M9 Colors at Night — Best Way to Shoot High ISO?

[Gerard]

Advertisement (gone after registration)

I have shot loads in low light at base iso. Well, it's what flash was invented for, right? :-)

[tomasis7]

Flash sucks

 

 

 

 

 

 

 

 

 

 

 

 

 

 

For some

[tomasis7]

tomasis7, as this thread makes clear from the outset the technique discussed here is with LR4/5, since other software may have different results. While I have looked at M240 colors, I have not done so in the context of the high-ISOtechnique discussed here.

 

—Mitch/Paris

Surabaya-Johnny

 

re pushed 4 stops photos, did you find banding or anything strange from m9 sensor?

 

do you have time to post 4 stops pushed b&w shots? from iso 640, maybe 1250 in order to find limits.

[Guido]

Yes, you'll likely see banding, as well as purple/magenta color blotches after noise correction. Here is how I deal with them in most cases. The example below is a "worst case": an extreme 5 stop push (the maximum possible in Lightroom). It doesn't get much uglier than that

 

I've also tried noise reduction in Nik Dfine, but found that LR gives better results.

 

All in all, I find the results achievable quite amazing... for a 5 stop underexposure and an "obsolete" camera

 

(Note that these are 100% crops.)

 

Original image out of camera... ISO 640, 1/60 sec handheld

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

 

Exposure slider +5 stops in LR

Note the banding (horizontal stripes) and noise

 

 

Noise reduction 50/50/0 (Luminance/Detail/Contrast) and 25/50 (Color/Detail)

Banding and noise reduced, but note magenta/purple color blotches in tree areas

 

 

Hue shift (in HSL tab): Purple -50, Magenta -50

Magenta/purple color blotches reduced somewhat

[Guest malland]

re pushed 4 stops photos, did you find banding or anything strange from m9 sensor?

 

do you have time to post 4 stops pushed b&w shots? from iso 640, maybe 1250 in order to find limits.

If you go through this thread, you'll find some pictures pushed 3-4 stops. The results were quite good. I don't post 100% crops because I'm interested in how pcitures print, not how they look at 100% on the monitor; indeed, a 50% view gives a reasonably good indication of how a print will look. I don't have any B&W conversion using this technique, as I don't use the M9 for B&W.

 

—Mitch/Paris

Surabaya-Johnny

[SoarFM]

What a wonderfully educational topic here. I recently applied some of the suggestions here and have been very pleased with the results. I too have struggled with the M9 and high iso's. What works for me as i apply the lessons of this thread, at least on the monitor, is to click "auto" in the basic exposure to get me in the ball park of the proper push, then I lighten the mid tones with the curve and go back to the basic sliders and bring down the blacks to a realistic level which is essentially increasing contrast.

 

Next up are some prints, that's where the rubber meets the road for me. So, we'll see.

 

Elsewhere, on GetDPI forums I recall reading one participants experience with testing a Phase One back where he quite convincingly concluded that increasing the iso in the camera back was essentially just a metadata notation telling the raw processor how much to push from the base iso. Could it be that is what we are seeing here with the M9? Could it be that allowing a powerful computer make the "push" in a raw processor is better than letting your less powerful camera processor chip do the work in a less flexible manner?

 

Fred

[JimKasson]

Advertisement (gone after registration)

Elsewhere, on GetDPI forums I recall reading one participants experience with testing a Phase One back where he quite convincingly concluded that increasing the iso in the camera back was essentially just a metadata notation telling the raw processor how much to push from the base iso. Could it be that is what we are seeing here with the M9?

Fred

 

I've looked at the M9 raw files a lot (mostly using RawDigger), and changing the ISO dial on the camera actually changes the rgb raw values. It's not a "metadata push". I believe changing the ISO dial changes analog gain prior to digitization. If there were digital gain, it would, in the absence of added digital noise, show up as histogram depopulation at the higher ISO settings.

 

Jim

[jcm0]

First of all, a big thank you to Mitch (Malland) for having initiated this thread. I got plenty of interesting infos. Looking at other forum talks, I get a list of factors for improving the M9 image quality in low light:

 

- Malland's great push in postprocessing technic (see first post)

- Do not use auto-WB. Setting color temp lead to less noise. Yes it works.

- Use Uncompressed DNG (useful up to 1280 iso included), which gives better dynamic range (dynamic range>8bits when iso<=1280), and will help noise suppression softwares.

- Use Auto Lens detection : this settings adds some internal processing (applied directly in the raw DNG file with no way to get rid of it) which improves the noise (and correct color cast) at picture corners. With uncoded lens, Cornerfix will help to an extent.

- Use the lightroom detail panel in the development tab to reduce color noise (astonishing efficience !) and luminance noise (not to much).

- Using Clarity in lightroom may help also to avoid pushing too much the other exposure sliders.

- Dfine 2 Nik software may help in additional noise suppression, but lightroom 4 does a great job.

 

Regards,

 

JC

[JimKasson]

Do not use auto-WB. Setting color temp lead to less noise.

 

The in-camera white balance settings simply change the metadata, not the rgb raw values. Setting color temp in post will yield the same results as setting the same temp in-camera if the raw processor uses rgb scaling for white balance.

 

Jim

[JimKasson]

The in-camera white balance settings simply change the metadata, not the rgb raw values.

Jim

 

Here's an example snip from M9 EXIF data with autoWB enabled:

 

http://www.kasson.com/ll/m9%20wb.PNG

 

And here's what the EXIF data looks like with the camera set for (nearly) UniWB:

 

http://www.kasson.com/ll/m9%20near%20uni%20wb.PNG

 

If you're new to UniWB, see this:

 

Using in-camera histograms for ETTR | The Last Word

 

Jim

[ddp]

Mitch - thank you for this thread. The shots you're posting look like what night shots should look like (to me). They have that certain ambience and what appears to me as being "natural". As someone who shoots with both an M9 and a Nikon D3s - I am aware of what each machine can do, and what they can't. The D3s has almost unreal ability in low lighting, but my experiences have been that it tends to make the scene look far different than what my eyes relay. Anyhow, thanks again!

[jcm0]

Thanks a lot Jim for your remarks .

 

Here is the updated list of factors for better image quality:

 

- Malland's great "push in post-processing" technic (see first post)

- Provide the highest - possible when shooting - exposure level without clipping (called ETTR) (see 02. Why ETTR? | The Last Word)

- Use Uncompressed DNG (useful up to 1280 iso included), which gives better dynamic range (dynamic range>8bits when iso<=1280), and will help noise suppression softwares.

- Use Auto Lens detection : this settings adds some internal processing (applied directly in the raw DNG file with no way to get rid of it) which improves the noise (and correct color cast) at picture corners. With uncoded lens, Cornerfix will help to an extent.

- Use the lightroom detail panel in the development tab to reduce color noise (astonishing efficience !) and luminance noise (not to much).

- Using Clarity in lightroom may help also to avoid pushing too much the other exposure sliders.

- Dfine 2 Nik software may help in additional noise suppression, but lightroom 4 does a great job.

[JimKasson]

Here is the updated list of factors for better image quality:

 

- Malland's great "push in post-processing" technic (see first post)

- Provide the highest - possible when shooting - exposure level without clipping (called ETTR) (see 02. Why ETTR? | The Last Word)...

 

All good, but note that the two approaches above can't be used simultaneously. There are times when each is the best plan.

 

Jim

[jcm0]

You're right Jim Thanks.

[mark_j]

Here are a couple of pictures taken with the Monochrom in Soho around dusk last week. Both had the same exposure (f8, 1/250) whilst the first was shot at ISO 1600 and pushed 2.5 stops in LR5 and the second at 3200 and pushed 1.5 stops. I didn't apply any NR but both pictures look a lot better for a +20 adjustment.

 

I realise this is not perfect but I am not one to do tests just for the sake of it. I hope it's better than nothing.

 

According to the theory the 1600 pushed shot should look better than the 3200 pushed shot. Having spent some time looking at these on a 27inch monitor I can honestly say I see no difference. I do think this is more to do with the structure and the look of noise in a true monochrome image rather than any problem with the underlying theory. I am sure if the images were forensically analysed then the difference may be noticeable.

 

I do think though that it does show how good the Monochrom is at higher ISO and for me that has been one of the big revelations of using this camera. I can now go out at my preferred settings of f8 and 1/250 and get good pictures in low light situations.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Guest malland]

Mark, thanks. Yes, M-Monochrom is excellent at ISO 1600 and ISO 3200. The issue here is that, if you shoot it at really high ISOs, such as 5000, 6400, 8000 or 10000, the dynamic range will be quite low. What we've been discussing is that most likely — I haven't tried it yet — the results in terms of dynamic range will be optimized by shooting at ISO 1250 and pushing Exposure in LR4/5 (as suggested by Jim Kasson earlier in the thread). I'll try this when I get a chance.

 

—Mitch/Paris

Tristes Tropiques [WIP]

[tomasis7]

malland ok thanks!

 

 

nightshift, thats very useful! sorry that I missed your post. Thanks a lot for very nice photos!! I mean those photos that I could analyse at closer detail.

 

I couldnt find any banding in your image4 . For example below in my example (rd1) at iso 400 pushed, there you see quite pronounced horizontal banding. Im really impressed by leica m9 sensor. it looks organic everywhere albeit controlled, very like Agx. Though I could sense very slight horizontal banding there but it is not much, like 98% less than Rd1. I could understand what you mean if we talk about colors.

 

the images are exposed in order to find banding. The other is rotated so I could check out horizontal eventual banding.

 

I will put hands on the camera

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[tomasis7]

played with the same img4 as it could be in real world. normal photo. I dont find nothing disconcerting. It is pleasing in fact. Im fan of Trix films, sometimes pushed to iso 1600-3200

 

Why B&W, because it is easier to analyse for eyes to find faults. Sorry Maland. I will leave the thread for colourists

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Guido]

A short update to my previous posting #184:

 

I recently played around with the current version of Photo Ninja, a raw converter for Windows and Mac. In first tests with some of my ISO 640 images, the results looked very promising. Compare the noise structure and highlight rendering in the Photo Ninja image below to the final Lightroom adjustments in my posting 184. Exposure is tuned with several sliders and switches in Photo Ninja ("Illumination", "Offset", etc.) as opposed to the single "Exposure" slider in Lightroom, and I haven't figured out all the details yet.

 

While the lack of asset management features makes Photo Ninja too cumbersome for me to use as a regular tool, I'm seriously considering making it the tool of choice for processing low-light images instead of Lightroom from now on. I'm unsure about the learning curve though, and I need to explore it a bit more before making a purchase decision. Any opinions?

 

PhotoNinja

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

Lightroom

[Advertisement]

Hallo,

Look here: Leica M

[Guest malland]

nightfire, sorry, but I don't really follow either your posting above nor #184. I presume that you're showing sections of your file at 100%. Using Lightroom 4 or 5, I have never seen anything that looks as blotchy as that. What version of Lightroom have you been using?

 

—Mitch/Paris

Tristes Tropiques [WIP]