M8 with 2.0 firmware high ISO better noise performance?

[sean_reid]

Advertisement (gone after registration)

Sean,

Of course my tests were quick&dirty and in no way can compare to yours. I also did not resize the files. Unfortunately I had no opporunity to read your tests. Can you point me there?

But I can safely assume, knowing and respecting your writings, that they have the correct results.

So Epson is equal to M8 in noise performance. Given the year of design/manufacture (2004 vs 2006) and the fact that the M8 Kodak sensor is custom built, it's quite an achievement. For Epson. And the price difference.

One other noise that I definitely prefer on Epson is that of shutter and recocking.

The manual recocking is quieter and can be performed any time after the picture.

This Leica toy engine noise is horrible. What a shame recocking was not addressed in the fw 2.0 upgrade.

 

Cheers.

Piotr

 

Even with the resizing, the R-D1 has a slight advantage at ISO 1600 but it is very minor. I think the overall point is that many of us would like to see an M8 successor with a higher S/N ratio in the sensor/processing engine/etc. pipeline. I certainly would like to eventually see files from a digital M that showed low noise through ISO 2500/3200. That's a reasonable goal, I think, for the next model.

 

I still have my R-D1 and I do like the manual shutter cocking lever.

 

Cheers,

 

Sean

[Woody Campbell]

I've found this subject puzzling. I accept Sean's statement that Leica hasn't actually done anything about high ISO performance in the new firmware release, but my own test shots (several pages above) showed ISO 2500 performance that was significantly better than I had remembered. (In my test I shot a Macbeth card in 5500K lighting at short shutter speed.)

 

Then it occurred to me that people (including me) might be seeing the results of different testing setups (and attributing the difference to firmware) unless they did controlled before and after shots. It turns out that high ISO performance is dramatically affected by color temperature, shutter speed and conversion software.

 

With the new firmware I shot a Macbeth target at ISO 2500 in controlled 5500k lighting and in tungsten (which measured 2750k) at shutter speeds ranging from 1/125 to 1/8 and processed in LR and the current version of C1 pro. I also underexposed by one stop in both lighting conditions.

 

There is a significant increase in noise as you go from 5500K to 2750K. In 2750K lighting there is a jump in noise as you go from 1/30 to 1/15, and another jump as you go from 1/15 to 1/8. By 1/15 color noise is aggressive and clumping has a very adverse effect on the lens's ability to record detail. Underexposure makes the issue much worse. So . . . images tend to be noisiest in the most common low light shooting situations: poor tungsten lighting at slow shutter speeds when you are pushing envelope and risking under exposure. This is consistent with my shooting experience.

 

The excellent rendering of low noise detail in the flash shot of the boy above in this thread (shot at 1250) now makes perfect sense to me - I would expect different results in available light.

 

If I get a bit of free time I'll measure standard deviation data and post them.

 

By the way C1 did a significantly better job at color noise reduction that LR - at the cost of some increase in luminance noise.

[Advertisement]

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

[robsteve]

 

The excellent rendering of low noise detail in the flash shot of the boy above in this thread (shot at 1250) now makes perfect sense to me - I would expect different results in available light.

 

If I get a bit of free time I'll measure standard deviation data and post them.

 

By the way C1 did a significantly better job at color noise reduction that LR - at the cost of some increase in luminance noise.

 

Woody:

 

I didn't use a flash in that shot of my son. It was window with overcast light coming in.

 

I think the real difference is the color temperature that you shoot at. In my case, between the tungsten and the light coming in the widows ,I was at about 4600K for that shot of my son.

 

When you do get into tungsten, trying to remove all the yellow cast to make the image neutral will make the noise worse. In this type of lighting there is very little blue and when you try to balance it to neutral, you bring up the very weak blue channel. It really wasn't any different with film. In tungsten you could only correct so far before the prints looked awful and grainy.

 

Robert