be of much use.

[greg s]

If this link works, there is a 2 page presentation of the sensor basics. (This would load on Win@k for me, but not XP for some reason).

 

"4th generation Super CCD link"

 

http://www.photo.net/bboard/uploaded-file?bboard_upload_id=16482884

[berg_na]

The Fuji SR pixel features 2 separate light sensitive regions that collect light at different rates. This is illustrated by the line graphs in the Fuji brochure, you'll find that the the slope of the response of the S region is much steeper than that of the R region.

 

The S region is more sensitive but it saturates sooner. When it saturates, it cannot collect any more light, so the highlights are lost. At this point the less sensitive R region is still able to function and collect light. The final image is formed from the information collected in both regions, so both the shadows and highlights are better preserved than in a conventional pixel structure.

 

The 12 megapixel image is generated by interpolation and has nothing to do with the 2-diode per pixel structure, the current S2 which has a single photodiode per pixel can also produce 12 Mp images. Fuji claims that the tight packing density of the hexagonal pixel layout improves the quality of the interpolated image.

[peter_lamotte]

So correct me if I am wrong, but Fuji has invented a better 6MP camera, rather than adding

more mega pixels? They have in essence improved upon the S2 while not really chasing

the Megapixel race. In a way (and this is just me theorizing) isn't this a path that all the

manufactures will eventually take. I mean, once we hit the "more megapixels than we

need" stage, the only thing left to do is improve the quality of the image? Just a thought.

[berg_na]

The main intent of Fuji's approach is to increase the dynamic range of the sensor, not its resolution, so I think that your conclusion is correct. Digital sensors typically have a narrower range of tonal values, from the darkest shadows to the brightest highlights, than negative film, and various techniques have been developed to address this issue.

[carl smith]

It isn't designed to increase spatial resolution specifically, however some of their papers about the sensor claim it can be used in a true 12mp mode. I have trouble seeing that mode actually produce anything more than a placebo effect.

[jason_busch1]

I've been having an interesting time with Fuji. They refuse to tell me the pixel pitch of their S2, and the S3. I've sent them dozens of emails, and even had 3rd party media outlets email them. Still, no reply. Why is this? Well, it has to do with a few things. I suspect in the S3, their so-called pixels have a smaller pixel area, even if you combine the S & R photodiodes. And this is another thing. Fuji calls these diodes pixels. In fact, Fuji has to make a reference to CIPA regulations for the pixel definition--in other words, they are splitting hairs and hoping you don't notice. If you want to know how the S3 will most likely fair, read Phil Askey's review on the Fuji F700. That camera has a SR sensor, and Phil noticed no improvements in this sensor over other CCD or CMOS sensors.

 

I'm even going to say I think the S2 will fair better than the S3. Why? Well, it goes back to physics. Since Fuji is not helping me at all, I have to guess if they are using pixels or photosensors/photodiodes. You see, a company named Foveon uses 3 photosensors per pixel, to get their image. And the image quality is very good (ISO limitations aside). However, Foveon is using the number of photosensors to represent their total megapixel resolution, which is entirely false (Polaroid's x530--Foveon imager inside--is touted as a 4.5MP image but it's actually only a 1.5MP image). You could theoretically have 100 photosensors per pixel, but only one pixel is resolved from those photosensors. So, suppose I have a 1MP chip and I have 100 photosensors per pixel. According to Foveon, I have a 100MP image. Fuji is doing the same thing, but they are covering their tracks in the sense they aren't being public with their pixel sizes. Every single company I have contacted to date is very nice with giving me information on the pixel sizes (Canon, Nikon, Olympus, Sigma). Fuji simply doesn't respond at all.

 

I think Fuji's "pixels" are actually photosensors. They have two photosensors per pixel. They have split their usual photosensor, into two parts. But this is problematic. The reason is, the area from the photosensor has been decreased. Generally, the photosensor is about as large as the pixel. Makes sense. But Fuji decided to split the photosensors into two pieces, leaving gaps in between those photosensors. This results in more vulnerability to the common attributes of any sensor, and negates the extra photosensor theory.

 

I would to add that this is just my observation and my experience with Fuji. Once Phil and other reviewers take a good look at the S3, we will know more. But I am really getting irritated with companies like Fuji and Foveon who are taking advantage of clever nomenclature, to disguise their sensors as something they are not. If Fuji would just say, "Hey, the S3 is a 6MP DSLR, and we have two unique photosensors processing the information instead of the usual one photosensor we had in the S2", I wouldn't have a problem at all. But saying one hand that they are not caving-in to the other manufacturers by having higher megapixel cameras, and then stating your S3 is a 12 effective megapixel DSLR, is not only ironic, but misleading. Is the S3 going to be a nice DSLR? I'm sure it is. But it will be far from the TRUE resolution of a 12MP image.

 

So, I don't want to irritate Fuji DSLR owners, but I thought I would share my thoughts on this subject.

[mva]

Hello,

Without going too much deep, and without comparing S3 to any other camera, I can say that I've compared the 12MP images coming out of the camera with the same ones downsampled to 6MP, and I see no difference at all :-( So for me this means that it is not useful to store the photos as 12MP TIFs.