The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Make some noise ...

Knorp

Well-known member
A new interesting article about noise from Dpreview.
Perhaps you already knew about this, but ETTR is the way to go for our relative small sensors.
Anyway, I think worth a read and looking forward to next article in this series: "What's that noise"

source: dpreview.com

Kind regards.
 

k-hawinkler

Well-known member
A new interesting article about noise from Dpreview.
Perhaps you already knew about this, but ETTR is the way to go for our relative small sensors.
Anyway, I think worth a read and looking forward to next article in this series: "What's that noise"

source: dpreview.com

Kind regards.
Thanks Bart.
If I am not mistaken ETTR seems necessary for older Canon sensors.
I found ETTR not helpful for my Leica M9 nor for my Sony sensor based cameras.
Those seem to be able to recover dark areas pretty well.
But one cannot recover lost highlights.
I think when using ETTR with the Sony sensors that also seems to change the color balance.

The article certainly points out why I also keep FF sensor cameras - namely for not so well lighted scenes.
When there is enough light MFT seems to work just fine.

One thing annoying about the Olympus OM-D cameras is the fine greyish line separating starkly differently lit areas.
That's very prevalent in the OOC JPG images. It takes some work to weaken this effect in images based on .ORF files.

What's your take on this? TIA.
 

Godfrey

Well-known member
Thanks Bart.
If I am not mistaken ETTR seems necessary for older Canon sensors.
I found ETTR not helpful for my Leica M9 nor for my Sony sensor based cameras.
Those seem to be able to recover dark areas pretty well.
But one cannot recover lost highlights.
I think when using ETTR with the Sony sensors that also seems to change the color balance.

The article certainly points out why I also keep FF sensor cameras - namely for not so well lighted scenes.
When there is enough light MFT seems to work just fine.

One thing annoying about the Olympus OM-D cameras is the fine greyish line separating starkly differently lit areas.
That's very prevalent in the OOC JPG images. It takes some work to weaken this effect in images based on .ORF files.

What's your take on this? TIA.
ETTR has been around a long while. Seems to me that it's often misunderstood and over-used as a concept. It's not just "overexpose and pull down in post" ... It's "get as much of the data into the upper two thirds of the sensor range without hitting saturation where you need detail." Exposing a sensor to saturation always loses data.

I'm curious about the "fine grayish line..." I've not seen it in my E-M1 photos. Can you post an example? Thanks!

G
 

raist3d

Well-known member
Another issue with ETTR (warning: I haven't read the article of dpreview in particular) is that it works best at the lowest ISO of the camera.

If you are, for example, going to use ETTR at ISO 400, you are better off to use it at the lowest ISO because you re letting the camera shift for you the exposure anyway and normally you are "darkening the exposure." Unless of course you really need a faster shutter speed, in which case it's a convenience by the camera.

- Ricardo
 

Godfrey

Well-known member
Another issue with ETTR (warning: I haven't read the article of dpreview in particular) is that it works best at the lowest ISO of the camera. ...
The reason for this is that at higher ISO settings, you are reducing the camera's dynamic range. ETTR as a technique works best when the sensor dynamic range is larger than the subject dynamic range. As soon as you step the sensitivity above a particular sensor's native point, you're cutting dynamic range.

ISO-less sensors (like the M9 and some of the Sonys) benefit less from ETTR than those that achieve higher ISO through hardware amplification.

The Kodak CCD sensor in the E-1 is an interesting case as it seems to be hardware amplification up to ISO 800 and software amplification beyond that (to its limit at ISO 3200). Prior to LR4 and Process 2012 raw conversion, I would always keep the E-1 under ISO 800, 400 for best results. With the new algorithms in LR after that, however, I found that good B&W was achievable right up to the ISO 3200 limit and color at ISO 1600 was no slouch: it retains remarkable dynamic range even at the elevated sensitivity settings.

The current sensors I've been using (the A7, the E-M1, the M-P, the E-PL7, the X) have amazing dynamic range*and remarkable exposure systems, to the point where ETTR becomes a somewhat academic thing. However, I do find that most cameras are generally set to underexpose and protect highlights more than necessary still; I'm often using +0.3 to +1.0EV compensation. The M-P, like my old Panasonic G1, is one of the few cameras where I find myself using fractional -EV nearly as much as fractional +EV settings.

G
 

Knorp

Well-known member
Thanks Bart.
If I am not mistaken ETTR seems necessary for older Canon sensors.
I found ETTR not helpful for my Leica M9 nor for my Sony sensor based cameras.
Those seem to be able to recover dark areas pretty well.
But one cannot recover lost highlights.
I think when using ETTR with the Sony sensors that also seems to change the color balance.

The article certainly points out why I also keep FF sensor cameras - namely for not so well lighted scenes.
When there is enough light MFT seems to work just fine.

One thing annoying about the Olympus OM-D cameras is the fine greyish line separating starkly differently lit areas.
That's very prevalent in the OOC JPG images. It takes some work to weaken this effect in images based on .ORF files.

What's your take on this? TIA.
Hi there K-H,

ETTR is in particular good practice for smaller sensors like our μ43rds, I think.
I frequently noticed noisy or blotchy backgrounds even in bright conditions and even with low ISO values.
Found my EM1 more prone to this phenomenon than my EM5.
It always kept me puzzled what was causing this behaviour, but I do tend to slightly underexpose my shots (ETTL).

As I read the article bottom line is to avoid your camera amplifying the signals because that will introduce the noise and catching light or photons as much as possible seems to be the way to go.
Anyway, you better read the article and the follow-up instead of me trying to explain this ... :eek:

Kind regards.
 

Knorp

Well-known member
ETTR has been around a long while. Seems to me that it's often misunderstood and over-used as a concept. It's not just "overexpose and pull down in post" ... It's "get as much of the data into the upper two thirds of the sensor range without hitting saturation where you need detail." Exposing a sensor to saturation always loses data.

I'm curious about the "fine grayish line..." I've not seen it in my E-M1 photos. Can you post an example? Thanks!

G
Hi Godfrey,

Indeed, ETTR is not about overexposure but keeping to the right just as much as possible while avoiding or accepting a small amount of highlight clipping.

Kind regards.
 

k-hawinkler

Well-known member
Thanks Bart. Yes I read the article before my above post. I mostly agree.
However, I also found with my M9 that without highlight clipping ETTR can damage tonal values. Furthermore, the M9 and Sony FF sensors permit recovering shadow areas extremely well. So, of course, one shouldn't deliberately underexpose shots. But overexposing them without highlight clipping is not necessarily helpful for all kinds of sensors.

My friends tell me that their Canon 5Dm2 benefitted from ETTR. However, they find that this particular technique isn't necessary anymore for their 5Dm3.
 

Knorp

Well-known member
Thanks Bart. Yes I read the article before my above post. I mostly agree.
However, I also found with my M9 that without highlight clipping ETTR can damage tonal values. Furthermore, the M9 and Sony FF sensors permit recovering shadow areas extremely well. So, of course, one shouldn't deliberately underexpose shots. But overexposing them without highlight clipping is not necessarily helpful for all kinds of sensors.

My friends tell me that their Canon 5Dm2 benefitted from ETTR. However, they find that this particular technique isn't necessary anymore for their 5Dm3.
Hi K-H,

It's m43rds sensors that are most likely to benefit from ETTR.
And it seems Godfrey also came to that conclusion often using +0.3 to +1.0EV.
With my M9 -0.3EV was standard and I can't recall the same kind of noisy results as I found with my EM1.
So for my EM1 it's now standard 0EV and when required I won't hesitate to go for +0.3 or +0.7EV.
That said one should also consider what kind of metering mode is used.
Mmm, time to do a bit of experimentation ...

Kind regards.
 

Knorp

Well-known member
One thing annoying about the Olympus OM-D cameras is the fine greyish line separating starkly differently lit areas.
That's very prevalent in the OOC JPG images. It takes some work to weaken this effect in images based on .ORF files.
I'm curious about the "fine grayish line..." I've not seen it in my E-M1 photos. Can you post an example? Thanks!

Hi there K-H,

I'm also interested in that 'fine greyish line' you mentioned: could you please post us an example ?

TIA :)

Kind regards.
 

k-hawinkler

Well-known member
Hi there K-H,

I'm also interested in that 'fine greyish line' you mentioned: could you please post us an example ?

TIA :)

Kind regards.
Sure, Scott Kirkpatrick pointed it out first for hi-res files. Here is his post. http://www.getdpi.com/forum/629997-post112.html
There is more discussion here. http://www.getdpi.com/forum/630121-post115.html
But these halos also show up in lo-res files, especially in OOC JPGs.
I am sure if you pixel peek your images you will see this stuff.

I have a bunch of images, hi- and lo-res of the same wires, that clearly shows this.
But so far I have not gotten around to writing this up. :eek:
Too much fun taking pictures with all sorts of gear! :D
 

Knorp

Well-known member
Sure, Scott Kirkpatrick pointed it out first for hi-res files. Here is his post. http://www.getdpi.com/forum/629997-post112.html
There is more discussion here. http://www.getdpi.com/forum/630121-post115.html
But these halos also show up in lo-res files, especially in OOC JPGs.
I am sure if you pixel peek your images you will see this stuff.

I have a bunch of images, hi- and lo-res of the same wires, that clearly shows this.
But so far I have not gotten around to writing this up. :eek:
Too much fun taking pictures with all sorts of gear! :D
Mmm, though pixel peeping is not my profession, I don't see those halos in my EM1 images, K-H ... :confused:
But there's plenty noise to be found ... :(

Kind regards.
 

k-hawinkler

Well-known member
Mmm, though pixel peeping is not my profession, I don't see those halos in my EM1 images, K-H ... :confused:
But there's plenty noise to be found ... :(

Kind regards.
Well Bart, in that case please ignore my post and enjoy your camera. :banghead:
 

Godfrey

Well-known member
Sure, Scott Kirkpatrick pointed it out first for hi-res files. Here is his post. http://www.getdpi.com/forum/629997-post112.html
There is more discussion here. http://www.getdpi.com/forum/630121-post115.html
But these halos also show up in lo-res files, especially in OOC JPGs.
I am sure if you pixel peek your images you will see this stuff.

I have a bunch of images, hi- and lo-res of the same wires, that clearly shows this.
But so far I have not gotten around to writing this up. :eek:
Too much fun taking pictures with all sorts of gear! :D
Interesting, but like Bart I don't see it in my files. Perhaps I just don't shoot subjects where it's an issue...

G
 

k-hawinkler

Well-known member
Interesting, but like Bart I don't see it in my files. Perhaps I just don't shoot subjects where it's an issue...

G
Yup, when I take pictures of my children and grandchildren I don't have the problem either. But as Scott pointed out one can see the halos also in Robin Wong's review for the E-M5 II. I need to check his other reviews though.
 

Knorp

Well-known member
Well Bart, in that case please ignore my post and enjoy your camera. :banghead:
Sorry K-H, didn't want to upset you :eek: ... really too bad about those halos.
And thanks for the example of course !
But again, it's the noise in my images that upsets me and the reason for starting this thread.

Meanwhile: mind your head ... ;)

All the very best.
 

Godfrey

Well-known member
Hi K-H,

It's m43rds sensors that are most likely to benefit from ETTR.
And it seems Godfrey also came to that conclusion often using +0.3 to +1.0EV.
With my M9 -0.3EV was standard and I can't recall the same kind of noisy results as I found with my EM1.
So for my EM1 it's now standard 0EV and when required I won't hesitate to go for +0.3 or +0.7EV.
That said one should also consider what kind of metering mode is used.
Mmm, time to do a bit of experimentation ...

Kind regards.
Photos from any digital camera benefit from proper exposure. Most manufacturers figure that the majority of users shoot JPEGs and calibrate the metering to protect highlights. The histograms, etc, are generally working and calibrated to look at the JPEG previews, and are not telling you what the raw files are getting unless you know the differentials.

When you're capturing raw files, you need more exposure. It's that simple. Whether you call it ETTR or proper exposure, the principle is identical. Proper exposure for a digital image is setting the Zone IX exposure range just shy (about half a stop) of sensor saturation.

My Olympus E-1 typically underexposes by almost a whole stop if I use a gray card. The JPEGs look fabulous, the raws get a little noisy that way. My Panasonic G1 not only had more accurate ISO markings than most, it had nearly spot on metering ... with the result that casual users overexposed all the time on the JPEGs. The raw files were near spot on. Later Panasonics went more conservative just like Olympus.

A friend with a Nikon D200 ages ago complained with how noisy his images were at ISO 400. I did some calibration testing and found that the meter needed a +0.7 EV baseline adjustment to expose properly. He couldn't believe the difference ...

Just enough exposure to do the job is always the best policy. The top end of a digital sensor has a hard limit (saturation), the bottom end (blackpoint) is always a soft limit based on how much noise is acceptable; you can shove it all over the place. The distance between saturation and black is the sensor's total dynamic range at that ISO setting.

What I do with my E-M1 is set up the JPEG rendering specifics to push the histogram (based on the raw thumbnail) so that it matches what I see in the .ORF files in Lightroom as closely as possible. Careful juggling of saturation, contrast, and sharpening will net a JPEG that looks almost identical to my .ORF preset defaults in LR. Once I have that calibration in place, I can rely upon the in-camera metering and histogram display to give me what I expect in the raw files. I don't care what the JPEGs look like most of the time... :)

Just keeping it simple.

G
 

k-hawinkler

Well-known member
hmm from what I gather performing ETTR is pretty much what ISO100 on the E-m1 does right?
Likewise the E-M5 II does seem to get the exposure pretty much right most of the time - and not just for base ISO. :)
Despite some of its short comings it's one of my favorite cameras at the moment! :clap:
 
Top