Is there a market for an 8mp crop body with excellent iso performance?

[Loswr]

If the sensor size plays such an important role in good ISO performance, then why does most medium format cameras have a really bad ISO performance?

You're comparing CMOS with CCD, along with a completely different set of design/engineering priorities (for example, if you think there's not noise reduction baked in to the hardware/firmware of dSLRs, I hear there's a large bridge in New York up for sale).

 

[Pi]

If the sensor size plays such an important role in good ISO performance, then why does most medium format cameras have a really bad ISO performance?

Because it is not the only one. MF sensors have low QE, and not well implemented high ISO. They do not even have "high" ISO by today's standards. They are designed for studio and landscape use, have very low native ISO (around 30 for the IQ 180), very high DR there (comparable to Nikon/Sony), and impressive color sensitivity. This is a niche market, with no much demand for high ISO performance. The low native ISO offsets the increased shot noise when SS is not a problem, and everybody is happy.

 

[TAF]

Depends on how you define "excellent" performance.

Leaving aside the technical issues involved (as discussed above), I would think there would be a market for a camera (body) which yielded ISO 100 level of quality at say ISO 204800, even if it 'only' had 8 MP.

Shooting perfect pictures in the dark would be lots of fun.

Even better would be if it included a very high speed mode - say 10000 fps video capability.

That would sell quite well indeed, especially if it was under $1K.

I'm guessing it isn't going to happen any time soon.

 

[Pi]

I'm guessing it isn't going to happen any time soon.

Never, actually. The shot noise is part of the image itself, not related to the sensor at all. What sensor can do is to make it worse by not being able to record all the information. With the Bayer design, there is less than 1 stop left for improvement, even if the sensor were ideal.

 

[Pi]

You're comparing CMOS with CCD, along with a completely different set of design/engineering priorities (for example, if you think there's not noise reduction baked in to the hardware/firmware of dSLRs, I hear there's a large bridge in New York up for sale).

If you think that there is, you need to present an evidence. What I have seen is some NR at extremely high ISO. It can be seen in the FFT of a black frame (which does not say what happens when the frame is not black). Aside from that, I am not aware of any evidence for Canon, at least. Nikon cuts negative values in the RAW which is not exactly NR (it makes things worse, actually). DXO looks for correlation in the noise, and when they find it, they tell you that they did.

More about that:

http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p4.html

 

[Loswr]

Let me add some emphasis, since it seems I've confused at least one person.

[quote author=me]
You're comparing CMOS with CCD, along with a completely different set of design/engineering priorities (for example, if you think there's not noise reduction baked in to the hardware/firmware of dSLRs, I hear there's a large bridge in New York up for sale).
[/quote]

While fast frame rates and expanded ISO ranges aren't a priority for most MFDB users, it's pretty evident that dSLR sensors have been pushing into ever higher ISO ranges, since that's a priority to many dSLR users, and an area of active competition between manufacturers. Designers of sensor hardware are therefore going to strive for noise reduction, for example cleaner amplification circuitry and better heat management. Needless to say, such design choices cannot be measured by analyzing images from the sensor after it is produced, and I doubt anyone outside of the R&D shops has access to prototype sensors for comparison. If anyone thinks the only place dSLR makers think about noise reduction is after the RAW image is generated, I hear there's some swampland in Kansas available as a bundle purchase with the aforementioned bridge.

 

[Pi]

While fast frame rates and expanded ISO ranges aren't a priority for most MFDB users, it's pretty evident that dSLR sensors have been pushing into ever higher ISO ranges, since that's a priority to many dSLR users, and an area of active competition between manufacturers. Designers of sensor hardware are therefore going to strive for noise reduction, for example cleaner amplification circuitry and better heat management. Needless to say, such design choices cannot be measured by analyzing images from the sensor after it is produced, and I doubt anyone outside of the R&D shops has access to prototype sensors for comparison. If anyone thinks the only place dSLR makers think about noise reduction is after the RAW image is generated, I hear there's some swampland in Kansas available as a bundle purchase with the aforementioned bridge.

You count cleaner amps as NR? Better heat management, too? How about higher QE - NR, too?

NR is a term typically reserved for some kind of averaging signal from pixels, roughly speaking. Anything done to reduce noise on each pixel channel is not considered NR. Actually, the latter is exactly what we want. When, and how NR is done, does not matter. RAW data is always a processed signal, the physical one being photons.

Deviation from the statistical nature of the noise can be measured but then we do not know whether the alleged NR activates in some situations only, in some parts of the tonal range, etc. My point was that when we do not know, we say - we do not know.

 

[RunAndGun]

"Is there a market for an 8mp crop body with excellent iso performance?"

Yes. That's basically what the C300(and 100 & 500) is. 8)

*I know you're talking about still cams*

 

[Sporgon]

Let me add some emphasis, since it seems I've confused at least one person.

Hmm, just one pierson it would seem.

 

[Pi]

Let me add some emphasis, since it seems I've confused at least one person.

Hmm, just one pierson it would seem.

Feel free to ask for help is that happens again.

 

[ajfotofilmagem]

I saw the pictures that a photographer did with 5D classic, processed with Lightroom, and I was happy with my humble APS-C 18 megapixel. : The noise in the shadows of the 5D classic is horrible and looked worse than any Canon Rebel that I have ever used. Yes, technology has evolved in the last decade, and with current technology, I believe it is possible to make an 8 megapixel APS-C with ISO 6400 that has performance as good as 5D mark ii. If this hypothetical camera was being sold now I would buy three or four of them. However, I understand that I am a minority, and that this is not a strategy of profitable market for Canon. :-\

You are mixing several thing here. Shadow noise is mainly read noise. With older cameras, at very high ISO, read noise is worse than say, ISO 1600, which increases the lead of 5D3 vs. 5D2 there, for example. If that is what is important to you, buy three of the D7100. It has slightly better DR at ISO 6400 than the 5D2 (and much better at base ISO). The shot noise however depends on the QE, and there is no crop camera yet that can beat the 5D2, for example.

Comparing formats is a bit tricky because same ISO with different formats does not mean much. For the same DOF, larger sensors do not have better noise performance but have more resolution which helps offset the noise. On the other, if you are shooting wide open on FF, say at 50/1.4, you cannot shoot with the same DOF on crop because there is not such lens. But if you need the DOF, is different, etc.

If I needed a camera for shooting primarily at ISO100, I'm sure I would meet D7100 very well, but at ISO 6400 and higher, do not see the superiority of D7100 compared to 70D.

 

[ajfotofilmagem]

Speaking of cameras unusual use a monochrome version of the 70D would be great. : Something not really unprecedented, because Leica already has a camera as well, and Canon has the exclusive 60DA.

 

[Pi]

If I needed a camera for shooting primarily at ISO100, I'm sure I would meet D7100 very well, but at ISO 6400 and higher, do not see the superiority of D7100 compared to 70D.

Of course, not. At high ISO Canon is comparable to Nikon, even in the shadows (DR).

 

[ajfotofilmagem]

If I needed a camera for shooting primarily at ISO100, I'm sure I would meet D7100 very well, but at ISO 6400 and higher, do not see the superiority of D7100 compared to 70D.

Of course, not. At high ISO Canon is comparable to Nikon, even in the shadows (DR).

Yes, so no manufacturer wants to make a APS-C camera noticeably superior to current models at ISO 6400. This would bring down the sales of full frame, and Canon wants exactly the opposite, forcing demanding amateurs to migrate to 6D (or more expensive) and L lenses.

 

[Pi]

Yes, so no manufacturer wants to make a APS-C camera noticeably superior to current models at ISO 6400. This would bring down the sales of full frame, and Canon wants exactly the opposite, forcing demanding amateurs to migrate to 6D (or more expensive) and L lenses.

I guess Pentax wants the same? Or Oly?

No manufacturer can make such a camera. If they could, they would, and then they will make better FF cameras as well.

This was said many times. The hard limit at a given ISO (more precisely, at a given exposure) is the area. Same exposure (a.k.a. "same ISO ", which is not quite correct) over a larger area means less noise per unit area. As simple as that. Sensors of the same generation have some differences in how they record that noisy image but the the differences are too small to beat the sensor size factor.