Histograms

Mostly, I overexpose by one stop (when I am not doing HDR, which I very frequently do).
 
The below diagram is for me a perfect explanation of the histogram
That's how I use it

View attachment 103906

Yes, a good explanation in the text, though it doesn't mention raising ISO in under-exposed situations that may well be the preferred option.

And without seeing the images to which those histograms relate, they're not only meaningless but misleading in that they imply a good histogram should have a big lump in the middle. That's a common misconception - the only tone that should always (normally) be in the middle, is middle-grey. So if you have an image of an elephant in long grass, that would have a huge lump in the middle, but other subjects may be completely different. Like the examples by Raymond Lin in post #17 - they would have big spikes on the right and left.
 
If you do ETTR, how many stops do you go on average?

It depends on the subject, and also on the camera and your processing regime. With ETTR, sometimes there is no room to push the exposure much at all, yet at other times you might go two stops or more.

The trick is to know your camera, and do a few tests to establish the point at which the blinkies begin to flash, and then by experiment in post processing, find out how far above that point you can really push the exposure before highlights actually blow. Blinkies are a warning that you're on the brink on blowing, and shooting to JPEG you'll need to be careful. But shooting Raw, you'll usually have at least one stop in hand, often more.

It depends on the subject, and whether you're happy to let some unimportant areas blow. For example, if the image contains small, very bright areas like white window frames, when they catch the sun it's hard to stop them blowing, but they probably don't matter much so let them go. On the other hand, if you have blinkies flashing on bright foreheads and cheeks - important stuff - you need to be more careful.
 
Have only just seen this thread and funnily enough I don't use it either and I thought it was just me being a bit of a luddite.

I've never been convinced by ETTR either; if you're shooting a dark subject you'll naturally get a histogram shifted to the left if you've exposed it correctly, and the histogram for a pale subject if exposed correctly will be shifted to the right. ETTR might have been good advice in the early days of digital (or possibly even now witha low-end camera) but with today's technology and software you have loads of latitude whatever your histogram says.
 
Yes, a good explanation in the text, though it doesn't mention raising ISO in under-exposed situations that may well be the preferred option.

And without seeing the images to which those histograms relate, they're not only meaningless but misleading in that they imply a good histogram should have a big lump in the middle. That's a common misconception - the only tone that should always (normally) be in the middle, is middle-grey. So if you have an image of an elephant in long grass, that would have a huge lump in the middle, but other subjects may be completely different. Like the examples by Raymond Lin in post #17 - they would have big spikes on the right and left.


Richard

I agree and disagree.

Those are general rules that people can use for indication. If you are looking at obtaining #17 pictures, you will have to be outside the general rules.
Those are for indication only.
After you need to add your eye into the pic.
 
It depends on the subject, and also on the camera and your processing regime. With ETTR, sometimes there is no room to push the exposure much at all, yet at other times you might go two stops or more.

The trick is to know your camera, and do a few tests to establish the point at which the blinkies begin to flash, and then by experiment in post processing, find out how far above that point you can really push the exposure before highlights actually blow. Blinkies are a warning that you're on the brink on blowing, and shooting to JPEG you'll need to be careful. But shooting Raw, you'll usually have at least one stop in hand, often more.

It depends on the subject, and whether you're happy to let some unimportant areas blow. For example, if the image contains small, very bright areas like white window frames, when they catch the sun it's hard to stop them blowing, but they probably don't matter much so let them go. On the other hand, if you have blinkies flashing on bright foreheads and cheeks - important stuff - you need to be more careful.

Thanks for the detailed reponse and apologies to the OP if this is slightly off topic.

What is the science behind ETTR and as mentioned above should it still be necessary with modern sensors? I tried the opposite earlier in the year when I was shooting inside a dark historic building on an old FF camera. I set everything manually using the highest settings I thought I could get away with knowing I would be under exposed. I always shoot raw. I was hoping to correct the exposure in post but unfortunately I got a lot of noise.

On my 7Dii some of my shots at ISO 1600 or even 3200 are fine for my purposes with a bit of noise reduction in post, but others just don't work. It seems to depend on the subject and the spread of colours. I will try to use ETTR to see if I get better results.
 
Histogram can tell you something is overexposed but not where, hence its not telling you anything useful

The very fact it is telling you that something is overexposed means it is telling you something useful. If it is useful to you, that is another matter.

The histogram doesn't tell you what the image will look like but tells you what brightness levels are present in the image. The type of scene you want to capture will determine what your histogram should look like.

If people don't use it, they should turn it off or ignore it, not wish it away.
 
That's a bit like saying its useful your doctor telling you you're probably ill, but not with what or what you can, or should if at all, do about it ;)

Dave

That's a bit far fetched, the histogram is not vague, it is very specific.

More like the doctor says that your blood sugar level is too high...do you continue eating copious quantities of sugary food and drinking fizzy pop?

The doctor doesn't necessarily know what you are getting up to and neither does the camera.

Like I said, it may not be useful to you, in that case ignore it ;)
 
That's a bit far fetched, the histogram is not vague, it is very specific.

I agree its very specific indeed, in fact telling you which (if not all) channels are blown, but as it doesn't tell you where in the image this is happening - i.e. the really useful bit - I can't see it ever being of any use at all

So yes, I do ignore it :)

Dave
 
I agree its very specific indeed, in fact telling you which (if not all) channels are blown, but as it doesn't tell you where in the image this is happening - i.e. the really useful bit - I can't see it ever being of any use at all
On my Nikon DSLR cameras I can zoom into the review screen with the RGB Histograms to bright areas, and the RGB Histograms redraw to show the tones in the zoomed in area on the individual RGB Histograms. It may be hard to see the Histograms going up the right hand wall, but you can see it. I'm not sure how other camera brands work when reviewing the RGB Histograms. :thinking:

If someone has already decided that they don't want to use the Histograms then such information is irrelevant. :rolleyes:
 
On my Nikon DSLR cameras I can zoom into the review screen with the RGB Histograms to bright areas, and the RGB Histograms redraw to show the tones in the zoomed in area on the individual RGB Histograms. It may be hard to see the Histograms going up the right hand wall, but you can see it. I'm not sure how other camera brands work when reviewing the RGB Histograms. :thinking:

If someone has already decided that they don't want to use the Histograms then such information is irrelevant. :rolleyes:


Blinkies make all this pointless :D

Dave
 
On my Nikon DSLR cameras I can zoom into the review screen with the RGB Histograms to bright areas, and the RGB Histograms redraw to show the tones in the zoomed in area on the individual RGB Histograms. It may be hard to see the Histograms going up the right hand wall, but you can see it. I'm not sure how other camera brands work when reviewing the RGB Histograms. :thinking:

If someone has already decided that they don't want to use the Histograms then such information is irrelevant. :rolleyes:

Can you do this with Nikon? This happens in Lightroom post processing, but not in-camera with my Canons.
 
V cool feature (y) I'm envious!

Agree, just checked to make sure my Sony can't do it, which it doesn't!

I'll say this now and expand later from a keyboard if necessary but the best digital exposure is exposed to the right without blowing the highlights in any colour channel with accurate white balance enabled, except as already mentioned some small spectral highlighlights that can be ignored at the photographers discretion.

What people misunderstand with expose to the right is the name confusing things, if it was expose to not blow highlights then you may have a graph on the right nearly touching but not blown or you might have a peak on the left, and a gap of tones then a tiny blip on the right just before blown. This is still exposed to not blow highlgihlights and you want that blip as far to the right as possible even if the main graph appears to the left, you have still exposed your key highlights to the right, and always should do so that the shadows don't suffer even more than they have to in post.

The name is confusing the technique is simple.

Caveat to this is extreme dynamic range situations where blending on static shots or blowing highlights on portraits or complicated situations where shutter speed is preferable to bright exposure with an ISO less sensor like a d810 or a7r then you deliberately place it elsewhere knowing what you are doing. For example my a7ii itself is ISO less from a point and pushes the files above ISO800, as I can do a better job in raw conversion if I need more than ISO800 I'll let the exposure darken then push later, you have more control over dynamic range too in this situation.
 
I prefer to ETTR and have Highlight Alert (Blinkies) on so in review I can see what, if anything, is clipping as overexposed and if nothing then I can adjust my exposure until something is or is very close to being overexposed

The beauty of Blinkies though is that it shows you where the overexposed areas are and you can then decide if it matters or not, the Histogram can tell you something is overexposed but not where, hence its not telling you anything useful

Dave

I respect nearly all of what you post on here but I feel you have this the wrong way round Dave.

I'm not criticising your shooting style, if quickly checking blinkies and adjusting exposure is the most effective way for you to keep taking the shots and get results then that's the main thing, we all use the tools how they suit us best. This method is especially ok when you know your camera and how the blinkies really effect the raw files.

The reason for me histograms tell you more are because, at least in the Sony, they are raw based rgb histograms. The blinkies are based on the JPEG, (notice how in standard picture mode they appear sooner than in neutral picture mode) and the JPEG itself will show clipping far before the digital negative is actually blown.

I've always ignored the blinkies and lived by the histogram, possibly the best approach is to look at both, blinkies to identify which areas, histogram to find out by how much, if at all, they are blown?
 
I respect nearly all of what you post on here but I feel you have this the wrong way round Dave.

I'm not criticising your shooting style, if quickly checking blinkies and adjusting exposure is the most effective way for you to keep taking the shots and get results then that's the main thing, we all use the tools how they suit us best. This method is especially ok when you know your camera and how the blinkies really effect the raw files.

The reason for me histograms tell you more are because, at least in the Sony, they are raw based rgb histograms. The blinkies are based on the JPEG, (notice how in standard picture mode they appear sooner than in neutral picture mode) and the JPEG itself will show clipping far before the digital negative is actually blown.

I've always ignored the blinkies and lived by the histogram, possibly the best approach is to look at both, blinkies to identify which areas, histogram to find out by how much, if at all, they are blown?


Careful - that started out like a compliment :D

And you are right in the respect that I use blinkies to tell me where the problem, if indeed it is one, is in the image - and I totally agree that the jpeg blinkies are an exaggeration on what the raw file will actually give me; you are right too that experience tells me by how much of an exaggeration that is and so I can work faster by checking the image at a glance only - as I often just don't have time to review the individual channels in consideration of the blinkies

In slower shooting though, I still find that Histograms tell me nothing useful to how I am shooting at the time, whereas having the merest hint of a blinkie tells me everything I need, and that's largely as the dynamic range of sensors these days is so huge

But its really a 'Whatever works for you' type scenario so if you (anyone) likes to use them and finds it helps that's fine by me :)

Dave
 
The reason for me histograms tell you more are because, at least in the Sony, they are raw based rgb histograms.
Interesting, I didn't know that there were any cameras that had Histogram generated by the RAW data. :eek:

I don't know if there is a technical reason for manufacturers not having the option for a Histogram derived from the RAW data, rather than from a internally generated Jpeg with all the Jpeg settings applied. :thinking:
 
Interesting, I didn't know that there were any cameras that had Histogram generated by the RAW data. :eek:

I don't know if there is a technical reason for manufacturers not having the option for a Histogram derived from the RAW data, rather than from a internally generated Jpeg with all the Jpeg settings applied. :thinking:

That may be incorrect actually, it was something I had read somewhere once...

I have just tested the camera and changing the picture styles does affect the histogram, but in neutral picture style with reduced saturation and contrast it is as close as possible to a RAW histogram as I've ever seen. I.e. almost identical to what you get in lightroom, especially if you soft proof to the same argb colour space.

In practice I use the neutral picture style with standard contrast and saturation, because a tad of headroom is not a bad thing. With the Sony compared to previous canons I do err a third of a stop darker than I used to. Easy to do when you have an extra 2-3 stops of shadows available!
 
I wouldn't have thought a raw would have a histogram, since the tones contained in it haven't been mapped to a tone curve?
 
I'm with Dave basically - ETTR technique, check blinkies, that's it.

But with all this talk about blowing highlights etc, it's worth remembering that ETTR technique is all about putting maximum exposure in the shadows - for less noise, more detail, better tone separation. That's why I might also look at the histogram to make sure the shadows have dragged themselves off the left-hand side, because traditionally there's been a heavy penalty for lifting shadows in post processing and it's better to do it at the shooting stage. The downside is you have to be careful and know the limits of your camera and processing regime so not to blow important highlights, and in practical terms ETTR usually means dropping the shutter speed so you have to watch that too.

However, with the best/newest sensors with wide dynamic range and ISO-less characteristics, you have more leeway and no additional noise penalty for lifting shadows in post - so easier, faster, better. I am looking forward to this new exposure Nirvana with my next camera :)
 
I wouldn't have thought a raw would have a histogram, since the tones contained in it haven't been mapped to a tone curve?

You are correct, I'd assumed wrongly because of what I'd read and the accuracy of the Sony histogram compared to previous cameras.

Thinking logically the histogram in lightroom has also had a gamma curve applied to it and as such is not a raw histogram either.

Interesting discussion here;

https://photo.stackexchange.com/questions/82548/why-dont-cameras-show-an-accurate-histogram
 
You are correct, I'd assumed wrongly because of what I'd read and the accuracy of the Sony histogram compared to previous cameras.

Thinking logically the histogram in lightroom has also had a gamma curve applied to it and as such is not a raw histogram either.

Interesting discussion here;

https://photo.stackexchange.com/questions/82548/why-dont-cameras-show-an-accurate-histogram

A Raw histogram is appealing, but overall I don't think it would actually be a move in the right direction. There are several fundamental changes that must be made to a Raw file before it's really worth working on (eg white balance and tone curve) and they change the effective exposure. The most annoying thing about the histogram is an inherent part of the JPEG process* that chops at least a stop off the highlights - the detail is there in the Raw, but you have to estimate how much headroom you've got when shooting.

The JPEG-derived histogram and blinkies are fine, provided you've done a bit of research first. Firstly, turn the contrast down in Picture Styles which compresses the histogram slightly and lowers the highlights clipping point a smidge. Then take a few test pictures (on a tripod, making sure the light doesn't change) at different exposures. Note when/where the blinkies just start to flash on the camera, then see where they actually clip in post processing. There will be at least a stop of extra exposure there, maybe 1.5 stops more, so that's the headroom you know you've got to play with.

*I'm not sure why in-camera JPEGs always chop the highlights. Guessing, it's a historical thing relating back to paper prints that have a relatively limited dynamic range compared to a good monitor and can't show very bright detail anyway. Rule of thumb with film and prints used to be a maximum tone range of seven stops - three above 18% grey and four below. And that's still a good yardstick IMHO, as even though many cameras and monitors can easily exceed that, it's too bright/dark to see much detail clearly. And that's the trick in post-processing - to push/pull the highlights and shadows so you can see the detail, while keeping it all looking natural.
 
Back
Top