Amateur Photographer Magazine

Quote:

---

13.5Mp?!.....

Clearly pixel count still looks to be the the single most important factor in marketing compacts.

When, O Lord, when is one of the major manufacturers going to realise that there is a market crying out for far fewer but better quality pixels?

---

Post Extras:

Quote:

---

...the real limit on image quality isn't so much having too many pixels, but having not enough square inches of sensor.

---

That's a bit like saying "I'm not fat, I'm just too short for my weight"...



--------------------
My Flickr Photostream

Post Extras:

Quote:

---

Yes, I'm interested. The high pixel count isn't ideal, but for low ISO shooting it might be worthwhile. As the ISO climbs the real limt on image quality isn't so much having too many pixels, but having not enough square inches of sensor.

---

Post Extras:

Quote:

---

Quote:

---

Yes, I'm interested. The high pixel count isn't ideal, but for low ISO shooting it might be worthwhile. As the ISO climbs the real limt on image quality isn't so much having too many pixels, but having not enough square inches of sensor.

---

If you had done the side by side tests I did a month ago, you'd be horrified at the fall-off in quality at 400 ISO. Happy to post a comparison if anyone wants, but will dent your faith in Nikon. This features-chasing should not be at the expense of the basics that the tool needs to deliver.

---

Sorry, but nothing I've said is an issue of 'faith', but of directly relevant technical knowledge and expertise. Everyone knows that quality falls off as ISO increases - that's not under dispute. What I meant by the statement you quoted was that total signal-to-noise ratio (which is essentially what 'image quality' meant in the context of rising ISO) is not very much affected by pixel count. Both signal and noise are integrated over area in the final printing/viewing process, and pixel density effectively disappears from the equation. The Mk 1 eyeball, coupled with the brain's signal processing, integrates locally, with groups of pixels or dots on the paper merging together to produce a general impression of details, tones, etc. plus whatever visible noise there is.

So for fixed reproduction size - certainly for common print sizes - and assuming similar technology, pixel count isn't actually terribly important - the total signal is more or less the same, and the total noise is more or less the same. Only by improving the original signal level, either by reducing the ISO and giving more exposure or by increasing the sensor area, is that parity significantly disrupted.

Put it another way - gien two sensors of the same receptive area, and the same basic technology, but one having, let's say, 6MP and the other having 12MP, then if you either interpolate the larger file down to 6MP, or interpolate the smaller one up to 12MP, the resulting images will be fairly similar in their noise levels. Expect some difference, certainly, in favour of the lower original pixel count, but that difference will be relatively small when compared to the difference made by actually increasing/decreasing the sensor area, because that, rather than pixel count, is the real limiting factor as the ISO setting rises.

--------------------
Huw Evans.

Post Extras:

Quote:

---

Megapixel haystack aside, I like the camera. It's got a proper viewfinder,records RAW and has a manual mode apart from all the other bits. Big thumbs up to Nikon from me.

---

You don't half pay for all that, though. Have you seen the RRP ???

--------------------
Too many cameras, too many lenses.......

Post Extras:

Quote:

---

Sorry, but nothing I've said is an issue of 'faith', but of directly relevant technical knowledge and expertise. Everyone knows that quality falls off as ISO increases - that's not under dispute. What I meant by the statement you quoted was that total signal-to-noise ratio (which is essentially what 'image quality' meant in the context of rising ISO) is not very much affected by pixel count. Both signal and noise are integrated over area in the final printing/viewing process, and pixel density effectively disappears from the equation. The Mk 1 eyeball, coupled with the brain's signal processing, integrates locally, with groups of pixels or dots on the paper merging together to produce a general impression of details, tones, etc. plus whatever visible noise there is.

So for fixed reproduction size - certainly for common print sizes - and assuming similar technology, pixel count isn't actually terribly important - the total signal is more or less the same, and the total noise is more or less the same. Only by improving the original signal level, either by reducing the ISO and giving more exposure or by increasing the sensor area, is that parity significantly disrupted.

Put it another way - gien two sensors of the same receptive area, and the same basic technology, but one having, let's say, 6MP and the other having 12MP, then if you either interpolate the larger file down to 6MP, or interpolate the smaller one up to 12MP, the resulting images will be fairly similar in their noise levels. Expect some difference, certainly, in favour of the lower original pixel count, but that difference will be relatively small when compared to the difference made by actually increasing/decreasing the sensor area, because that, rather than pixel count, is the real limiting factor as the ISO setting rises.

---

Post Extras:

Quote:

---

Just did the same comparison with the Leica Digilux 3 and that beats the things that beat the 5100....by a long way.

---

Well it probably would, wouldn't it - for precisely the reasons I discussed above. The Digilux 3 has a 4/3" sensor, which is huge by comparison to the 1/1.7" sensor of the P5100 - about 6x the receptive area. What did you expect?

--------------------
Huw Evans.

Post Extras:

Quote:

---

Quote:

---

Just did the same comparison with the Leica Digilux 3 and that beats the things that beat the 5100....by a long way.

---

Well it probably would, wouldn't it - for precisely the reasons I discussed above. The Digilux 3 has a 4/3" sensor, which is huge by comparison to the 1/1.7" sensor of the P5100 - about 6x the receptive area. What did you expect?

---

Sorry, mean the D-Lux 3. 1/1.65 CCD according to the spec.

--------------------
Mike

My flickr site: http://www.flickr.com/photos/rovingmike/sets/

Post Extras:

Quote:

---

The D-Lux 3 is (I think it's widely believed) identical under the skin to Panasonic's LX2, which has a fairly poor reputation for noise.

---

Mechanically and optically I think they pretty much are the same but IIRC the image processing side is different. I imagine that the Leica probably is as noisy as the Panny but I would strongly suspect that the noise-reduction routines in the Leica are more sophisticated in terms of maintaining resolution and detail. I know from my FX30 that the NR routine can be a bit heavy handed (and patchy). TBH I've come to the conclusion that sometimes the 'noise problem' is actually made worse/more obvious by rather clumsy attempts to alleviate it...

--------------------
Nigel

Completely BSRIPN

ElSid Gallery

A camera in the hand is better than one in the cupboard........

Post Extras:

Quote:

---

TBH I've come to the conclusion that sometimes the 'noise problem' is actually made worse/more obvious by rather clumsy attempts to alleviate it...

---

Well I certainly leave all noise reduction options set firmly to 'off' - but then I have a DSP background! :-) Ultimately there is no difference between an electron that gets counted because a photon hit the sensor and one that arose from some electrical or other noise source - the camera or software cannot possibly know which is which. So all post-processing noise-reduction efforts inevitably process out detail as well as noise, because by that time they are quite simply the same thing.

One can potentially use some sort of subject recognition algorithms to spot things like, say, a clear blue sky, and then apply noise reduction more heavily in such areas, confident that there won't be much real detail lost. But that kind of approach almost inevitably runs the risk of producing 'cartoony' images. I'd rather my pictures were just noisy! :-)

As with grain on film, I think we still have pretty much the same three options, and for the same fundamental reason - keep the ISO low, get a bigger format camera, or learn to love it! :-)

--------------------
Huw Evans.

Post Extras:

Quote:

---

Mechanically and optically I think they pretty much are the same but IIRC the image processing side is different. I imagine that the Leica probably is as noisy as the Panny but I would strongly suspect that the noise-reduction routines in the Leica are more sophisticated in terms of maintaining resolution and detail.

---

Post Extras: