Crazy comparison: sensors - Pentax Q vs Mamiya ZD (and others in between)

Started Jan 22, 2013 | Discussions
peevee1 Veteran Member • Posts: 6,247
Crazy comparison: sensors - Pentax Q vs Mamiya ZD (and others in between)

Disclaimer: this is mostly for fun, but also try and see through the marketing talk about latest and greatest technology in some very expensive cameras. If you are not inclined to have some fun with numbers, better skip it. Some say the modern sensors in those expensive cameras are as good as they get, let's see...

Benchmark: DxOMark's "low-light ISO" score. As DxOMark explains, it "An SNR value of 30dB means excellent image quality. Thus low-light ISO is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of 9 EVs and a color depth of 18bits." 9 EV and 18 bits is certainly well above abilities of paper, ink, printers and LCD monitors to correctly reproduce the pictures.

Looks like a combined metric corresponding to how well the sensor converts light it gets into useful signal, vs just noise. As the amount of light which gets to the sensor is proportional to sensor area (given the same scene illumination and lens transmissivity), I will divide the "low-light ISO" score by the sensor area to see the sensor efficiency.

1. First contestant - Pentax Q, 2011. Sensor used is 1/2.3" class, 6.17mm x 4.55 mm, 28 sq. mm 12 mpix BSI_CMOS. I am pretty sure it is the same 3 y/o Sony unit first appearing in tiny Sony WX5 and TX9, and then in many other point-and-shoots ignored by DxOMark. It is certainly not state of the art, since superseded by 14 mpix, 16 pmix and 18 mpix sensors. But in Pentax Q it was analyzed, getting ISO score of 189, or 6.7 ISO/sq.mm. Let's take it as 100% of P&S efficiency.

2. Next up is Fujifilm X10. So called 2/3" sensor, 8.8 x 6.6 mm, 58 sq.mm. (or 2.1 times the area of Pentax Q sensor). 245 DxOMark ISO, 4.2 ISO/sq.mm, or 63% efficiency.

3. Nikon V2. 116 sq.mm (4.1 times Pentax Q), 403 ISO, 3.5 ISO/sq.mm, 52%.

4. Olympus E-PM2. 225 sq.mm (8 times Q), 932 ISO, 4.1 ISO/sq.mm, 62%.

5. Canon 7D, 332 sq.mm (11.8 times Q),  854 ISO,  2.6 ISO/sq.mm, 38%.

6. Nikon D800, 862 sq.mm (30.7 times Q), 2853 ISO,  3.3 ISO/sq.mm, 49%.

7. Canon 5D Mark III, 864 sq.mm (30.8 times Q), 2293 ISO,  2.7 ISO/sq.mm, 39%.

8. Leica M9, 864 sq.mm (30.8 times Q), 884 ISO, 1.0 ISO/sq.mm, 15%.

9. Pentax 645D, 1452 sq.mm (51.7 times Q), 1262 ISO, 0.9 ISO/sq.mm, 13%.

10. Finally, with huge sensor, Mamiya ZD back, 1728 sq.mm (61.6 times Q), 245 ISO, 0.1 ISO/sq.mm, 2%.

Fun. Obviously, 4 y/o Canon 7D and newish Canon 5D3 use the same technology. Ancient, but not nearly as ancient as the Kodak tech in Leica M9 and Pentax 645D.

How ancient? Electronics factories switch to new node every 2 years, with low power consumption=lower heat dissipation=lower noise. Now it is 22nm (for processors, regular simple structures like RAM, flash and sensors can use even smaller node, but let's go with processor tech). 3 years ago it was 32nm, 5 years 45 nm, 7 years 60-65 nm, 9 years 90 nm, 11 years 130nm, 13 years 180 nm (this is the Sony big sensor and Canon small sensor node), 250 nm in 1997, 350 in 1995, 500 nm (Canon large sensor node) about 1993? 20 years old? I don't know of what junk the Kodak sensors are made of, let alone the Mamiya.

Steen Bay Veteran Member • Posts: 7,418
Re: Crazy comparison: sensors - Pentax Q vs Mamiya ZD (and others in between)
1

Interesting. Worth to notice though that the Pextax Q RAWs are 'cooked' (or "smoothed") a bit, according to DxO.

OP peevee1 Veteran Member • Posts: 6,247
Re: Crazy comparison: sensors - Pentax Q vs Mamiya ZD (and others in between)

Steen Bay wrote:

Interesting. Worth to notice though that the Pextax Q RAWs are 'cooked' (or "smoothed") a bit, according to DxO.

I remember that about Pentax K-5, not sure about Q. I guess the algorithm must be similar, and it did not help K-5 much compared to D7000 with the same sensor in the "low light ISO" metric, within the margin of error of DxOMark tests.

Steen Bay Veteran Member • Posts: 7,418
Re: Crazy comparison: sensors - Pentax Q vs Mamiya ZD (and others in between)

peevee1 wrote:

Steen Bay wrote:

Interesting. Worth to notice though that the Pextax Q RAWs are 'cooked' (or "smoothed") a bit, according to DxO.

I remember that about Pentax K-5, not sure about Q. I guess the algorithm must be similar, and it did not help K-5 much compared to D7000 with the same sensor in the "low light ISO" metric, within the margin of error of DxOMark tests.

Think it's likely that it helped Pentax Q quite a bit in this comparison. Isn't it almost too good to be true that the Pentax Q should have app. twice the QE of e.g. Nikon D800? (the D800's QE is 56% according to sensorgen.info)

OP peevee1 Veteran Member • Posts: 6,247
Re: Crazy comparison: sensors - Pentax Q vs Mamiya ZD (and others in between)

Steen Bay wrote:

peevee1 wrote:

Steen Bay wrote:

Interesting. Worth to notice though that the Pextax Q RAWs are 'cooked' (or "smoothed") a bit, according to DxO.

I remember that about Pentax K-5, not sure about Q. I guess the algorithm must be similar, and it did not help K-5 much compared to D7000 with the same sensor in the "low light ISO" metric, within the margin of error of DxOMark tests.

Think it's likely that it helped Pentax Q quite a bit in this comparison. Isn't it almost too good to be true that the Pentax Q should have app. twice the QE of e.g. Nikon D800? (the D800's QE is 56% according to sensorgen.info)

It is not all about QE, there is read noise too, right? It seems pretty easy too see that the tech used for bigger sensors is older, cheaper. The trend is clear.

I guess with the fall in P&S production some better electronics lines should become available for larger sensor production. It especially applies to Canon, as they apparently missed all the cellphone sensor production which took up the slack on the best sensor lines for Sony. Simple economics should force they hands and they should start production of some high-end APS-C sensors on the same lines which produce their pretty good BSI-CMOS P&S sensors and are underutilized now. 7D Mark II could be quite interesting. Rebels will still be produced on those ancient 500 nm lines for a while, they are dirt-cheap until parts for them are still available. It funny to imagine working there, production lines run by some old mainframes or mini-computers... How do you support a VAX? Or you just reengineer everything to run on a PC for a 20- y/o line? But it is very costly, consumers don't make their own production equipment usually...

Steen Bay Veteran Member • Posts: 7,418
Re: Crazy comparison: sensors - Pentax Q vs Mamiya ZD (and others in between)

peevee1 wrote:

Steen Bay wrote:

peevee1 wrote:

Steen Bay wrote:

Interesting. Worth to notice though that the Pextax Q RAWs are 'cooked' (or "smoothed") a bit, according to DxO.

I remember that about Pentax K-5, not sure about Q. I guess the algorithm must be similar, and it did not help K-5 much compared to D7000 with the same sensor in the "low light ISO" metric, within the margin of error of DxOMark tests.

Think it's likely that it helped Pentax Q quite a bit in this comparison. Isn't it almost too good to be true that the Pentax Q should have app. twice the QE of e.g. Nikon D800? (the D800's QE is 56% according to sensorgen.info)

It is not all about QE, there is read noise too, right? It seems pretty easy too see that the tech used for bigger sensors is older, cheaper. The trend is clear.

Yes, the read noise is important too, especially for the DR, but SNR 18% is determined almost entirely by the sensor size and QE. So assuming the same QE, then the D800 should have about 15 dB (= 5 stops) better SNR 18% than a 5.6x crop camera like the Pentax Q if comparing at the same measured ISO (= same exposure), but at measured ISO 100 the D800 is only app. 11 dB better (44 vs. 33 dB), meaning that the Pentax Q should have more that twice the QE of Nikon D800, and that simply isn't possible since the D800's QE is more than 50%.

(Notice that the SNR 18% graph shows the "Pentax Q smoothed" results)

I guess with the fall in P&S production some better electronics lines should become available for larger sensor production. It especially applies to Canon, as they apparently missed all the cellphone sensor production which took up the slack on the best sensor lines for Sony. Simple economics should force they hands and they should start production of some high-end APS-C sensors on the same lines which produce their pretty good BSI-CMOS P&S sensors and are underutilized now. 7D Mark II could be quite interesting. Rebels will still be produced on those ancient 500 nm lines for a while, they are dirt-cheap until parts for them are still available. It funny to imagine working there, production lines run by some old mainframes or mini-computers... How do you support a VAX? Or you just reengineer everything to run on a PC for a 20- y/o line? But it is very costly, consumers don't make their own production equipment usually...

Looking forward to see DxO's test of Fuji X20 that has the biggest BSI-CMOS sensor yet (2/3"). Though, not quite sure that we'll see a DxO test of the X20, because it has a X-Trans sensor (with a different CFA pattern than Bayer sensors), and they haven't tested any of the other X-Trans sensors  yet.

OP peevee1 Veteran Member • Posts: 6,247
Re: Crazy comparison: sensors - Pentax Q vs Mamiya ZD (and others in between)

I suspect the "QE of 56% for D800" is just an estimate with large margin of error.

Keyboard shortcuts:
FForum MMy threads