Need to Reset Over Exposed Pixels to Luminance 0

[Gerald Reisz]

I have a number of total eclipse photos shot at different shutter speeds. I want to black out (set to luminance of 1 or 0) the totally over exposed pixels in each image before combining them. With the hundreds of tools in PhotoShop, I imagine there is an easy way to do this but I haven't figured it out. I am a beginner at PhotoShop. I normally use Lightroom. I am running Release 25.7.0.

Thanks for the help!

[c.pfaffenbichler]

Please post sample images. 

A black Solid Color Layer with the appropriate Blend If-setting might suffice. 

Edit: 

[Gerald Reisz]

Thanks for the response. I have attached a sample file. The original file is a Fujifilm RAW file (RAF extension). I opened it in PhotoShop which opened the file in Adobe Camera RAW. I then saved the image file in the PSD format since it looks like I can do far more editing in PhotoShop vs Camera RAW. Now using PhotoShop I cropped the image to reduce the file size (113 MB to 14.4 MB). This is a typical total solar eclipse photo. It is one image from a set of nine bracketed images shot at various shutter speeds to capture the details of the solar corona. The dark circle in the center is the moon.

The area around the moon is burnt out with luminance values of 250 to 255. It has no useful data to include in the final combined image. I want to reset the pixels in this area to a value of 0 or 1, completely black. I would repeat this process for each image, save these images with over exposed areas set to 0, and then combine the images to get a view of the total corona. The Threshold slider seems to be in the right area except it is doing the opposite of what I want to happen.

[r-bin]

The Threshold slider seems to be in the right area except it is doing the opposite of what I want to happen.

By @Gerald Reisz

I don't know how this will help you, but you can do this.

Make a copy of the layer and convert it to a smart object.

Apply Image->Adjustment->Threshold 250 to the layer.

Apply Image->Adjustment->Invert.

Set the layer's blending mode to Darken.

If everything is ok, flatten the layers or change the Threshold value by double clicking.

[Gerald Reisz]

Your image looks exactly like what I have imagined. I will try it. I also found a Threshold mode of Subtract which may do the same. Do you know if Subtract is equivalent to Invert plus Darken?

[r-bin · Accepted solution]

Your image looks exactly like what I have imagined. I will try it. I also found a Threshold mode of Subtract which may do the same. Do you know if Subtract is equivalent to Invert plus Darken?

By @Gerald Reisz

Your method is much more convenient than mine. )

No, Darken and Subtract modes are different things.

In Darken mode, the result will be the darkest (per channel) color of the pixel from the current layer and everything under it.

The Subtract mode is the subtraction of pixel values (in each channel). Since in your case the original treshold layer gives only white or black, then subtracting black (this is zero) does not change anything in the resulting pixels, and subtracting white (i.e. subtracting 255) just resets everything to zero (and even makes it negative, as it were, but in reality there is no less than zero)

[c.pfaffenbichler]

The area around the moon is burnt out with luminance values of 250 to 255. It has no useful data to include in the final combined image. I want to reset the pixels in this area to a value of 0 or 1, completely black. I would repeat this process for each image, save these images with over exposed areas set to 0, and then combine the images to get a view of the total corona. The Threshold slider seems to be in the right area except it is doing the opposite of what I want to happen.

I am not sure I understand. 

[D Fosse]

I still dont't understand the end goal of this exercise.

As for the original question ("I want to black out (set to luminance of 1 or 0) the totally over exposed pixels in each image"), that has been fully answered by c.pfaffenbichler. His procedure does exactly that.

But what is supposed to come out of this? How does this make it any easier to combine these multiple exposures into one?

[D Fosse]

And one more thing: There is much more highlight information in the raw files than what you get straight up at default slider positions. This is where you would use the Highlight slider in ACR to recover those highlights.

It may not be enough to cover the full dynamic range here, bit it should reduce the number of frames needed considerably.

You say, "I can do far more editing in PhotoShop vs Camera RAW" - but that's not correct. It's the opposite. Opening into Photoshop throws away a lot of data, permanently.

A raw file is linear sensor data. A modern sensor can capture up to around 14 stops dynamic range. A gamma-encoded RGB file can only hold about 7 or 8 stops and still retain a natural-looking contrast curve. But that extra data in the raw file, although initially hidden, can be brought out with the Highlight slider, and remapped into the 8 stop gamma representation.

The golden rule with raw files is always to do as much as possible in ACR. Only when those options are exhausted, when you can't do anything more with the ACR sliders, do you move to Photoshop.

(there are a few special exceptions to that rule regarding color corrections, but this isn't one of them).

[Gerald Reisz]

Thanks for your comments. The end goal is to get the best possible image of the solar corona from a set of photos taken during a total solar eclipse. The normal starting point is a bracketed set of photos taken at various shutter speeds keeping the aperature and ISO constant and saving the images in a RAW format. In my case I used shutter speeds from 1/1250 second to 1.3 seconds. The result is some images with excellent detail of the edge of the sun with the rest of the image close to black (1/1250 sec) and other images with great detail of the far corona but the area closer to the sun completely burnt out (1.3 seconds). The other images give detail of the corona at various distances from the sun.

There are various videos on YouTube (e.g. two by Naztronomy) showing a process for combining the images. Most of the videos I have watched use PhotoShop or PhotoMerge. For example, the procedure by Naztronomy starts with the PhotoShop (not ACR) command "Files / Scripts / Load Files Into Stack". You select all the RAW image files and click OK. His images were Canon CR2, mine were Fujifilm RAF. PhotoShop brings all the RAW images into a stack. You then get a Load Layers window. You then align the images (critical step), and crop as needed. Skipping a lot of detail, you finally execute "Layer / Smart Objects / Stack Mode / Mean" to get a combined image and then do a Save As. The interesting aspect to me was this process, and others, leaves all the burnt out pixels in the individual images so they are still there contributing to the final combined image. 

My experimental question was "Will I get a better end result if I first get rid of all those burnt out (clipped) pixels before combining the images?" That question lead to my search.

Relative to your comment that my question "has been fully answered by c.pfaffenbichler. His procedure does exactly that." Unfortunately my meager knowledge of PhotoShop left me clueless as to how to execute what he shows. Extensive searching of the User Guide has not helped much. Once I figure out how to do what c.pfaffenbichler shows I will try it.

In the interim, I found the Threshold adjustment in PhotoShop but not in ACR. In its Normal mode, it seems to cut out all pixels with a luminance below the threshold level. I found the Substract mode does the opposite. It cuts out all pixels above the threshold level (I think?). I have found no documentation of exactly what Threshold does. The Threshold Exclusion and Difference modes seem to produce a feathering action around the threshold value whereas Substract is a hard cut off. The above response by r-bin helped me greatly.

At this point I have centered and cropped the images in ACR. I then saved them as PSD files which I understand retain all the RAW image data. I will next open the PSD files in PhotoShop and apply the Threshold adjustment and save the results. I will then go to the "Files / Scripts / Load Files Into Stack" command and proceed from there.

I do not understand "Opening into Photoshop throws away a lot of data, permanently." Are you saying when I save my RAF/RAW image as a 16 bit PSD file in ACR that I am throwing away data? That would be a show stopper for me. Concerning my comment "I can do far more editing in PhotoShop vs Camera RAW", editing was the wrong word for me to use. For this particular case of solar eclipse images, I don't want to "edit" any of the individual images. Editing will come later after the images have been combined. I am primarily using ACR as a way to get the images in a format that I can use in PhotoShop to then apply the Threshold adjustment.

Hopefully this experiment will yield a useful result.

[D Fosse]

Ah, OK. Stack mode > Mean should probably work. This is something I haven't used much, but perhaps I should look more into it. If there's a tutorial using this method, I'm sure it works, by discarding the black pixels.

But:

I do not understand "Opening into Photoshop throws away a lot of data, permanently." Are you saying when I save my RAF/RAW image as a 16 bit PSD file in ACR that I am throwing away data?

By @Gerald Reisz

Yes, that's exactly what I'm saying. This is in fact the whole point of shooting raw. The raw file contains a lot of data, especially in the extreme highlights, that an RGB file cannot hold because it has a much steeper contrast curve. So it will eventually be thrown out when you render it into an RGB file - unless you retrieve/remap it in ACR using the sliders.

Raw processing in ACR means selecting what data you want to keep, and what to throw out. That's what all the sliders are there for. You're mapping the available data into the gamma-encoded RGB file.

The secret is the fact that raw data from the camera sensor are linear. This means a much flatter contrast curve, which in turn allows a much wider dynamic range. A modern sensor can capture and register up to 14 stops of dynamic range. But there's no way you can squeeze all that into a finished RGB file without remapping.

And that's why a raw file has to be processed, you can't use it as it is. If you could see it directly, it would be a very flat, very dark and very tonally compressed monochrome image. It would look something like this (the green tint is because there are twice as many green-sensitive sensor pixels):

[Gerald Reisz]

Many thanks for this additional information. I thought PSD preserved 100% of the RAW image data plus the filters, masks, etc used by PhotoShop. After re-reading several articles on PSD, I am not sure why I thought that. Maybe because my "little" RAF files double or triple in size when saved as a PSD 

even if I have done no editing in ACR other than cropping and centering the moon.

[D Fosse]

The file size increase is mainly by going from one channel to three. That alone triples the size.