D800 RAW File Size

[AchimT]

There's one more configuration for image sizes, FX NEF 14 bit lossless compressed. This results in images of a little over 35 MB for me.

I'd like to comment on the sensor tech aspect. I've been told the D800 uses an expensive technology called Silicon on Insulator (SOI) which makes it possible that the D800 beats the D600 WRT DR and ISO performance. Apologies if everybody knew this or it is irrelevant, I have not personally read up on this to verify.

Regarding the pixel size's effect on detail. I understand where sk66 is coming from when he says smaller pixels will provide more detail. A smaller pixel would capture a smaller part of the overall image. So in the end, there would be finer detail in the image.

But I don't think it's true in the case of the D7100.

In photography, I think everything is about colour, right? Because that's what we see. Each sensor pixel stores a piece of colour information (which colour a point in the image had). A larger pixel, or one with better sensor technology, can store finer (= more) nuances of colour. And finer nuances of colour in photography I think mean more detail.

Because imagine this: if you have a lot of pixels showing very small pieces of a colourful surface, in theory you would have finer detail, but if those pixels can only store (just an example) 256 shades of colour each, then probably a lot of pixels would deliver the same colour information (i.e. a value between 1 and 256) to the processor.

If this is true, detail information would be lost, because an image is not created through the information of how many quadrants of space are occupied by an object. The detail in the image is created by the fine nuances of colours that distinguish one point of space from the next one.

I'm not an expert at all, so I may be completely wrong and in that case I apologize for wasting your time, but if the above is true, a sensor with larger pixels which can store more colour nuances, or one with pixels which can store more nuances of colours due to better technology, might deliver more detail information than one with very small pixels which lose colour information because they're not capable of storing all the nuances.

The precise transition ratios when this ratio of pixel count versus pixel size would start to turn around I don't know, of course.

Apologies to everybody if all the above is a load or ru....

 

[sk66]

In regards to sensor technology improving. I don't think Nikon's made any significant changes since the D3s. (The D800 is still a CMOS sensor AFAIK) The D4 only has a slight advantage due to it's slightly higher MP's when downsampled.

Sensor Pixels are not quite the same as image pixels. Each sensor pixel can only record brightness information for one color. Everything else is calculated in the conversion. But no, more MP's cannot bring out more detail where there is none.

Most of the image degradation from high ISO (loss of DR/Color sensitivity/noise/etc) comes from a lack of sensor site(pixel) saturation. And the randomness of light hitting the sensor in low light environments (photon noise). Imagine you have 3 cups set together on the ground and you throw three drops of rain at them. Each cup could have as many as 3 drops or as few as none assuming all three drops hit a cup. You have a wide variety of "saturation." Or, all three cups could catch one drop and each would be equally "saturated." Now consider a single much larger cup alone which catches all three drops. That single cup is more "saturated." In the previous senario you could have 1 cup with three drops (as equally saturated as the larger cup) but then the other two have none. That's essentially how photosites work at the analogue stage.

The rest of the issues arrise due to digital manipulation (the digital ISO ranges). Or maybe it would be more correct to say the same issues are "magnified."

My D800 is back and I was going to do the test shots. But I found this DP Review Studio Shot Comparison Tool which allows you to select a camera, an iso, and then view the raw files at 100% without resampling. Because they have not been resampled they show different "amounts" of the original image (but the same pixel dimensions). The link should open w/ the D4/700/800/7000 @6400 pre-selected. IMO, the examples are "generous" because they are of a well lit scene.

As I said before, you can gain a much better image by downsampling. But if you are going to downsample the question becomes "why?" Why deal with the large file sizes and everything else if another camera can do essentially the same with less hassles?

And you also have to keep in mind that doesn't account for cropping or anything else. Using the D800 in DX mode essentially gives me a 16MP D7000 image...on image IQ/noise/etc it's a pointless "upgrade" to only use the D800 images heavily cropped.

 

[AchimT]

I can't really comment on the noise issue, and I also don't want to, as it is not such an important aspect for me. I've seen comparisons where the D800 outshone the others, while your's show it the other way around; so I just don't know what's right or wrong, and why sometimes the one camera shines, and sometimes another. I guess its - apart from the test setups - maybe also down to one camera being better under certain conditions and worse under others?

What I can comment on is that it appears that the D4 uses a Nikon sensor, whereas the D800 uses a Sony sensor with Nikon circuitry and software. Apparently this sensor (Sony sensors are now being used everywhere left and right because they are so revolutionary, and Canon falls behind a bit in terms of DR because they don't use one and can't match it with their own) and circuitry were considered to be quite revolutionary in techie circles because they gave the ability to have this high DR using such small pixels. Note that the D600 has a bit less DR although its pixels are approx. 20% larger. It does have a less refined (and expensive) sensor (and maybe also less refined circuitry and software, this I don't know).

So yes, I think at the time of its release, the D800's sensor and circuitry were considered to be very new technology, and to an extent they still are.

Regarding the famous water cup analogy, AAMOF I think it could be understood as proving that potentially, larger pixels are better but seriously, I think you and I are both pretty much saying the same. Each pixel stores intensity information for the colour of the point of the image the lens throws onto it.

If you have more pixels to show any given sector in the image, you could indeed have more detail - as long as those pixels are also capable of distinguishing subtle differences in the colours or their intensities.

Which is why I think the D800 will show more detail - its pixels are capable of distinguishing more subtle nuances in colour intensities.

An please correct me if I'm wrong, but if I point an FX camera and a DX camera at the same scene from the same distance and angles, using lenses that result in the same net focus length (considering the conversion factor for DX/FX, i.e. e.g. a 50mm on DX, and an 85mm on FX?), will I not get images of the same size? Not filesize, but showing the same section of the world in front of me? If this is the case, then the D800 has more pixels to show it, and hence would have a finer resolution, even if its pixels are larger, and due to the larger pixels it can on top of it store more intensity detail for each point it records as well

That said, please take all of the above as being said in a friendly manner. I have no intention to be right or to defend the D800 at all cost. I just find it an interesting topic.

 

[G.K.Jnr.]

An please correct me if I'm wrong, but if I point an FX camera and a DX camera at the same scene from the same distance and angles, using lenses that result in the same net focus length (considering the conversion factor for DX/FX, i.e. e.g. a 50mm on DX, and an 85mm on FX?), will I not get images of the same size? Not filesize, but showing the same section of the world in front of me? If this is the case, then the D800 has more pixels to show it, and hence would have a finer resolution, even if its pixels are larger, and due to the larger pixels it can on top of it store more intensity detail for each point it records as well

Hi, You are perfectly correct.

That's why when film was the norm with all things being the same ie Equivalent Focal Length Lens/Film/Processing/ etc, etc the larger format camera would always produce better results.

In my book the same applies to Digital images.

 

[AchimT]

Thanks for confirming this George.

 

[sk66]

An please correct me if I'm wrong, but if I point an FX camera and a DX camera at the same scene from the same distance and angles, using lenses that result in the same net focus length (considering the conversion factor for DX/FX, i.e. e.g. a 50mm on DX, and an 85mm on FX?), will I not get images of the same size? Not filesize, but showing the same section of the world in front of me? If this is the case, then the D800 has more pixels to show it, and hence would have a finer resolution, even if its pixels are larger, and due to the larger pixels it can on top of it store more intensity detail for each point it records as well

AFAIK, the most "revolutionary" thing about the D800 was the number of pixels they could fit and the lack of an AA filter on the E model. I don't *think* the D800 is using Sony's "stacked" sensor technology (which is really just a size advancement for phones).
The main reason the D800 *can* outperform all other sensors is "oversampling." That's essentially collecting a ton of information and compressing it, and that's what happens when you "down sample" an image (i.e. use it at a size less than 100%). The more information you "compress" the greater the color bit depth, DR, etc..

As to crop factor and detail....it's not that simple really. As far as detail due to MP's it comes down to how many pixels remain in the captured image. I.e. If I use my FF D4 (16MP) against a DX D7000 (16MP) then the pixel density(relative size) will be equal between the two captures. D800 against any other DX (currently) and it will have more pixels. Compare the D800 to an even smaller sensor and it may have fewer pixels.

BUT, in order to capture the same FOV I also have to use a lens of greater magnification on the FF camera. That means the details as presented to the sensor are actually larger. If details are too small to discern, then there is no detail regardless of how many MP's are displaying it. The bigger lens usually wins. This little difference is why crop factor/MP's is not the same as focal length.

And then there's "perceived" detail/sharpness which is largely due to contrast and not actual resolution/MP's. And larger photosites which gather more information have more "contrast" (and everything else) at their native size.

Smaller photo sites/higher MP's *require* more light to perform their best. If you don't have that you can downsample the mage to bring the quality back up (essentially using the "3 cups" as if they were one). Higher MP's also require faster SS's in order to handhold.

I've got no problem w/ the D800. I own and use one. The D800 is great when it can be used at it's best. Low ISO, high SS/tripod, "correct" focal length. When it can't be used "ideally" you can "throw away" the extra MP's by downsampling to bring the quality back up. Currently, if you throw away those "extra" MP's you get a minor improvement in image quality (DR/color) over most other cameras, but there's a lot of hassles for that gain.

If the type of work you do is typically outside of the "ideals" for the D800, then it's probably the wrong camera. There's absolutely nothing wrong w/ a 12MP image for most uses. An extra stop of DR, or bit of color depth probably won't even be discernible by most, certainly not in an 8x10 print or 1024x display.

 

[AchimT]

...in order to capture the same FOV I also have to use a lens of greater magnification on the FF camera. That means the details as presented to the sensor are actually larger...

That would mean that on an FF camera you can have larger pixels and they will still capture the same degree of environmental visual detail as smaller pixels would on an APS-C camera?

Well, I think we won't agree on these questions, so maybe let's just agree to disagree. Thanks for all the info you provided!

 

[sk66]

That would mean that on an FF camera you can have larger pixels and they will still capture the same degree of environmental visual detail as smaller pixels would on an APS-C camera?

Yes, *if* you make those details larger by using a longer lens on the FF camera.
Assuming the same final MP size (i.e. D7000/D4):
With a DX body you are using smaller pixels to resolve smaller details from a smaller FOV.
With a FF body and a longer lens you are magnifying the "smaller FOV" so that it fills the frame. You are using larger pixels to resolve larger details from the same smaller FOV.

The end result is that a detail that occupies 1px in one capture also occupies 1px in the other and the resolved detail is the same.
The problem (and why many use DX/DX mode) is that long FF lenses of high IQ are VERY expensive. You could buy 3 D800's for the price of my 400 f/2.8.