Histogram question regarding clipping
- Thread starter kat.hayes
- Start date
As far as i can tell there is no debate here about whether or not the Lightroom and Adobe Camera RAW Histograms are the same. That is something you read into the conversation, so it seems you are confused. or am I missing something? I think it's common knowledge that both programs share some technical components, including the RAW engine. I see nobody claiming something else.snoke said:
Sulla's point was that Lightroom's Histogram shows the information from the current state of the edited picture and not the information from the underlying RAW file. Which can be seen as a distortion of information, and therefore be called a "lie" in some sense. Another term that you brought into the conversation, by the way.
So what's your point? I feel like we're discussing a misunderstanding.
Upvote
0
Precisely. The histogram for both programs changes as you make edits to your image. The reason for that is that the histogram on display shows the information of the edited picture. It shows the information of the output image, if you will. Just like the camera shows the histogram of the JPEG it would output.snoke said:
Neither shows all the information from the actual RAW file. That's all that was ever said here. Do you disagree with that? It that case, run the experiment described by Sulla to see the proof. I still think there is a misunderstanding here.
Upvote
0
I already wanted to retreat from this discussion, but now that I see it becoming more productive again, I somehow feel that I am obliged to reengage, as my earlier contributions lead to considerable misunderstandings. So, I would like to add this:
I don't consider the "additional headroom of ~1 stop" in the RAW files over the in-camera displayed histogram an "additional bonus" you get from shooting RAW, but I consider the canon / adobe algorithm of calculating and displaying the histogram as hiding from the user ~1stop of dynamic range that is actually present in the sensor data. That algorithm is what I - and apparently Juoules, too - am not really happy with and that leads to the situation that there is no Canon way to know in the field whether an exposure acutally clips or not and that makes exposure-to-the-right quite difficult.
But we learned from earlier contributions that there is a MagicLantern way to tell right away whether sensor data clip.
I take away from this passionate discussion that I really should give MagicLantern a try. So far I unfortunately haven't.
I don't consider the "additional headroom of ~1 stop" in the RAW files over the in-camera displayed histogram an "additional bonus" you get from shooting RAW, but I consider the canon / adobe algorithm of calculating and displaying the histogram as hiding from the user ~1stop of dynamic range that is actually present in the sensor data. That algorithm is what I - and apparently Juoules, too - am not really happy with and that leads to the situation that there is no Canon way to know in the field whether an exposure acutally clips or not and that makes exposure-to-the-right quite difficult.
But we learned from earlier contributions that there is a MagicLantern way to tell right away whether sensor data clip.
I take away from this passionate discussion that I really should give MagicLantern a try. So far I unfortunately haven't.
Upvote
0
A large part of the dissonance in this thread is based on an opinion that is not entirely correct, which is based on a misunderstanding of what Lightroom is actually showing you. In the Library module it is, effectively, showing you a jpeg in a small color space and the histogram reflects that, in the Develop module it is showing you something entirely different.
Lightroom does a lot of heavy lifting under the bonnet to make things seem effortless in the interface, trouble with that is it leads to many false assumptions. If you want to see your histogram from the RAW data look at the Develop module, the histogram there is the full depth information mapped into the Melissa RGB color space, which is the ProPhoto color range with an sRGB gamma curve applied (it is a custom color space used in the Develop and Print modules).
Here is an illustration. First screenshot from the Library Module with a clearly clipped (but missing spike) histogram, second screenshot same image no adjustments but from the Develop Module with highlights that are clipped but present.
So why does it do this?
The Library module is working in a different color space with a smaller gamut and the histogram is a reflection of the image as you see it, WYSIWY have. The Develop module is working in a much bigger color space and again is showing you a real time analysis of what is on the screen. By the way, the Develop module only does this for RAW files, it is quite amazing how much thought Adobe put into color management within the software so we don't need to give it a thought.
Lightroom does a lot of heavy lifting under the bonnet to make things seem effortless in the interface, trouble with that is it leads to many false assumptions. If you want to see your histogram from the RAW data look at the Develop module, the histogram there is the full depth information mapped into the Melissa RGB color space, which is the ProPhoto color range with an sRGB gamma curve applied (it is a custom color space used in the Develop and Print modules).
Here is an illustration. First screenshot from the Library Module with a clearly clipped (but missing spike) histogram, second screenshot same image no adjustments but from the Develop Module with highlights that are clipped but present.
So why does it do this?
The Library module is working in a different color space with a smaller gamut and the histogram is a reflection of the image as you see it, WYSIWY have. The Develop module is working in a much bigger color space and again is showing you a real time analysis of what is on the screen. By the way, the Develop module only does this for RAW files, it is quite amazing how much thought Adobe put into color management within the software so we don't need to give it a thought.
Attachments
Upvote
0
I knew that the develop module uses different data than the library module, as, e.g., it builds its own preview, during which at first it displays the library-module preview. I did not know, however, that it uses a different colour space and that it displayed a histogram with a greater dynamic range than the library module. I never noticed that colours and tones shift as soon as the develop module switches from the library-module-preview to its own preview. Next time I'll try to observe that carefully.
Upvote
0
That's insanely helpfull suggestioon, thanks a lot! For non-Magic Lantern cameras and people who understandingly don't feel like trying it, that seems like a nice work around from my first little test.neuroanatomist said:
I just gave it a try. I shot the blue, sunny sky at 1 second, f/1.4, ISO 100 to get a completely blown out image and used that to set my custom white balance. Works the way it should, now the Live View looks terrible but the histogram is a deal more helpfull.
I put my Rebel T3i on a tripod and aimed it at a big white cloud. I made three observations:
1) If I expose the image so that the Canon RGB LiveView histogram shows that I'm just shy from clipping highlights and take the shot, the histogram in the Image Viewer shows a great deal extra headroom, about 1.5 stops. The LiveView hsitogram is less wide than the Image Viewer histogram! They just cut away the right side! In other words, the LiveView histogram on the Rebel T3i / 600D is entirely useless for ETTR.
EDIT: Seems like additionally using the Technicolor CineStyle Picture Style reduces the difference between the Image Viewer and Live View histogram.
2) Taking shots at increasing increasing shutter speed (Longer exposure time) and looking at the Image Viewer histogram afterwards to determine the brightest exposure possible, I arrived at 1/1250 seconds, f/2.8, ISO 100. That's pretty much precisely what I would have chosen with the Magic Lantern RAW RGB histogram, which showed the clipping indicator, but without a number in it, meaning less than 1% of pixels were blown out.
Switching back to daylight WB and taking the same shot, the Image Viewer histogram touched the right border and the blinkies where on, so it thought the image was clipping even though it wasn't. Pulling back the exposure in LR, the cloud had tons of detail... I wonder if even the ML RAW histogram might not show the entire dynamic range ... a mostly white cloud isn't the best DR test subject though, is it? ;D
3) I have yet another dust spot on my sensor :-\ (I tried some smaller apertures before going to 2.8 )
Also thanks to privatebydesign, some more knowledge about Lightroom is very nice too. I guess I really have to look into color and the science behind it more at some point.
Upvote
0
Similar threads
