SONY must stop the noise!

Quote:

---

Originally Posted by dr4gon

You mean a900 and d3x?

---

No, at this point my 10.2 MP is quite enough, and I'd rather buy a vertical grip for the D700 for a 10th of the price of the d3x. D3X is an incredible camera, but the D700 is the same features, probably better fps, smaller, and way cheaper. What you lose with the D3 or D700 is the resolution. Thats major for some people, but not for me... yet.

So e-dawg makes a good point I haven't really been thinking about. This is merely a software/algorithm thing. Sony makes most or all of Nikon's sensors. With good glass a D900 should be able to take anything any Nikon can take. Better since they (theoretically) have access to the engineers that design the sensors.

It seems like Sony should be able to do what e-dawg said, poach a few engineers who know how to write noise reduction algorithms, release firmware updates, and sell the best bodies on the market.

To be sure it's more involved than that. I actually had to do a dinky sharpening and softening algorithm for an introductory C++ class once, and I got a taste of how complicated it can be.

But at any rate it seems odd. And I see Nikon and Canon pictures at ISO 3200 that look better than my A300 at ISO 800.

Good on you Don, this is the same prob I mentioned in that other thread.
I don't know why they bother to have ISO above 400 because the pictures come out looking like photos from a 1MP webcam :p

Quote:

---

Originally Posted by laydros

No, at this point my 10.2 MP is quite enough, and I'd rather buy a vertical grip for the D700 for a 10th of the price of the d3x. D3X is an incredible camera, but the D700 is the same features, probably better fps, smaller, and way cheaper. What you lose with the D3 or D700 is the resolution. Thats major for some people, but not for me... yet.

So e-dawg makes a good point I haven't really been thinking about. This is merely a software/algorithm thing. Sony makes most or all of Nikon's sensors. With good glass a D900 should be able to take anything any Nikon can take. Better since they (theoretically) have access to the engineers that design the sensors.

It seems like Sony should be able to do what e-dawg said, poach a few engineers who know how to write noise reduction algorithms, release firmware updates, and sell the best bodies on the market.

To be sure it's more involved than that. I actually had to do a dinky sharpening and softening algorithm for an introductory C++ class once, and I got a taste of how complicated it can be.

But at any rate it seems odd. And I see Nikon and Canon pictures at ISO 3200 that look better than my A300 at ISO 800.

---

I mentioned that since you said if price was no object. But isn't the D700's sensor the same as the A700?

The D700 is the same full frame as the D3. The D300 might be the same as the A700, but I'm not sure. I expect the new 24 MP sensor in the $8000 D3X is the same one in the A900 that many consider overpriced for $3000.

It's also possible that Sony makes a whole other line of sensors custom for Nikon than they use in their own cameras, but that seems stupid.

The Nikon's are Sony Sensors (A900/D3x) and I'm pretty sure the D700 is too, not sure about the D300.

Quote:

---

Originally Posted by dr4gon

The Nikon's are Sony Sensors (A900/D3x) and I'm pretty sure the D700 is too, not sure about the D300.

---

Here's what I know about Nikon and Sony sensors (Most of it is based on what I have read, as I am not privy to the actual arrangements that are in place, but I have combed through most of what is available and have attempted to separate fact from fiction):

• They are all at least based on Sony sensors, although not all of them are Sony sensors.
  
• Nikon's 12 MP FF sensor (D3 and D700) is actually a Nikon sensor, but rumour has it that it is manufactured for Nikon by someone else (probably Sony).
  
• I think the other bodies use Sony sensors (in both design and manufacture), but Nikon does request certain specs and designs that are different from the ones Sony uses in their bodies.
  
• For example, the sensor in the D300 is the same base sensor (the sensel, or silicon light sensing surface and its photosite structure) as in the A700, but the D300 sensor has 14-bit A/D conversion and probably its own microlenses, AA and IR filters, Bayer filter (and thus demosaicing algorithms in conversion).

I have been tweaking, adjusting and playing with my A700, for several weeks now for 'acceptable' low light no flash indoors sports photos. I have been impressed with what I have been able to do so far, Then I look at the images on my buddys canon, and this other guys pentax, our photos are nice, pretty bright, good color the lighting isn't to weird, so far so good, then the NOISE, I am at ISO 2000, 2500, 3200, they are at 1600 1800 and far less noise!! but hey i am having fun! and they go and switch over to flash units for the serious shots.

The poor "Noise" performance of the A700 really lets the side down.

I knew about the problem when I purchased the A700 but, at the time I thought that future firmware updates would easily address the issue.

Now I know a bit more it sems that, unlike Canon and Nikon, the analog/digital conversion happens in two stages in the A700. The first stage is built into the Exmoor Sensor while the second bit is done in the Bionz processor. This means that the RAW files already have some noise reduction applied, again unlike the competition.

What this means in terms of a Firmware fix, I'm not entirely sure but I now suspect that the first bit of the process can't be fixed. If so that's bad news and may mean that the A700 may never compete on level terms.

Quote:

---

Originally Posted by Peekayoh

Now I know a bit more it sems that, unlike Canon and Nikon, the analog/digital conversion happens in two stages in the A700. The first stage is built into the Exmoor Sensor while the second bit is done in the Bionz processor. This means that the RAW files already have some noise reduction applied, again unlike the competition.

---

Technically, all sensors have some level of NR and signal manipulation applied on-chip -- the option is usually to turn "off" the inter-pixel NR (which the A700 never had until the v4 firmware, but with it off, it should be comparable to other mfrs in that respect), but per-pixel NR remains active with any mfrs' sensors.

Filtration is always applied at the analog stage (i.e., after the light "signal" is generated by the photosite but before amplification and A/D conversion) and NR is often applied at the digital stage (i.e., after the on-chip A/D conversion but before readout from the chip) too. Both the amount and type of NR applied can vary between manufacturers.

Quote:

---

What this means in terms of a Firmware fix, I'm not entirely sure but I now suspect that the first bit of the process can't be fixed. If so that's bad news and may mean that the A700 may never compete on level terms.

---

Now that I have had time to think of this some more, many "hardware"-related things can affect the signal produced by the imaging sensor... unfortunately, most of them can't be changed by a firmware update:

• the aggressiveness of the AA filter (aggressive AA filters reduce detail before it even reaches the sensor)
• the transmittance vs colour response characteristics of the Bayer filter (some designs may allow less light to pass through)
• the light-gathering ability of the microlenses (they focus light from the surface of the sensor into the photosite "wells")
• the characteristics and quality of the analog filtration and amplification stage (analog signals are filtered and amplified before conversion; raising the ISO mostly achieved through increasing the gain on the analog amplifiers at this stage)
• the performance of the A/D converters
• the on-board digital signal processors (which will perform the NR)
• the electrical engineering of the sensor and its environment (circuit design, power supply, and grounding, EMI/RFI, and shielding concerns)

Some of these elements are different between Nikon and Sony versions of the same sensor, and it is possible that some of these elements are suboptimal on the Sony Exmor sensor relative to other manufacturers, which conspire to produce an inferior result.

Don't forget that Canon has been a leader in CMOS sensor technology for years before Sony (and therefore Nikon) joined the party. In fact, the 12 MP Sony CMOS sensor found in the D300 and A700 is arguably their first "successful" CMOS sensor, while Canon has been pumping out good ones since the 20D.

Again, a hardware-upgrade seems totally possible, with replacement of the SSS sensor. Let's face it, it is not "welded" down ...
because it moves! :eek: (who knew?) It is not part of the circuit board. If SONY were to release a "pro" level upgrade to the popular
α700 ... we could get some more mileage out of these cameras BEFORE having to purchase a whole new unit and effectively
tossing the old one into the "bit bucket" of dust and rememberance.

I don't know about you, because lenses are one thing, but it does not make me all that happy to have all this older body hardware
constantly piling up. Along with the α700 & α100, I have 7 camera bodies ... enough, I say!

http://i147.photobucket.com/albums/r...f-the-line.jpg

The idea of being able to actively recycle the DSLR with a quick internal parts exchange is a lot more desireable. I do not have to
relearn the position of controls ... and other nuance changes. The α700's 3-inch LCD is about as big as practical, at this point.