Petition to Nikon: low-light DX camera

Nigelous

Perhaps this test inaccurate and sony does not win. But nobody loses, approximate parity in sensitivity. I just wanted to prove that it is technically possible to produce low-noise sensor APS-C albeit at a lower resolution, but with a sensitivity of not worse than D4. By the way to compare the noise of the video is quite reasonable because the image is read from one and the same pixel count of 1920x1080. Look at another example, what can the not-full-frame sensor with large pixels in low light

Msmoto

@ SquamishPhoto

Thank you...I had no idea....a video camera for USD$6500...

I think the comparison with a flame remains as less than scientific.
And, to you, Nigelous, you probably need to answer the initial question I asked you about the link to the petition. Is this yours or not?

TaoTeJared

@TaoTeJared Those numbers having nothing to do with a rating of the cameras sensor, but rather the QE depending on the size of the sensor. Based on QE, and the size 1/1.7" sensor size, the G15 is 59% efficient. The 5D MKIII is 49% at it's size. I think you've totally missed the point of the site altogether.

Think of it this way.
D4 53% Efficiency: Meaning the sensor still has room to improve at lot, based on the Bayer design.
D5200 65% Efficiency: Meaning it has less room to improve because, based on the sensor size it is closer to being 100% efficient.

Does that clear things up a little better? This shows how much room the sensors have left to improve, and how efficient the technology is, at a given photosensitive surface (photosite). The FX sensors are less developed than the DX's, yet they still outperform them.

Dang it - I answered the phone during the post and left a whole thought process out. (DXOs numbers don't actually measure electrons.) I didn't miss the point at all, but I do not think it can be called "quantum efficiency" as that is a actual measurement of electrons and could not be used as a "catch all" term from what I have always understood. It is an "Efficiency of something" rating, but just not "quantum efficiency."

It is interesting for sure. I can see that as one part of the actual equation as photo sites are not just "holes" and have various parts, almost gap-less Micro lenses for example, that help increase the light gathering.

Meinrad

@TaoTeJared: You are quite right when you say that the quoted measure is not true "quantum efficiency". A quantum efficiency of unity (QE=1) means that every photon triggers an electron that does not get lost on its way to the analog-to-digital converter. But: in a Bayer array, 2/3 of the photons have the "wrong" color for the filtered photosite (i.e. they are green or blue, when striking the red site, for example) and are filtered out. That drops the QE by definition below 0.333 to begin with. Add losses (especially at the long wave end, where the energy is marginal to straddle the band-gap), imperfections of the micro-lens screen, lattice gaps between the micro-lenses, filter imperfections etc, and I would bet a large sum of money that even a D4 is not at QE=0.1 yet.
The Foveon principle has the potential to improve this (because the filters are "switches", rather than absorbers of photons). Somehow, I doubt that the current developers of the Foveon have the necessary R&D muscle to utilize that inherent advantage.

PB_PM

The math might not be perfect (no idea, I'm not a math wiz), but they are all coming from an equal playing field, so it does give an idea of where things are.

Nigelous

There is 2 ways to boost SNR:
1) improvement Q.E. (very difficult and more expensive);
2) increase of the microlens area (easy unexpensive way).
By the way, the sensors majority are gapless, include D4. Therefore pixel area must be increase for better sensivity.

Godless

There are a few low-light Nikon DX cameras already: D7000 and D5200. And soon there will be a D7100 as well. If you need more sensitivity, get a D600 and/or better lenses (some f/1.x stuff will help)

TaoTeJared

I too think the Foveon had the "future" design - it is a shame Sigma has it where it doesn't get the due research support backing of a much larger company. As it is such a different design, I'm not sure what the limitations are for low light and why it does not perform well. I love the photos from the sensor though. I just don't have the patience or the cash to buy any of their camera's for such a limited working use.

Most people are forgetting that back-end software/firmware right now is the King on handling noise. If you look at the Sony vs. Nikon where the sensors are in the same release generation, Nikon almost always trumps Sony to see it is software gains. At some point, sensors will leap foreword (i.e. D3 sensor) again but then they will get 2-3 years of releases where tweaked software will really be the "big gains."

Personally I think once they get a processor in the camera that is just for "editing" then we will see massive gains. But I'm sure that is release 1-2 generations away.

donaldejose

Very good point about a separate processor in the body just to edit the image. Denoising can be done right in camera and automatically be set to on when ISO reaches a certain level.

PB_PM

I don't see much future in Feveon sensors, it just doesn't have the performance. There is a reason that Nikon, Panasonic, Canon and other major chip manufactures have largely ignored it, despite some of them having patents for similar technology. The D2H, for example, had a Feveon like sensor, but for some reason Nikon decided not to continue development.

If anything a sensor like Fuji's X-Trans are most likely to be the future.

TaoTeJared

If you figure big money/research has gone into Bayer sensors for almost 40 years now and we are where we are at with them & Foveon is just over 16 years old (+/- a few years) and has not had any "real" money thrown at it, I would say it has done quite well. It seems Sony and Canon both threw out patients in the last 2 years that are similar so they must think something is out there about it. Only time will tell.

Msmoto

Thank you Harold

Nigelous

There are a few low-light Nikon DX cameras already: D7000 and D5200. And soon there will be a D7100 as well. If you need more sensitivity, get a D600 and/or better lenses (some f/1.x stuff will help)

CAP reminds: D7000/5200 worse than D4 about 3 stops sensitivity because it have very high pixel density, the same as D800. Also, their sensors optimazed for low ISO. A small sensor must have smaller resolution for save the same sensitivity.

Nigelous

By the way, D600's video noise similar D7000. And 12 MP D3s from 2009 kills D600 of 2012. Don't believe DxO. It is corrupted by Nikon. Believe your eyes!

heartyfisher

I had a look (and think) at the the faveon sensor when it was introduced a few years ago. I thought it was very interesting.. I believe the reason that it is weak in low light may be because of how it works.
As you may know it works with 3 layers of the same light sensitive silicon.
1) the first layer gets all the light hitting the sensor and absorbs all colours
2) the second layer below gets part of the light that is not absorbed by the first layer.
3) The third layer absorbs the light that is left over from the first and second layers.
The reason that works is silicon, like water, absorbs the lower frequency more so than the higher frequency light. . if you have ever been in one of those tourist under water submarine tours you will see that the deeper it goes the less red you can see.. and at the deep parts you only see blue and UV(you know .. glowing clothes).

So when processing the data from the 3 layers they need to subtract the value(by some factor) of the lower layers from the upper layers to find the value of the red colour, and green. so you are throwing away some volume of light from the upper layers. I think theoretically the volume of light "lost" this way is still less than in the bayer system.. but who knows.. maybe its not as theoretically predicted.. maybe too much of the light is really lost as its goes through the silicon, converted to heat not electricity( the Q.E ?).

Dredden85

@ Nigelous
We all want the most bang for our buck. Let me know when the D4 is $399 and Harry Potter can wave his wand and remove every ounce of noise out of the shot! Only then we'll all be happy!
I'm not the sheep that follows you and signs your stinky petition. First, there must be a real, accurate, and scientific comparisons.
+1 heartyfisher

heartyfisher

@Dredden85 lol and i bet you don't vote either !

Msmoto

@ Nigelous

Your quote:
"Don't believe DxO. It is corrupted by Nikon. Believe your eyes!"

Do you have evidence of this? In doing my research on DxOMark, I could find nothing to suggest they are "corrupted by Nikon".

PB_PM

Indeed. Look at the lens ratings, if DxO is corrupted by Nikon why are many Canon, and for that matter Sigma, lenses rated higher?

Nigelous

Dredden85, Flock are follow the shepherd in the face advertising and marketing. Do You support megapixel war?
Do you justify 24 MP in APS-C?
Msmoto, Why the D600 sensor have higher ISO than D4? This contradicts the reality. Why the Nikon's cameras are testing immediately after the announcement and Canon's too late? Also there is information about underestimate scores of 1D X.

Dredden85

@ nigelous: Megapixel war? I don't care. A 24MP in a APS-C, sure why not. I care about performance and if my pictures look great (and if the darn thing focuses). I feel its more about composition than gear. Yes, good gear helps but technique is half the picture.
Why are you so driven to strong arm us, to see it your way and to agree with you? I found out about my products through research; not an advertisement or marketing.

Nigelous

If you don't care, why don't you ignore this thread?

spraynpray

OK girls, stop slugging it out with your handbags before one of you smudges your lipstick! 8-|

@Nigelous: You will get better responses if you can be less confrontational - take your time before you post.

MikeGunter

Hi all,

The current crop of DX cameras is pretty good.

I don't see Nikon or any of the others jumping into the fray to improve upon them much. The current trend is in the FX world. The tea leaves point to more and better FX cameras.

Mene, Mene, Tekel, u-Pharsin

The current DX cameras will be as good as they are, but they wont be any better. I plan to wear my out, then upgrade to FX. I hope I'm wrong. The rumors suggest a D7200. We'll see.

My best,

Mike

Msmoto

@ Nigelous

You have presented no objective data to support your contention "Don't believe DxO. It is corrupted by Nikon"

If you have some reference articles which support your position, present these. Otherwise, this is only your opinion.