ETTR - do you use it often?

[davidedric]

Thanks for all the contributions. I wasn't trying to start World War 3, but asked in the general spirit of enquiry

It just struck me that sensor design has improved so much over the years, that practice could have changed - and I expected a variety of opinions.

I don't know enough to have an opinion to offer. When shooting, I am usually more concerned with avoiding blown highlights (especially on those darn white birds), than in thinking about ETTR - but it is something that I can now take into account when the conditions seem to demand it.

Dave

[inkista]

My only quibbles with this article are that the author assumes one's using an extremely low-noise camera. If you have to use a high ISO, I think ETTR suddenly becomes way more attractive (improved SNR)....

... except for the fact that everyone thinks they can detect blown highlights at the time of shooting when it's very possible they can't. I think that's the bit that ctein is saying that needs to get hammered home. ETTR is great in theory if you actually can tell when you're blowing highlights. But look again at the histogram of his image, and look again at the bit where he shows where the highlights are actually blown. There's no way I would have seen that on a camera back. For me I'd rather put up with noise than blown highlights (or motion blur). You can fix one in post, it's a lot harder to fix the other. But, as I said, my tolerance for noise and doing fancy digital things to bracketed sets of frames is higher than most; YMMV.

I far prefer a technique like Guillermo Luijk's ZeroNoise combination of multiple exposures for noise reduction, and I'm kinda overjoyed to hear the Magic Lantern team has succeeded in pulling off their "dual iso" trick (similar to ZeroNoise) in-camera, and that they're tackling getting it into the single-processor cameras now that they've cracked it for the dual-processor 7D/5DMkIII.

[RustBeltRaw]

... except for the fact that everyone thinks they can detect blown highlights at the time of shooting when it's very possible they can't. I think that's the bit that ctein is saying that needs to get hammered home.

Very true. My camera's blinkies definitely have a higher threshold than Photoshop, and only respond to the average histogram, not each channel. So maybe ECTTR is the way to go. Expose close to the right.

[NorthernFocus]

... except for the fact that everyone thinks they can detect blown highlights at the time of shooting when it's very possible they can't. I think that's the bit that ctein is saying that needs to get hammered home. ETTR is great in theory if you actually can tell when you're blowing highlights. ...

Like anything else, one has to understand the equipment and know how to use it. If you do, you can detect blown highlights.

[inkista]

Like anything else, one has to understand the equipment and know how to use it. If you do, you can detect blown highlights.

I state again: look at that histogram and KNOW that there are blown highlights in that data. And that's the in-Photoshop histogram, accurate to the RAW data. Both the in-camera histogram AND blinkies would been even less accurate and would both indicate there were no blown highlights. Remember, both that histogram and the blinkies are derived from the LCD preview JPEG image, which is severely downsized. The disposal of over 90% of the image data before analysis is done shifts the histogram and blinkies until they lie to you.

So, what exactly is the magic understanding of my camera and the technique I need to learn how to use that would work for me, here? I'd really love to know.

[Colin Southern]

There's more to blinkies ...

- Yes, they're based on an in-camera JPEG, but

- Don't forget that pushing past that threshold effectively removes a lot of safety margin with regards to approaching the sensors saturation point AND THUS FLIRTING WITH IT'S NON-LINEAR REGION.

Personally, I use blinkies to highlight areas I need to be aware of (think of it as a "caution flag" for those areas - no more, no less) - I also use the histogram to indicate areas of under-exposure (with "under-exposure" in this case being defined as data levels less than the sensor can handle).

But in all cases, it comes down to the dynamic range of the scene to be captured. If it's just a reflective scene - and you're only capturing 4 stops with perhaps another couple for shadows/"good luck" then ETTR is pointless as even a standard exposure will keep you 4 stops above the noise floor (assuming base ISO & a 12 stop DR modern camera). It's when you start doing things like using high ISO (reduced DR capability) or shooting into the light and still wishing to retain foreground shadow detail (high DR requirement) that ETTR becomes essential. At other times it just increases the risk (of weird & impossible to correct colour shifts due to entering the sensor non-linear response regions) and it increases the workload (in that tones then need to be moved back down - right back to where they were in the first place!) -- all for a theoretical benefit that not realised in a real-world result (ie you get cleaner data that you can't see).

Sensor data is gambler territory -- ya gotta know when to hold em - know when to fold em - know when to walk away - and know when to run. Blinkies / histograms / ETTR -- they're all just tools. Each has it's place, but no tool is right for every job.

[Wayland]

If you are worried about blowing almost specular highlights then pop a couple of bracketed shots off.

It's what we had to do with film and it's easy enough if you are on a tripod.

You can always blend them back in during processing if needed but it means you will retain better tonal range in your shadow areas.

Of course, if you are not on a tripod you are probably not that bothered about the quality anyway...

[dje]

Sensor data is gambler territory -- ya gotta know when to hold em - know when to fold em - know when to walk away - and know when to run. Blinkies / histograms / ETTR -- they're all just tools. Each has it's place, but no tool is right for every job.

Oh no, so it's not just LeAnn Rimes but Kenny as well (I have a long memory). Oh well, to each his own.

Good quote though Colin !


And Dave (davidedric), intentional or not, I think you have started WW3. But it's the fault of those dam Nikon shooters with their ever so superior sensors

Dave

[Colin Southern]

Oh no, so it's not just LeAnn Rimes but Kenny as well (I have a long memory). Oh well, to each his own.

Moved on from Leanne, although I still love the voice. Running with early Neil Diamond work at the moment (Juliette, Sunday Sun, Two-Bit Manchild etc)

[NorthernFocus]

...So, what exactly is the magic understanding of my camera and the technique I need to learn how to use that would work for me, here? I'd really love to know.

Sorry if I touched a nerve there. By "you" in my last post I intended the global you, not YOU. I also casually chose my words. Predict may have been more appropriate than detect.

Unfortunately I can't share the magic formula . But the good news is it doesn't require one either. One just has to develop an understanding of one's equipment and how it performs(e.g. metering, dynamic range of sensor) in different lighting conditions. Understanding the equipment includes how to interpret the histogram, blinkies etc. For example sitting down and looking at images and comparing the histo on the camera back with what it looks like on one's software of choice. Particularly when one ends up with different results than expected. For the way that I set up my camera and the subject matter that I typically shoot, the histogram is pretty darn reliable. Unfortunately now I'm getting at that age that I can't see the darn thing without reading lenses so recently I've been studying the "blinkies" which historically I've never used. In the context of studying the equipment, what I've seen so far is that I can comfortably shoot 1/3 stop into blinkies and not do any clipping. But again, that works for that one body with my normal wildlife set up. I would not arbitrarily assume the same results if I was shooting saaaay... the Taj Mahal at night.

One problem with all articles like the one referenced is that the authors, by definition, are writing from their own frame of reference on the subject matter and covering a broad audience. We all shoot different stuff under different conditions. Personally I'm always amazed at how much there is to learn with every new subject that I decide to shoot. And yet the pundits (and people in internet forums) go on line and proclaim how we should all shoot. For example making statements that noise is less of a problem than blown highlights. Really? Doesn't that really depend on the subject? And noise due to underexposure isn't that much worse than base level noise. Again, REALLY? Frankly that particular author lost all credibility at that point.

[NorthernFocus]

... Blinkies / histograms / ETTR -- they're all just tools. Each has it's place, but no tool is right for every job.

IOW, one needs to understand lighting conditions and one's equipment

Speaking of knowing when to walk away, that's what I call this image:

[Colin Southern]

IOW, one needs to understand lighting conditions and one's equipment

I would put it like this ...

"one can never have too much knowledge -- one can however have insufficient knowledge".

Is it advantageous to understand lighting and ones equipment? Absolutely. I'd add to that, a knowledge of general digital photography principles as well.

To be honest, I think many photographers are pretty weak when it comes to understanding things like scene & sensor dynamic range, and the implications therewith when it comes to things like ISO selection and ETTR.

[Colin Southern]

One problem with all articles like the one referenced is that the authors, by definition, are writing from their own frame of reference on the subject matter and covering a broad audience. We all shoot different stuff under different conditions. Personally I'm always amazed at how much there is to learn with every new subject that I decide to shoot. And yet the pundits (and people in internet forums) go on line and proclaim how we should all shoot. For example making statements that noise is less of a problem than blown highlights. Really? Doesn't that really depend on the subject? And noise due to underexposure isn't that much worse than base level noise. Again, REALLY? Frankly that particular author lost all credibility at that point.

I'd go so far as to say that photography is always about making compromises. The more one understands the limitations of the equipment, the better the choices they can make. Extreme contrast night scenes are a good case-in-point; point light sources will always blow, and noise can usually be clipped - usually it's far more important to get the midtones right.

Look at this old image - it has lots of blown highlight - clipped blacks to control noise - and I've sold lots of copies on 22 x 44" canvas

[Brownbear]

I've been following this thread with interest, and today is my day to ask what is likely to be a dumb question.

I'm shooting jpeg and raw, and because my focus is on photographing birds in flight and wildlife, often in low light and hoping to avoid excess noise at high isos, I'm trying to expose to the right and stock piling raw files for the magical day when I'm a whiz at editing my raw files.

I've noticed that the histogram for all my jpeg files is farther to the left side then the histogram for my raw files... This morning I also just noticed that although I can't see any clipping of highlights in my jpeg files, (or in my view finder) I can see clipping in some of the raw photos typically in the white bits of bird wings in some of my raw files but not the jpegs. Why is this?

Thank you.

[NorthernFocus]

Christina, the jpeg version of the photo has been processed by whatever picture control setting you are using. Though most of those setting would tend to increase contrast therefore increase likelihood of clipping the right side. But if you have D-Lighting turned on it typically reduces dynamic range by compressing the histogram. Are you comparing the histograms in LR?

[Brownbear]

Dan,

Thank you for your reply. I don't have D Lighting turned on, so far the only modes/settings I've used on my camera are S, A and Manual.

Yes, I'm viewing my histograms in LR and all the jpegs are towards the left and the raw files are on the right.

[Brownbear]

Oops..

I am viewing the histograms for the jpegs and the raw files in Picasa. After I upload my files to my pictures file in Windows, I view them in Picasa and delete the poor shots, then I upload to LR for editing. In LR I only work with the raw photos. (the jpegs are not uploaded into LR)

Dan,

Thank you for your reply. I don't have D Lighting turned on, so far the only modes/settings I've used on my camera are S, A and Manual.

Yes, I'm viewing my histograms in LR and all the jpegs are towards the left and the raw files are on the right.

[Cantab]

Oops..

I am viewing the histograms for the jpegs and the raw files in Picasa. After I upload my files to my pictures file in Windows, I view them in Picasa and delete the poor shots, then I upload to LR for editing. In LR I only work with the raw photos. (the jpegs are not uploaded into LR)

Christina, what happens when you compare blinkies in camera with the clipping in LR? I have several memory cards on the go so I was able to put my bald eagle card back in the camera to see which images showed up as clipped there. But as Dan noted earlier, the histogram (and presumably blinkies) are based on the JPEG, with whatever jpeg picture style your camera has baked into the image. But as he said most of the picture style settings (eg. contrast) would increase the likelihood of the jpeg clipping.

(I think what Dan referred to as picture control setting is what in the Canon world are picture style settings. These are not modes such as shutter priority or manual but things that the camera does to the jpeg image before baking it. But since I know next to nothing about Nikons, I'll be interested to see what others say.)

[Brownbear]

Hi Bruce,

I'm pretty good about adjusting my exposure as I take photos, so no clipping or very little seen...

It might be the way Picasa displays photos. In Picasa I can view the jpegs and raw files, and the jpeg histogram is to the left with no clipping.. The histogram for the raw files in Picasa is much more to the right (presumably because of exposure compensation) and show clipping, but I don't see the clipping in LR in the raw files so I'm guessing that maybe it is just something to do with Picasa's program. No clipping is seen in the jpegs in Picasa. So maybe it is nothing for me to worry about.

Yes, that makes sense the jpegs are edited by my camera and perhaps to the left to correct for me exposing to the right? Truly I'm just guessing.

Christina, what happens when you compare blinkies in camera with the clipping in LR? I have several memory cards on the go so I was able to put my bald eagle card back in the camera to see which images showed up as clipped there. But as Dan noted earlier, the histogram (and presumably blinkies) are based on the JPEG, with whatever jpeg picture style your camera has baked into the image. But as he said most of the picture style settings (eg. contrast) would increase the likelihood of the jpeg clipping.

(I think what Dan referred to as picture control setting is what in the Canon world are picture style settings. These are not modes such as shutter priority or manual but things that the camera does to the jpeg image before baking it. But since I know next to nothing about Nikons, I'll be interested to see what others say.)

[rhadorn]

I use ETTR most of the time to cover the mostly huge dynamic range of landscapes. In addition, I am not fond of spectacular global contrast effects but rather in fine local contrast over the whole tone range, so I need detail all over the range.

Now, the main issue is to me that usually the default tone curve of raw developers compress the highlights so as to simulate pictures taken with the good old film. This may be o.k. for a sunny holiday picture on the beach, yet at the cost of highlight detail, which may be of importance in other pictures.

My response to this is to reduce this compression, f.e. using the 'linear' developement curve in Capture One rather than the normal one. This moves the histogram a great deal to the left and 'opens' the highlights by spreading them over a greater segment of the tone range. Now I build up tone range, global contrast, local contrast and saturation to my taste.

To be noted: the 'linear' curve in C1 does not set back Gamma to 1, data are just less compressed towards the high end of the scale.

It is actually a pity that we have to juge exposition looking at histograms, which show what would be the JPEG image if we would not shoot raw... Changing from Normal to Linear in C1 gives an idea about how far from overexposing we actually are when we shoot to the right on the basis of the JPEG histogram... at least when using a modern broad range camera.

So, ETTR still makes sense to record the most extended tonal range the camera can 'see' but should not be dissociated from postprocessing strategies.

Reto