A Question for the Experts...

[barjohn]

Last night I was experimenting with low light shooting to better understand how the camera reacts and the amount of noise in the image at high ISOs. I had read about the trick of setting the shutter speed to +1 and was trying to evaluate different settings.

My findings (still a little preliminary) is that in low light at high ISO the camera needs longer exposures than the meter would indicate. In fact as I switched to manual mode and went for longer exposures (say the metering system suggested 1/125 at f1.4, and I set the shutter speed to 1/15 I would get a cleaner image than at 1/60 or 1/30th. Eventually I would reach over exposure. If shot at the suggested exposure the image would be dark and noisy.

My question is this: Is this due to a light meter that does not compensate for the characteristics of its sensor and ISO? In other words, should the camera software recognize that the ISO setting is high, and change its suggested exposure value based on the high ISO and low light level to give a proper reading to produce high quality images? It would seem that Epson's software engineers could have programmed in a compensation curve that better matched the sensor's need for more light. Thanks for your comments.

John

 

[wintoid]

I'm not putting myself forward as an expert, but I'll offer my thoughts...

I've heard it said that digital cameras cannot retain detail in the highlights if the image is overexposed, whereas if the image is underexposed you have a chance of pulling the detail out of the shadows. As a result of this, some cameras (my Canon 20D for example) seem to deliberately underexpose.

 

[ampguy]

I've also noticed that in low light, the meter sometimes reads a stop or two lower than other cameras, but still usable settings (and in some cases preferred depending on workflow -- no hot spots), and reflects so in the image and histogram and lack of blown highlights anywhere.

I previously thought it was the funky bessa metering pattern, but would be interested to read others experience here. I'm using v1 firmware, but likely to move to v2 soon. I realize I get the extra segment in the histogram with v2 and that its better somehow.

 

[barjohn]

I was particularly interested in noise reduction. Given the low light levels I am talking about, it would seem like blown highlights is less of an issue than noise in the image. It seems to me that rather than +1, it really needs +2 to get relatively clean images at ISO 800. I haven't found a setting that works at ISO 1600 but that may be due to my lack of experience with this camera.

 

[jim_buchanan]

Use an incident light meter to compare with the R-D1 metering

Use an incident light meter to compare with the R-D1 metering

I am always glad I have kept my incident light meter from my M3 days. Except for more complicated methods of calibrating a metering system, an incident meter can help "cut to the chase".

I used it last night, and the scene needed more exposure than the R-D1 indicated.

 

[Ben Z]

it really needs +2 to get relatively clean images at ISO 800.

Have you compared the in-camera metering to a handheld meter or another camera? If they agree, then what you're implying is that ISO 800 in the RD1 is really 200. If the meters don't agree then you're implying the RD1's meter fails in low light. So far I haven't found either to be the case, so perhaps it's your specific camera.

 

[barjohn]

I will run more tests this evening and see what I get. When I use flash or in the day time exposures appear to be either on or just slightly under.

 

[Topdog1]

I will have to run some metering tests, as I have not done it, but I have been doing alot of theater photography with bright highlights and deep shadows. I would say that my R-D1s deffinitely underexposes by about 1 stop. I notice I have to consistently set the exposure +1 from the meter or the shadows will go completley dead in that situation.

/Ira

 

[Topdog1]

Here's a shot taken at iso 1600 with my canon 50mm lens @ f2. I had the EV set at +1 but wanted to keep the shutter speed up so it was underexposed and had to push the raw file exposure +1 in lightroom. You can get away with this but expect some shadow noise.

Tim

View attachment 39421,

Amazing how little noise there is. Did you do any post-processing noise reduction?

/Ira

 

[Didier]

John

RAW files are better underxposed and then processed during converting - this also minimizes the noise. Overexposing in dark light is the best way to PUSH the noise, IMO. I believe the R-D1 metering in low light is quite smart.

This picture was shot at available light, ISO 1600, f1.4. The JPG looked very black and I first thought it's waisted disk space. Then I edited in the RAW converter, see below. I first believed it's still quite noisy, but then compared it to M8 and 5D shots - since it looks not that bad at all.

Didier

 

[Topdog1]

Didier,
Very nice. Yet another reason I must begin shooting raw and not just jpg.

/Ira

 

[Didier]

Very nice. Yet another reason I must begin shooting raw and not just jpg.

Ira
I knew I gonna convice you one day! 😀
Didier

 

[Didier]

Here's another crop from the same picture, of a more noise-critical area:

 

[Topdog1]

Didier,
It is hard to believe the raw converter pulled that photo out of that black goo. That's really magic. (Now where did that Epson raw converter go?)

/Ira

 

[ampguy]

Picasa can do it, no reason a fancier program couldn't. No need for raw here, unless V2 firmware screws things up in JPG.

 

[barjohn]

Didier,

Those results are similar to what I have seen (looking at the 100% crops) but when you expose longer the images at 100% are much cleaner. As long as you are below the level that blows desired highlights (sometimes you may not care) you get a cleaner image I think. Like I say, I need to do more controlled experiments to understand the relationship between the camera's metered value and the value I chose to use to overexpose rather than underexpose. I don't have a handheld meter any longer (years ago was last time and no other camera that gives me the data) to compare. Maybe some of you with a hand meter can give us a plot at verious light levels comparing the R-D1's readings to the light meter's.

John

 

[Topdog1]

Picasa can do it, no reason a fancier program couldn't. No need for raw here, unless V2 firmware screws things up in JPG.

Picassa is my standard photo editor. I don't think it can do that without introducing a bunch of noise. At least that's been my experience with it. What function would you use in Picassa to brighten Didier's dark jpeg? Better yet, why not get a full res jpeg from Didier and see what you can do with it using Picassa. I would be very interested to see the resullts.

/Ira

 

[ampguy]

noise settings in camera?

noise settings in camera?

John,

Keep in mind that if NR is set to Hi in film settings, then for exposures longer than 2 seconds, NR in-camera is automatically performed.

Didier,

Those results are similar to what I have seen (looking at the 100% crops) but when you expose longer the images at 100% are much cleaner. As long as you are below the level that blows desired highlights (sometimes you may not care) you get a cleaner image I think. Like I say, I need to do more controlled experiments to understand the relationship between the camera's metered value and the value I chose to use to overexpose rather than underexpose. I don't have a handheld meter any longer (years ago was last time and no other camera that gives me the data) to compare. Maybe some of you with a hand meter can give us a plot at verious light levels comparing the R-D1's readings to the light meter's.

John

 

[ampguy]

Hi Topdog

Hi Topdog

I find the "I'm feeling lucky" works well, but sometimes I manually move some of them there sliders by hand. You can upgrade to the latest version on my blog site. Thanks.

Picassa is my standard photo editor. I don't think it can do that without introducing a bunch of noise. At least that's been my experience with it. What function would you use in Picassa to brighten Didier's dark jpeg? Better yet, why not get a full res jpeg from Didier and see what you can do with it using Picassa. I would be very interested to see the resullts.

/Ira

 

[mwooten]

Picassa is my standard photo editor. I don't think it can do that without introducing a bunch of noise. At least that's been my experience with it. What function would you use in Picassa to brighten Didier's dark jpeg? Better yet, why not get a full res jpeg from Didier and see what you can do with it using Picassa. I would be very interested to see the resullts.

/Ira

Ira,
Download a copy of Adobe Lightroom for free (while you can) to develop your raw images. It works wonderfully, and is a great "fileing system."

Back on subject; I am no expert, but in normal AE use I leave my camera set on +3/4. I read last week or so where one of our members (I sorry I forgot which) wrote that when set to 1600iso it is best to over expose 2 stops. The sun is going down soon, and I'll get out my camera, handheld meter and run some test shots to see if I can find any weirdness.

Take care,
Michael