Does the M9 have good and bad ISOs?

[pop]

Advertisement (gone after registration)

I believe that the main difference lies in the absence of the Bayer filters, which yields a difference in sensitivity of about one stop, i.e. a factor of two. I think the curves are really close together once they are normalised with respect to the base sensitivity.

[IWC Doppel]

One other thought is how and why the base ISO of the M9 is slightly lower noise than the MM, which doesnt make sense to be honest, perhaps the noise calculation has some subtleties to it that arent apparent from these graphs.

 

From what I have seen the real life gap appears bigger

[pop]

Given that the MM and the M9 use the same sensor, there's still one difference which - after all - determines the MM's strength. While each and every "site" (is that what it's called) produces one monochrome pixel in the MM, a weighted average of four "sites" are used to calculate one colored pixel in the M9. I think that this averaging process just conceivably could result in a very slightly reduced noise level.

 

Afterthought: If this is true, just comparing the noise of the sites of one color - say, green - should give a still smaller difference.

[flyalf]

This might help, posted in a LFI magazine. Green MM, blue M9

 

Thanks for sharing!

 

There is also some indications that one could (should) treat M9 as ISO-less from 640 (or better 500) and do post-prosessing push in LR instead of increasing ISO in M9. Please see: ETTR — Just crank up the ISO? Part 13 | The Last Word

[IWC Doppel]

Thanks for sharing!

 

There is also some indications that one could (should) treat M9 as ISO-less from 640 (or better 500) and do post-prosessing push in LR instead of increasing ISO in M9. Please see: ETTR — Just crank up the ISO? Part 13 | The Last Word

 

Yes an interesting thread, I have been involved in this thread on LuLa

 

Leica M9 ISO 1,000 and above NR settings

[douglasf13]

Interesting, thanks guys. As I mentioned on the first page, I've just been shooting my M9 at ISO 160 for everything, and then boosting in LR, but it looks like ISO 640 may be the sweet spot for the cutoff. Maybe I'll just use ISO 160 most of the time in good light, and then switch to ISO 640 in lowlight.

[IWC Doppel]

Advertisement (gone after registration)

How do you choose shutter speed ? I set manually with the dot/arrows so match for the 'correct' speed at a chosen ISO

[algrove]

Maybe I'll just use ISO 640 most of the time in good light, and then switch to ISO 640 in lowlight.

 

Did you mean this?

[douglasf13]

Did you mean this?

 

Whoops, sure didn't. Fixed. Thanks!

[douglasf13]

How do you choose shutter speed ? I set manually with the dot/arrows so match for the 'correct' speed at a chosen ISO

 

Once you go "ISO-less" (a term that has caught on, but isn't really accurate,) you almost think of it like using auto-ISO in M mode. Just set whichever aperture and shutter speed you like, of course knowing that, the more light that hits the sensor, the better (unless you're clipping in bright light at ISO 160.) Then you just push to taste in LR. Of course, you don't want to really push more than 3-4 stops in LR4, if you start with ISO 640, because things will get really noisy. After a bit, you'll start to get the feel for it.

 

The great thing is that you'll rarely blow highlights with this method, accept maybe when you're in bright light at ISO 160. The big negative is that your review images on your camera's LCD (and when importing into your raw converter) will be dark.

[Lindolfi]

How do you choose shutter speed ? I set manually with the dot/arrows so match for the 'correct' speed at a chosen ISO

 

One way of working with this is

 

1] Set aperture

 

2] Set camera to base ISO = 160

 

3] Adjust exposure time for correct exposure (using dot and arrows)

 

4] Now if this exposure time is too long (for subject or camera motion), turn it with a maximum of 8 clicks (4 stops) of the exposure dial shorter

 

5] If you need more than 4 stops, go to step 2] increase ISO and repeat 3] and 4]

 

6] After loading the RAW correct the image with the stops underexposure used

 

(I've checked it with the M9 and you get a visibly lower noise level)

[algrove]

This is close to what Overgaard does as he uses ISO 800 indoors, but does not do LR pushing from what I can tell.

[douglasf13]

One way of working with this is

 

1] Set aperture

 

2] Set camera to base ISO = 160

 

3] Adjust exposure time for correct exposure (using dot and arrows)

 

4] Now if this exposure time is too long (for subject or camera motion), turn it with a maximum of 8 clicks (4 stops) of the exposure dial shorter

 

5] If you need more than 4 stops, go to step 2] increase ISO and repeat 3] and 4]

 

6] After loading the RAW correct the image with the stops underexposure used

 

(I've checked it with the M9 and you get a visibly lower noise level)

 

That's basically what I do too, Burt, although I go to 5 stops, since LR4 goes to +5EV. Steps 3 and 4 start becoming second nature, so I rarely bother to do them.

 

Over on Jim's blog that was linked above, he mentions that, in the shadows, there's really no advantage to raise ISO at all in-camera. In the mid tones/highlights, you gain less than a half stop improvement by raising ISO up to 640, after which everything starts going downhill.

[IWC Doppel]

This is close to what Overgaard does as he uses ISO 800 indoors, but does not do LR pushing from what I can tell.

 

Yes, I think this was an ease of managing ISO settings (200 outdoors and 800 indoors) I have read somewhere about fixing WB to improve noise performance over AWB, which sounds interesting as I am 99% B&W any way. I might try fixing WB in the evening and experiment

[douglasf13]

Yes, I think this was an ease of managing ISO settings (200 outdoors and 800 indoors) I have read somewhere about fixing WB to improve noise performance over AWB, which sounds interesting as I am 99% B&W any way. I might try fixing WB in the evening and experiment

 

I believe that the fixed WB thing was a misconception when the M9 was newer, but has since been disproven. If you shoot raw, it shouldn't matter what your WB is at capture, but certainly try it out if you're curious.

[rscheffler]

The question about WB really should be more about whether to fix it in post via the RAW file, or to filter it 'correctly' with an actual filter on the lens when doing the exposure. For example, in tungsten lighting conditions, the problem is the lack of blue light and the corresponding underexposure of the blue channel. Using an 80A/B would somewhat equalize exposure to the RGB pixels/channels, but of course with the cost of 1-2 stops light loss... My feeling is this is the primary reason for the M9's blotchy low-light colour quality, because often we're doing strong colour balance shifts in post.

[douglasf13]

The question about WB really should be more about whether to fix it in post via the RAW file, or to filter it 'correctly' with an actual filter on the lens when doing the exposure. For example, in tungsten lighting conditions, the problem is the lack of blue light and the corresponding underexposure of the blue channel. Using an 80A/B would somewhat equalize exposure to the RGB pixels/channels, but of course with the cost of 1-2 stops light loss... My feeling is this is the primary reason for the M9's blotchy low-light colour quality, because often we're doing strong colour balance shifts in post.

 

Agreed, although most camera's I've used haven't had near the issues with this blotchiness. That being said, I can usually get rid of most of it with subtle WB shifts, in terms of both temp and tint.

[dwbell]

The question about WB really should be more about whether to fix it in post via the RAW file, or to filter it 'correctly' with an actual filter on the lens when doing the exposure. For example, in tungsten lighting conditions, the problem is the lack of blue light and the corresponding underexposure of the blue channel. Using an 80A/B would somewhat equalize exposure to the RGB pixels/channels, but of course with the cost of 1-2 stops light loss... My feeling is this is the primary reason for the M9's blotchy low-light colour quality, because often we're doing strong colour balance shifts in post.

 

This is very interesting. Do you happen to have a link, or book title, to "old school" filtering for light type etc? Care to elaborate more? Would welcome it.

[rscheffler]

Sorry, no links to books or other articles at the moment. It's one of those things I read somewhere, at some time that kind of stuck because it made sense.

 

I believe the theory is that digital sensors are in some respects like colour film. They have one best, most optimized zone for white balance. And the idea is you use filtration in non-ideal lighting situations to correct for the ambient light colour imbalance. Most films are/were optimized for 'daylight' which would be in the 5200-5600 Kelvin range. Not sure about digital sensors, and it might be sensor dependent. Assuming roughly daylight optimization, then one would use an 80 series blue filter, either 80A or 80B in tungsten lighting conditions to bring the roughly 3200 Kelvin up to around daylight equivalent. The blue of the filter absorbs a fair amount of red, reducing its transmission to the sensor, while transmitting green and a lot more blue.

 

The challenge when shooting under tungsten lighting, and one of Jono's M240 images illustrated this somewhat (the one of the girl hugging the dog inside), is that the strong red shift of the light will cause red channel clipping/over exposure quite quickly while on the other end, the low amount of blue wavelength light will result in blue channel underexposure. If one exposes for the red light to prevent clipping, then the blue channel receives even less light. When one goes to adjust the WB in the RAW conversion, the information from the blue channel will be quite weak and my guess is more interpolation of the intended colour quality is required by the software, resulting in some blotchiness/colour noise.

 

In Jono's image, the red channel in the skin tone areas was already clipped. Using an 80A or 80B filter would have absorbed the excess red light and would have prevented the red channel clipping while allowing more green and blue light to those pixels. Assuming white subject matter, If the red, green and blue pixels get roughly equal light, then the theory is none will be dramatically underexposed. When it comes time to process the RAW file, the necessary WB adjustment will be minor and won't incur undue colour interpolation. Whether this actually works perfectly in practice, I'm not 100% sure. I also recall reading something about how the red and blue pixels/channels have greater gain applied to their values, I believe to offset the lower pixel count in a Bayer array vs. the green channel.

 

It would be interesting to test this theory... If I can find my old Wratten filter sets, I might be able to set up a test. Of course the drawback to this method is the light loss from the filter. Usually when one is working with low level tungsten lighting, there isn't enough and one's at a high ISO already trying to squeeze out as much possible performance from the lens and camera.

 

Maybe someone with some actual facts can jump in and offer a more definitive answer.

[Advertisement]

Hallo,

Look here: Leica M

[nggalai]

This is very interesting. Do you happen to have a link, or book title, to "old school" filtering for light type etc? Care to elaborate more? Would welcome it.

 

The term you’re looking for is “color balancing filters” or “color correction filters (CC)”. Usage has somewhat dwindled, but it was necessary when primarly shooting film.

 

For example, see section 4 in the data sheet for Fujichrome Velvia:

 

http://www.fujifilm.com/products/professional_films/pdf/velvia_50_datasheet.pdf

 

 

I’m unconvinced color balancing filters will produce better results than correcting color/white balance in post, but Ron’s posting certainly made me reconsider a couple of experiments. You may need to boost ISO to 2500 rather than sticking with 640, but noise should be more uniform.