50,000 ISO Maistro

[dkmoore]

Advertisement (gone after registration)

Hello all,

 

I have been reading about the new Q and it having up to 50K ISO and wanted to see if the consensus is that we may see a firmware/software update for the M/M-P 240 cameras?

 

I am not sure if you can update software via firmware or not but I would love some education on the topic. This would be a huge upgrade and advantage to all users if possible.

 

Hopefully this topic hasn't already been started. If so, my apologies!

 

Keith

[Advertisement]

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

[jaapv]

Not a chance. The last time this happened was on the DMR, from 400  to 800...

[dkmoore]

Bummer! Thanks for the reply. I guess I was a little too optimistic. 

 

In general, would it be difficult to do?

[JonPB]

In general, would it be difficult to do?

 

Keith,

 

Yes and no. (You mentioned that you'd like to learn more on the subject. I'm no expert, but here's how I understand the issue.) Yes, it is difficult to create a good image with less light, which is what higher sensitivity implies. If it wasn't hard, then every camera manufacturer would be offering cameras that could produce high quality images with ISO numbers in the millions. No, it isn't hard to make it an option on a camera, but the problem with that approach is that you're throwing away four stops of information from an already sketchy ISO 3200.

 

With a firmware upgrade, you can't change the underlying hardware and so you can only optimize noise reduction algorithms. That takes quite a bit of processing power and produces results that aren't sufficiently better to make up for the lack of light. Try underexposing an ISO 3200 image by four stops, increasing the exposure in post by four stops, and applying all the tricks and techniques you can to get an image with high technical quality. That's pretty much what doing it in firmware would look like, best case scenario without any processing or time limitations.

 

With a new camera, there is the possibility of modifying the analog-to-digital electronics so that they can produce better images with less light. The easiest way to do this is by reducing the strength of the color filter array, which increases luminance sensitivity at the expense of color sensitivity. (In other words, make a sensor somewhere between the M 240 and M 246.) Another way to do it is to change the way the sensor element itself is made; perhaps the clearest illustration of this is with backside illumination. The technology is there to do it--making the wells shallower, enlarging the sensor within the well, shortening the path between sensel and ADC, etc--but that takes increasingly high resolution lithographic machines and/or jumping to a different architectural design, which means much higher prices.

 

As an aside, one of my "conspiracy theory"-level hypotheses is that the Retina display is what has caused the stagnation of sensor technology since Sony released their 16 megapixel APS-C sensor tech about five years ago. I'm under the impression that most sensor fabrication plants use microprocessor fabrication systems that are a few generations old. E.g., Intel develops a 14 nanometer process, makes premium processors for our servers, desktops, and laptops with it for two years, then uses it to make flash memory for a few years, then sells the devices to another fabricator; that company then uses the same equipment to make second-tier integrated circuits, like video cards; after a while, it gets too old for them and they sell it to another company, which uses it to make imaging sensors. For the last four years, companies have been demanding high resolution displays for their mobile devices, and I think they've been buying the lithographic machines to make phone and laptop displays--which are more profitable than cameras--leaving sensor tech to stagnate at somewhere between 60 and 90 nm. In about the same timeframe, SSD has become the default storage technology for most computers, creating another source of demand that prevents sensor manufacturers from getting their hands on improved lithographic machines. That said, I suspect the demand for high resolution displays and storage has largely been addressed, given that ever-smaller processors are still tricking out of Intel and AMD, allowing more advanced process lithographic devices to continue down the river to sensor manufacturers, so I expect substantial improvements in sensor tech over the next year or three.

 

So. ISO 50,000. From other manufacturers, I'd be skeptical of getting quality results at that sensitivity, but Leica has consistently been conservative about the ISO ranges it offers in order to ensure respectable quality at all factory options. But, if Leica does release the Q with an ISO setting of 50,000, I'd expect it to be fairly good. At that sensitivity, you're also encountering shot noise, but that's another issue entirely. This jump in quality would have to be due to different hardware, which may or may not be superior at base ISO, though again I doubt Leica would sacrifice much on that count. It is also possible, as you seem to suggest, that Leica employs aggressive and time-consuming noise reduction algorithms with a new microprocessor, which would fit with their dark-image noise reduction approach to long exposures. At any rate, it is something that requires new hardware to make practical, if not literally possible.

 

All that said, if ISO 50,000 would make a huge difference for the way you shoot, do experiment with underexposing and pushing in post. It may only be able to produce high contrast, noisy B&W images, but that style has a following of its own from the days of ISO 1,600 B&W film and it may be suitable to your subject matter and you may be able to produce such images today with the camera you already have.

 

Hope this helps. Also, I hope that I'm not propagating any horribly incorrect information. :-)

 

Cheers,

Jon

[CheshireCat]

Beware of marketing gimmicks.

 

By the info we have now, this claim seems to be based on software denoise. So you could just apply the denoise algorithm later to your M240 raws, no matter how slow the CPU on your camera.

 

You may also want to read this:

 

http://blog.kasson.com/?p=3138

[Exodies]

Beware of marketing gimmicks.

 

By the info we have now, this claim seems to be based on software denoise. So you could just apply the denoise algorithm later to your M240 raws, no matter how slow the CPU on your camera.

 

You may also want to read this:

 

http://blog.kasson.com/?p=3138

 

Does the graph for a camera which isn't ISO-less slope upwards from left to right? Do you have real data which shows this?

[250swb]

Advertisement (gone after registration)

This would be a huge upgrade and advantage to all users if possible.

 

 

 

Huge? The number sounds big, but only two or three stops is all it amounts to, about the equivalent of switching the lights on or getting a faster lens, each with fewer downsides under current technology. And you know what, even when big numbers like that are offered photographers still keep their camera at base ISO because they don't want to be seen making a noisy grainy looking images that their peers can scoff at.

 

Steve

[dkmoore]

Thank you Jon for the very detailed response and interesting read. Much of what you wrote makes a lot of sense in general with technology, not just camera technology. I figured it would be more than just a simple firmware/software update but I am uneducated at that level of science/tech.

 

Steve, I am an exaggerator at heart, which you wouldn't know without meeting me! But, I get your point and agree. I prefer to keep my camera at base ISO as well but also like the idea of having "usable" higher ISO. Don't many of us always want more (even with little reason to)? In reality, there have only been a few circumstances that I would have liked more than ISO 3200 (maybe a handful). Then again, I have also spent a million pence on SX lenses to assist :  )

 

Like Jon mentioned, Leica is generally conservative with what the ceiling is for settings so my question was more asking from a technology standpoint would it be possible to update and retain similar performance of 3200. As I read the responses it sounds clear that this most likely would not be possible without hardware changes and huge losses in performance without those hardware changes.

[mjh]

I have been reading about the new Q and it having up to 50K ISO and wanted to see if the consensus is that we may see a firmware/software update for the M/M-P 240 cameras?

The Q has a different sensor than the M and M-P (Typ 240) and no firmware update could change that. The Q also features a different CPU.

[CheshireCat]

The Q has a different sensor than the M and M-P (Typ 240) and no firmware update could change that. The Q also features a different CPU.

 

Which sensor is in the Q ?

[jdlaing]

Which sensor is in the Q ?

None of your business.