Lower ISO

[glacierparkmagazine]

Advertisement (gone after registration)

I know there's a lot of flack on the M9 and it high ISO performance (or lack thereof), but what I really want is a true lower ISO, like a 50 or 25, so I can shoot those wide open shots in broad daylight without ND filters.

 

All shot in the M6 was Velvia anyway.

 

Anybody else with me?

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[ho_co]

I can't argue with the request, but I don't think it's going to happen.

 

Check out Michael Hußmann's article (part 2 of 3) in the current LFI (3/2010).

[Shade]

Don't think it will happen too, at least not for the m9. Maybe an m10 later on?

[lars_bergquist]

What's wrong with ND filters?

 

The old man from the Filter Age

[adan]

It would require a complete redesign (in a retrograde manner - going against the tide towards higher ISOs) of Kodak's pixels. Base (lowest) ISO is hardwired into how the silicon is doped and arranged and all that - how many photons it can accept before the electron wells saturate and the highlights blow out.

 

And unlike film, where lower ISO meant smaller grain - the smaller the pixel, the FEWER the electrons it can hold - so it's the dynamic range that shrinks, not the ISO.

 

If Leica's choice is a) invest another €5,000,000 with Kodak on a major sensor redesign program for (very) limited use - or leave it to you buy ND filters.....well, whaddaya think?

 

OTOH if you really want it bad - pony up the 5 mill yourself and I'm sure they'll consider making you one.

 

I guess Leica could put a 4x ND filter right on the sensor - but as we know from the M8/IR experience, Leica thinks filters on the sensor degrade image quality more than filters on the lens, so you're not likely to get a sympathetic ear.

[250swb]

How do Canon fake an ND filter on a P&S like the G10? I've never used the feature on mine properly, but a quick test seems to suggest the IQ isn't degraded by the software option. Just maybe it wouldn't be so radical as to need a full re-design of the camera, just some processing code for JPEG users?

 

Steve

[jaapv]

Advertisement (gone after registration)

If it is pulling down the amplification it will not be obvious on low-contrast shots, but it will take down the dynamic range. Iirc the base ISO for those sensors is 100, so you will lose one stop of DR @ ISO 50 - nothing dramatic. In fact, exactly the same as ISO 80 pull on the M9.

[mjh]

How do Canon fake an ND filter on a P&S like the G10?

As far as I know it isn’t faked; these are real ND filters. Canon has a long tradition of compact cameras where the fastest shutter speeds can only be used when stopping down. If you are after a shallow DOF in a brightly lit scene, you need to use the ND filters as the shutter speed is limited.

[cbretteville]

All shot in the M6 was Velvia anyway.

 

Shortest shutter speed on an M6 1/1000, on an M9 its 1/4000. So in that sence you're better off with the ISO80 Pull on the M9. As the base ISO160 is more like ISO200 in real life that should give you about the same shooting conditions as your M6 with Velvia 50 given the same lens.

Carl

[mjh]

what I really want is a true lower ISO, like a 50 or 25, so I can shoot those wide open shots in broad daylight without ND filters.

It isn’t that easy to build a low ISO sensor, at least not in a way that offered real benefits. One could fake it, like many DSLRs and the M9 do, with an ISO pull setting that’s one f-stop below the sensor’s native sensitivity. But you lose one f-stop’s worth of dynamic range that way and you don’t get anything in return. It’s just that you can do without an ND filter. More on that in my LFI article (current issue).

 

There are basically two ways to reduce a sensor’s native sensitivity. One way would be to reduce its quantum efficiency. An integrated ND filter would do the trick, but it would be simpler to just leave out the microlenses. Without microlenses, the sensors in the M8 and M9 would have a native ISO of about 80. Microlenses create a lot of issues of their own (especially vignetting), so doing without has its attractions, but apart from this, the lower ISO sensitivity wouldn’t confer any advantages. Noise and dynamic range at ISO 80 would be just the same as they are at ISO 160 now. More importantly, noise and dynamic range at ISO 1250 would be equivalent to the performance at ISO 2500 now. Surely not an attractive proposition.

 

The second method is to increase each sensor pixel’s storage capacity for electric charges. This would not just lower the native ISO value, but also reduce noise and increase dynamic range. Moreover, the sensor’s performance at higher ISO settings wouldn’t be compromised – while it wouldn’t improve, at least it wouldn’t get any worse either. Quite obviously this were the way to go – if we only knew how.

[jaapv]

I wonder how Kodak did that on the DSC 14. Iirc one could dial it down to ISO 20 with the quality increasing all the time

[menos I M6]

I wonder how Kodak did that on the DSC 14. Iirc one could dial it down to ISO 20 with the quality increasing all the time

 

… which was capped at ISO 800 unlike todays digital sensors.

[mjh]

When I was reviewing the DCS Pro 14n in the spring or summer of 2003 it was still very much a work in progress, but at that time its ISO range went from ISO 80 to 400. The manual mentioned ISO 640 that wasn’t actually available (except at a reduced resolution), also by fiddling with the dials I sometimes managed to select both values below ISO 80 and up to ISO 6400 that weren’t real either. Admittedly it was hard to tell exactly as the ISO value wasn’t stored in the EXIF data. At the time, Kodak didn’t recommend ISO settings above ISO 200, promising these would become usable with later versions of the firmware (4.2.2 was the version I had). Exposure times were limited to 2 seconds max; longer times could be selected but prompted a warning that this wasn’t recommended at the current ISO setting (regardless of the ISO setting). Indeed the resulting images did look strange.

 

The DCS Pro SLR/n allowed for longer exposures but actually combined the image data of a series of shorter exposures. For some reason this also resulted in a reduced ISO value, from ISO 50 for a 2 second exposure down to to ISO 6 for exposures between 15 and 60 seconds. I have no idea what exactly was happening there; maybe the camera was doing some kind of HDR to make up for the reduced dynamic range of the ISO pull shots. The 14n could be upgraded to the sensor of the SLR/n so I suppose similar options were available with an upgraded 14n.

 

All in all the FillFactory sensors performed strangely; the results at lower ISO settings were excellent (if you could get rid of the moiré that is), but there’s a reason why we haven’t seen those sensors ever again.

[adan]

"One could fake it... with an ISO pull setting ..... But you lose one f-stop’s worth of dynamic range that way and you don’t get anything in return."

 

Actually, Michael, my experience is that there is a slight reduction in noise using the "Pull 80" ISO over ISO 160. As with Dolby noise reduction, if you boost the signal (2x the light) and then de-amplify both the signal and noise, the noise goes down. But it does put you (as your articles have noted) one stop closer to burning out the highlights.

[mjh]

Actually, Michael, my experience is that there is a slight reduction in noise using the "Pull 80" ISO over ISO 160. As with Dolby noise reduction, if you boost the signal (2x the light) and then de-amplify both the signal and noise, the noise goes down. But it does put you (as your articles have noted) one stop closer to burning out the highlights.

In my measurements I have found a little less chroma noise at ISO 80 vs. ISO 160 (a reduction by 11 percent) whereas luminance noise was the same – actually a tiny bit higher at ISO 80 but well within the error margin of these measurements. So yes, the compression inherent in digital pull processing does reduce some kinds of noise.