Started by imagineD, October 16, 2016, 03:10:54 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.


Hi Everyone.
This is kinda my first post on the forum and it's about one of the most exciting features/modules on ML. Auto ETTR.
I had to do a lot of research and reading until i learned how this module and all those confusing settings work. I also noticed that there are a lot of other users who ask and dont understand all the settings. So i decided to make a post where i will share the most important information i learned in one post so anyone who is interested in learning this will be able to learn it faster and easier. Some of the info is from other web sites and the most important part is from the ML forum in all different posts and comments....

I own a Canon 650D that we all now is not quite good especially above ISO 800 and more. I was always disappointed from the noise levels this camera had especially for indoor shoots. And I was thinking to sell it and invest in more expensive camera.
Thanks to the ML forum and everyone involved, starting from the developers of the FW and the modules, I learned a lot how actually sensors capture light and generate noise.


Even I am experienced user of Photoshop and Lightroom and i understand histograms quite good, I just didn't had an idea how the camera sensors records the information and details and how that is related to the levels of details (shades of gray or however it is called).

I will share some information/links/images that I think are really important and can help somebody who want to learn all this.
"To fully understand this concept let us see how the digital camera sensor records information. The digital camera sensor which is actually a linear device with over 4000 levels of sensitivity uses the first 2048 (50% or half) of the available 4096 levels to record the first stop where the highlight information is recorded. Each subsequent stop records half the light of the previous stop, always using half of the remaining levels. So the next darker stop gets 1,024 levels, the next gets 512, the next 256 and so on."
Basically this is an image that you have to memorize in your had.

When i learned this i was like OK this is awesome, I can have more details in my image by pushing the details to the right of the histogram. Lets test it.
I made a couple shots in one with ISO 100 metered by the camera and then another with ISO 1600 manually correcting the exposure to push the histogram to the right without having too much HL clipping (I was comparing both the RAW histogram and the cameras histo - beware the second one is your enemy). Imported the shots in lightroom, and first on the ISO 100 image i raised the exposure and shadows sliders to the point i wanted my image to look. This is how I (and i suppose everyone else that have no idea of ETTR) shots and corrects the images.
I zoomed in and checked the shadows and the noise. I was like ok I already know this, low budget camera you cant expect quite clean image especially when you need to recover the shadows.
Then the fun part. Moved to the other image, and it looked awful, too bright, highlights looked blown it made no sense. The first think I wanted to check is if the RAW histogram was correct. It didn't show any clipping so i wanted to compare that with the Lightroom histogram. I turned on "Show Highlight Clipping" and i was a little surprised. Even if the image looked blown there were no highlights clipped. Next thing on my mind was to correct the image and see if all this makes sense and is it worth trying... I lowered the Exposure and Highlights sliders and all those details that looked blown started to appear. I moved to the shadows, the same part i checked in the ISO 100 version of the image.
And.... i was like  (hahaha sorry i had to post this :D )
I was really amazed, the noise in the same part of the image was so much less than the ISO 100 version. I just couldn't believe it, my poor little T4i can make this usable image on ISO1600  ???
That was the moment I just figured out I had a lot to learn and had to start right away and also not sell my low budget canon :D because now with this in my head i had completely another picture for the camera.

---- AUTO ETTR ----

So the next step,  i already knew there is AUTO ETTR module so i was like great i will just turn this on and make it work for me and all the noise levels in my future images will be optimized.
I also turned on the ETTR Hint in the RAW histogram as i wanted to experiment and find the "perfect combination". I did a lot of test indoors, outdoors, high contrast scenes, backlit subjects...
And the more i experimented the more i got confused. Sometimes it was working perfectly other times i just destroyed and completely missed the exposure. I decided to learn what are all this settings in the AUTO ETTR module. They might seem quite easy and self explanatory but once i stared testing, a completely another story...
OK Enough talking, you might be annoyed enough so far so in the following 'sections' I will paste some important info mostly from another post in the forum.

The MIDTONES SNR LIMIT varies from 1EV to 8EV.  I think of it 1EV = lots of noise and 8EV = no noise.  So if you want strict control of the noise in midtones then set it to 8 and the module will raise the highlight ignore until the midtones are perfectly exposed to the right.  If the scene has lots of dynamic rang then the highlights will be blown out (ignored) to maintain a proper exposure to the right in the midtones.

The SHADOWS SNR LIMITvaries from 1EV to 4EV. I think of it 1EV = lots of noise and 4EV = no noise in the shadows.   So if you want strict control of the noise in shadows then set it to 4 and the module will raise the highlight ignore until the shadows are perfectly exposed to the right.

The most important feature of this options is that you can independently control these two limits, one is how much noise is in the shadows, and the other is how much noise in the midtones. If we use both functions then the resulting image will depend on which of the two limits will be hit first.  By having two options measured at once, there is better control of the highlight ignore function in a much quicker way than setting the h.i. from  0 to 50%.

The module is enabled as default midtones at 6EV and shadows at 3EV.  This will ignore a great amount of the highlights in return of properly exposed midtones and shadows.   This is the equivalent of setting highlight ignore of 2 to 3%.  But if the scene has lots of shadows then the shadow limit  will kick in and raise the highlight ignore to say 10%.  This means that the module is automatically adjusting the highlight ignore to meet my shadow requirements.

From my testing:
Indoor photography ignoring highlights from windows set midtones = 6, shadows =3
For outside photography in bright lights only ignoring extreme highlights m=4,  s=2
If you have AETTR set to no clipping (0% highlight ignore, clipping mode set to no clipping, midtone and shadow limits off), there will be no clipped pixels.  Here, the EXPOSURE LEVEL TARGET defines where to set that white point. An exposure level target of -0.5EV, tries to place the brightest pixels in the image as far to the right of the histogram as possible without pushing any pixels to clipping.
This might appear as if you have pushed to much detail to white.  It has, but you need to be aware that none of that detail has actually been clipped to white.  There is a defined difference.
Have a read of this page.   -->

You can see that there are 3 zones above the highlights that still retain detailThese 3 zones are the areas where the camera is performing it's best. This is where AETTR is very useful, as it's designed to make it easy to push as much of the detail into these 3 zones as possible.
When you take one of these images into lightroom, you might see an extreme amount of very bright pixels, this is a good thing. Remember earlier that AETTR was set to ensure that no pixels have actually been clipped to white.  To fix such an image in LR, you simply have to lower the exposure.  Take the exposure slider in LR and move it to the left.

If you were to set the exposure target level to -2EV, you are telling AETTR to to make the brightest pixels in the image -2EV below clipped white.  The image might look like the correct exposure when taken directly into LR, but you have wasted 2 full stops of the sensors best performance.
Now, if we decide we want to allow some clipping (specular highlights, or just a simple exposure decision), the exposure target now becomes the point where pixels at this set clipping point are mapped.  So if we decide to allow 3% of pixels to be clipped and we set the exposure target to -2EV, those 3% of pixels will be allowed to be above the -2EV point and all other pixels will be below this point.
Now, if we decide to allow no pixels to be clipped to white (as per the first example above), but we have the midtone SNR limit set to 6EV, we are now saying, I don't want any pixels clipped to white, but if the midtone point of the exposure has a Signal to Noise Ratio lower then the point set in the midtone SNR limit settings, let's allow some pixels to be clipped to white to ensure that midtone level has the request Signal to Noise Ratio.
-0.5 EV is in a way a safety setting to go close to pure white but not 255 white.
That's correct.  And if AETTR is setting the exposure to -0.5EV, you can enable a little amount of highlight ignore to push the histogram even further to the right.



Ver usefull. Thank you for your contribution. Iam trying to use AETTR in a canon 60d for bird photo. I use high speed shooting normally at f8, só i need high ISO and my camera is not very good at this. O think he as the same sensor than the 650d. Iam Just beginning and iam still trying to understand How the software thinks.
Thank you once again. :-)👏👏👏👏

Enviado do meu thl 5000 através de Tapatalk


@imagineD, @Fdelgada

@imagineD thanks for taking the time to write down your experiences. As a landscape photographer, I think AETTR, RAW Spotmeter and Auto bracketing, used together, is THE killer photography combination.

For example, I will decide what part of the scene I wish to place in a given (Ansel Adams) zone, eg where, say, I wish the darks to show detail I would use my RAW spotmeter to exposure to, say, -4 to -3 EV, ie 4 stops down from the 0Ev highlights. Using the RAW (Ev) spot in ML is a great way to control your exposure. Also the ML spotmeter is narrower than, say, a 1 deg, spot meter, so you can target and explore an area of the scene.

[Hint to the gurus, it would be great if the ML Spotmeter could be made to work in Canon zoom mode  :)]

Having established my base (shadow) exposure, ie I have positioned the darkest bracket at the exposure I want, relative to the 0Ev highlights, I now finish my workflow by using ML Auto Bracketing which handles the highlight end, ie it captures as many images as the scene requires.

@Fdelgada personally I would think twice about using AETTR for bird photography. IMHO the critical thing in bird photography is the bird. Thus, when I do bird photography, I want a 'well' exposed bird and the 'motion control' off the bird, especially for BIF shots. I therefore spot meter for the bird, ie AETTR does a full scene exposure setting. I personally would use Tv mode and let the ISO float, at least on my 5D3.

As usual, the above is how I approach my photography  ;)


Quote from: garry23 on April 08, 2017, 10:09:28 AM
[Hint to the gurus, it would be great if the ML Spotmeter could be made to work in Canon zoom mode  :)]

5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109