To ETTR or not to ETTR...?

Started 3 months ago | Discussions thread
knickerhawk Veteran Member • Posts: 6,235
Re: example
2

bobn2 wrote:

jimkahnw wrote:

The example also demonstrates what a fool's errand we are all on. There's little difference between the 0EV and the +3EV, at least in the mid- and 3/4-tones. I know with a little more processing, there would be none. However, what about the highlights? That's the whole thing about ETTR; increase exposure without clipping--recover in post. In this example, what does the sky look like for the +3EV exposure? I would bet that the highlights are unrecoverable.

I still think ETTR is a misconception that leads to overexposure and clipped highlights. Our 4/3 sensors are fantastic at preserving highlights and shadows. We can get great images without fretting over the shadows to preserve highlight details. I'm continually amazed how much 3/4- and full-tone values can be lifted in an image exposed for the highlights.

You post illustrates the crux of the problem discussing ETTR. The technique itself was based on a false premise (that giving more exposure would result in the image being encoded in more digital 'levels' which would give better image quality. In fact, any benefit is due to maximising exposure, and nothing to do with digital encoding. However, despite being based on a fallacy, the idea and name has stuck, and I'm not sure why, because as a technique, it isn't a very sound way of maximising exposure.

While it's certainly true that Reichmann got the reason to "ETTR" wrong in his articles, it's pretty rare these days to see improved digital encoding raised as the reason to ETTR. For instance, I don't recall seeing that argument made anywhere in this thread, although admittedly, I haven't read every post here with that in mind. We've come a long way in understanding how to optimize raw exposure (in no small part thanks to your efforts and others and buttressed by Martinec's paper). For better or worse, though, the terminology has stuck even as the meaning and application of the terminology has evolved.

The reason it's stuck, I think, is that it forms a sort of a link between two very different models of exposure management.

Model 1 posits that the purpose of exposure management is to control the output lightness of the final image. This idea is so entrenched that many people do not even acknowledge that there is a distinction between exposure and final image lightness. It is from this model that there often used terms 'underexposed', 'correctly exposed' and 'overexposed' derive. They generally mean that the final image is deemed to be not light enough, a perceptually pleasing lightness or too light.

Your "Model 1" is really just the JPEG exposure model. It serves its purpose pretty darn well as a working mental model for JPEG-only shooters, especially for the majority of whom are interested in doing little or no postprocessing. However, it breaks down (indeeed, is even conceptually counterproductive) as soon as one starts thinking in terms of optimizing raw exposure. The frequent problem I see for photographers trying to bridge the two models is how they approach shooting JPEG+raw while using the JPEG exposure model. They then express surprise that their renderings from the raws aren't all that much better than their OOC JPEGs and declare that shooting raw is much ado about nothing!

Model 2 posits that the purpose of exposure management is to ensure the maximum possible information about the scene in the raw file, and if there are constraints that limit how much information be captured, that those are managed to the best effect for the desired output image. In general this means maximising exposure, except where that is in conflict with practical or aesthetic requirements. The terms 'under', 'over' and 'correctly' exposed would have completely different meanings within this model of exposure management.

ETTR provides a method (though not a great one) for using the tools and vocabulary of Model 1 to achieve at least some of the results of model 2. That, however leads to it being fundamentally inconsistent and contradictory.

I agree, but the real obstacle here is the cameras we have today are almost completely designed to reinforce the JPEG exposure model. In that regard, would you be as critical of using "ETTR" as a succinct short hand for optimized raw exposure if cameras with actual raw histograms started being released?

And, by the way, no sensor is particularly 'great' at preserving highlights or shadows, and mFT ones are certainly no greater at it than others. The question is more about the exposure you chose in relation to the capabilities of the sensor.

Post (hide subjects) Posted by
XRF
XRF
XRF
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow