M4/3 Noise

[FrankG]

I'm sorry. I think I didn't write that clearly. What I meant was I always shoot Raw

[FrankG]

Thanks for the tip on experimenting with auto bracket. I find , so far, that +2/3 brings the histogram all the way to the right. +1 stop makes it crawl up the right side. So for now +2/3 exposure comp is where I'll leave it.

But if the histogram is on the jpeg rendition of my raw file then it isn't all that accurate - right?
There's likely still some highlight wiggle room

[nma]

Hi Frank,

If you are referring to the in-camera histogram displayed with the jpeg image on the back of your camera, then I would say that it is only a rough guide. There is additional headroom before clipping the highlights, somewhere between +1 and +2 stops. Think about that in relation to your original complaints about the noise. Properly exposed you will have about 1.5 more stops of exposure in the shadows. That works for me.

I would also add, if you switched to a full frame camera, with a high dynamic range sensor without fully exposing your images, your results will be suboptimal. You can still end up with noisy shadows, or at least the IQ will not be what it could be.

[FrankG]

Gotcha.
So I can almost always have the Exp Comp set to +1 quite safely (ie without clipping highlights)
And then just reduce the exposure when bringing the raw file into ACR.
I'm pretty sure it's going to look too light/over-exposed as it is shot

Yes I was referring to the histogram rendition on the GX7 camera. I have set it to standard brightness (mode 2) and kept all other other settings neutral.
 That histogram is  the only thing I have to go by to judge exposure

anyone mind showing a screenshot of an image Exposed to the right by 1 or 2 stops Before you reduce exposure in Lr or ACR

[SZRitter]

My standard method, when using the histograms and having time to adjust, is to manual adjust exposure until the highlights are just inside the histogram, then increase the exposure by a stop to two to get about where there is full sensor saturation. A bit cumbersome, but pretty reliable. You'll need to run a few tests to see how it works on the GX7.

How do you like that camera? Been thinking of adding it or the GX8 to go with my E-M5.

[FrankG]

Oh wow. You +exposure compensation until histograms hit right edge, which is likely already +1. And then another +1 or more on top of that. And no burnt out highlights?
Can you show a screenshot of a sample before you make exposure corrections?

I love the GX7 - I love the flip screen, silent mode, flip eyepiece, touch screen (focus/shoot),wireless, menu layout, dial/button layout. It basically just suits me. Your mileage may vary.
I also have a Canon 5D3 kit but this is my take everywhere camera.

[SZRitter]

Oh wow. You +exposure compensation until histograms hit right edge, which is likely already +1. And then another +1 or more on top of that. And no burnt out highlights?
Can you show a screenshot of a sample before you make exposure corrections?

I love the GX7 - I love the flip screen, silent mode, flip eyepiece, touch screen (focus/shoot),wireless, menu layout, dial/button layout. It basically just suits me. Your mileage may vary.
I also have a Canon 5D3 kit but this is my take everywhere camera.

The method I'm describing will pretty much never burn highlights, that is the point of figuring out where they are in your histogram. Then you just need to get a good idea of how far you can push it past the histogram (and this will vary by camera) to not clip anything. This is great for landscapes and architecture, not so good for portraits, action and the likes. For portraits I usually expose properly based off a light meter, or expose +1. Also, if you know you have a highlight that will clip, say the sun, just try to find the brightest thing you don't want to clip and work from that.

What are you looking for as examples? I may have a few images I could show you on this computer.

[FrankG]

I just tested a few shots here indoors with bright snow outside (both interior & exterior in the shot)
aperture priority f4, and exp comp +1.3, 200iso
Also, some interior shots of kids playing. But these at 800iso. Also 'slightly' cleaner shadows / noise than before the +1.3 exp comp.
In ACR it only took a moderate decrease of the highlight slider to prevent clipping the highlights.
The shadows could still use a little lifting, but are very clean without the objectionable noise I was getting before.

Is there an iso limit to using ettr for cleaner noise / shadow detail ?
Or does this only work well at 200iso (800 was a wee stretch I think, but more testing to come)


has anyone with a GX7 established just how far above the right you can push the exposure and not lose highlight detail?
If not, is there an established way to shoot tests? Like. for example, shooting a white terry towel and seeing how far you can 'over expose' until you cant bring back the detail in post ?

[stamper]

FastRawViewer is the program you want.

http://www.fastrawviewer.com/

[FrankG]

FastRawViewer is the program you want.

http://www.fastrawviewer.com/

Could you please give me a quick summary of what/why I want it?
Ok just installed it to see.
Seeing the % over and under exposure for each rgb channel must be why you recommend it to me?
I can test pushing exposure to the right until I start to clip one of the channels and then back off a bit - correct?
Thx

[stamper]

Take some images of a scene focusing on the same place and increase your EV in steps of 1/3 all the way to +3. Load them into the viewer and see which one clips and then you will know where your clipping point is.

[SZRitter]

FastRawViewer is the program you want.

http://www.fastrawviewer.com/

Another software to look at is RawDigger. Looks to be the same price.

[FrankG]

there are 3 columns - under, over, & not sure what the third one is (possibly the over but with raw converter correction?)

so I just photographed some snow outside in +1/3 increments and got up to +2 stops without over exposure.
Does that mean my ETTR limit is +2 stops?
Regardless of the scene or metering method?

Here's ascreenshot

[SZRitter]

there are 3 columns - under, over, & not sure what the third one is (possibly the over but with raw converter correction?)

so I just photographed some snow outside in +1/3 increments and got up to +2 stops without over exposure.
Does that mean my ETTR limit is +2 stops?
Regardless of the scene or metering method?

Here's ascreenshot

Nope. Just curious, do you know how your camera's metering modes work?

Basically, it tries to adjust whatever you have it pointed at to 18% gray (i.e. medium gray). This is what gray cards should be calibrated to. So if you use the full scene metering, it takes all the values and averages them to 18% gray (give or take, there is a bit more to it, but that is the jest of it). If you go center or center weighted, it will use only the center to adjust to 18% gray.

Now, applying this to a scene, in full scene metering, that is 90% snow, it will have a tendency to underexpose because it is trying to adjust that white snow to neutral gray. So adding +1 or +2 probably only brought it back to where it should be in the first place.

[FrankG]

yes, I understanf meters bringing white or black even surfaces to middle grey and that one has to over or under exxpose to make them white or black.

I was just looking at the raw histogram to find the max exposure before it indicates any over exp

[SZRitter]

I would start with a gray card and a white card, if you have them. then meter for the gray card and keep upping from there, manually, until the white card blows out. That should get you a pretty good idea. Albeit, you will need to know the exposure difference between the white and gray card.

[BobDavid]

Topaz Labs Denoise 6 is nice.

[Pete Berry]

yes, I understanf meters bringing white or black even surfaces to middle grey and that one has to over or under exxpose to make them white or black.

I was just looking at the raw histogram to find the max exposure before it indicates any over exp

Frank, what you need to determine practically is how far the highlight histogram spike in-camera can "crawl" up the clipping limit and still preserve the RAW highlight detail you want. Try some skies, and the old white textured towel in the shade test to get a good idea.

Pete

[mecrox]

I tried this yesterday with my Oly em-d 5 mkII and founded +1 usually OK but +1.7 often not OK, taking cityscapes on a bright morning with sunshine and clouds. Cloud highlights blew out too easily, or at least became far too bright for my taste. But then I tend to prefer more somber shades.

My experience is that strong ETTR can introduce a few gotchas, at least with my Oly: skies can become too cyan, for example, and some images require quite a lot of PP to get things back into balance. Highlights reduced too much in LR can acquire a sickly, waxy look, at least on the honey-coloured stone around here. I can see this being a fair system for shots with a wide DR where I know I will need to raise shadows a lot, but for more general shots I'm thinking that sticking with the camera metering is a simpler system, albeit adjusting for bright and dark scenes where reducing things to an 18% grey will give an incorrect exposure (snow, black doors, etc). Or maybe +0.3 or 0.7 for general things, but not going for broke with +2 - too much work in PP.

Bracketing and then combining using LR's HDR feature is another way to do this, especially since LR now outputs a fresh RAW of the result. That way, I still have a useable shot at default settings even if the second one is blown out.

[AFairley]

With the Oly E-M5 I found that I could pretty reliably shoot at +2/3 without hardly ever clipping, if I went to +1 I would get some clipping in maybe 1 in 15 scenes.  (That's just relying on the cameras matrix metering.). Of course this is highly dependent on the kind of stuff you photograph.