Lightroom images "underexposed"?
MileHighAko
Registered Users Posts: 413 Major grins
Still learning here...
I just noticed when I import a NEF image into LR, regardless if I convert to DNG or not, the image seems to be slightly underexposed as compared to ViewNX. Even the histograms show this. If I export to at TIFF from ViewNX, and import into LR3, the image has the exposure expected from ViewNX.
Why are images imported directly into LR3 darker and less exposed than images imported directly into ViewNX?
Example:
#1 Histogram of DNG
#2 Histogram of TIFF imported from ViewNX Export
#3 Histogram of NEF inside ViewNX
I'm considering changing my workflow to import to ViewNX first, convert to TIFF, then import to LR3. ?? I'm crazy. :dunno
I just noticed when I import a NEF image into LR, regardless if I convert to DNG or not, the image seems to be slightly underexposed as compared to ViewNX. Even the histograms show this. If I export to at TIFF from ViewNX, and import into LR3, the image has the exposure expected from ViewNX.
Why are images imported directly into LR3 darker and less exposed than images imported directly into ViewNX?
Example:
#1 Histogram of DNG
#2 Histogram of TIFF imported from ViewNX Export
#3 Histogram of NEF inside ViewNX
I'm considering changing my workflow to import to ViewNX first, convert to TIFF, then import to LR3. ?? I'm crazy. :dunno
0
Comments
Author "Color Management for Photographers"
http://www.digitaldog.net/
Thanks for the quick response. I think I might set my standard import preset to auto-exposure. Heretofore I haven't been using any type of import preset. What do most people do on import as far as pre-processing?
Keep in mind, that when you initially import the raws, there is either a default out of the box, a default you create or update or the ability to add a preset on import. But they are all just starting points. If you did not actually over expose (clip highlights in the actual raw you wish to retain) or under expose, the Exposure slider settings by and large will control overall exposure rendering. Having a default you build or pick on import that gets you closer to the goal saves you time, but you can do it upon import or after. Each converter will have differing such settings too. You can always futz around with the default rendering (any of the settings in Develop) and update that as a new Lightroom default (alt/option click on the Reset button in Develop).
Author "Color Management for Photographers"
http://www.digitaldog.net/
I just read through this article, and it was really interesting!
It seems really strange to me though, that to make the best use of the camera sensor that I would need to over expose ever image that I take. Shouldn't this be something that is baked into the camera, that It looks at the settings that I have and automatically exposes it to fit in the most amount of data possible without clipping highlights, unless I intentionally overexpose?
It seems that with the method outlined in the article I would always be guessing as to what is correct for a particular scene. Let's say that a normal shot I over expose by +1 to get a normal exposure. Then I would need to see if there was anything in the scene that I wanted actually overexposed so that the subject was properly exposed. Then I would need to guess as to what the in-camera metering was doing with regards to the exposure, and then add in exposure comp. to adjust for that. And this whole time I actually have no idea if what I am adjusting for is having the effect that I want, since the screen does not reflect it!
What do you pros do? How do pros capture the most amount of data for a given image (given that more information is available in the top half of the image)
This is an interesting discussion that I think very few photographers probably truly understand what is going on behind the scenes, myself included.
Don’t think of it as over exposure. Think of it as ideal exposure for low ISO raw capture.
The camera is designed to meter and expose for the JPEG which is not a linear capture. Its very similar to film which also isn’t linear (anyone remember the H&D curves)?
Like the old film days, you need to run a test of the exposure and development process.
If you are shooting at low ISO values and want the least amount of noise, you don’t under expose the raw data. That’s the basics of the article.
Author "Color Management for Photographers"
http://www.digitaldog.net/
It isn't "overexposure" as long as you're not clipping...
Here are more articles on the same subject. ETTR is well known/debated in the industry...
http://www.luminous-landscape.com/tutorials/expose-right.shtml
http://schewephoto.com/ETTR/index.html
I think the reason it is not built into cameras is that people expect a "normal" looking picture out of the camera. Too many students drop the exposure when shooting, because they like the moodiness and intense colors, but all that does it set them up for LOTS more noise. They should be going the other way, like Ansel Adams, shooting for a high-quality negative and not for the back of the LCD.
Also, most people shoot JPEG in their cameras, and JPEG is best exposed "normally."
The issue is, the histogram on the camera is showing you the JPEG processed from the raw by the camera and in terms of the actual raw data is a big fat lie! You can have a camera Histogram that shows clipping and there isn’t a lick of such clipping of the raw data. You can easily adjust the Exposure slider in LR or ACR and see no clipping at all. If you expose properly for JPEG you under expose for raw. One’s gamma corrected (H&D, Toe and Shoulder) data, one is linear. The Histogram shows you only one interpretation. There’s no reason at all the manufacturers could not show you a linear Histogram of the raw data. Use the camera LCD for JPEG interpretation but for raw, its simply not showing you the actual data.
FWIW, proper exposure is photography 101! Even with digital (and film of course). You could under expose or “over expose” film, based on what your meter told you and process the data (push or pull for chrome) and for neg, be a mile off. At least in my day shooting film professionally, we always purchased film in bricks of the same emulsion and did a film and process test to nail the exposure. This is not at all new for digital capture. But treating a JPEG and raw the same is like treating E6 chrome and C41 neg, of differing ISO’s as if they are the same. Sure, you can have a neg that’s a stop down and make a print, but its hardly ideal in terms of getting the best neg. Same with digital capture.
Author "Color Management for Photographers"
http://www.digitaldog.net/
I never thought about that. You're saying that the histogram shows you the representation of the JPEG displayed in the viewfinder, and NOT the actual raw data. Humf... that's a bummer.
Exactly.
Author "Color Management for Photographers"
http://www.digitaldog.net/
Absolutely not. In fact, that’s one reason whey Adobe provided a means to build DNG profiles to more closely mimic the JPEGs from the camera. And if you alter picture styles (which have zero effect on the raw), those settings do have an effect on the JPEG and the Histogram! You might try to interpolate the JEPG Histogram as some kind of rough basis for the raw data, much like looking at an old Polaroid could be interpolated to reflect the chrome, but its a stretch and requires a lot of mental gymnastics and experience. Further, the data provided in on the camera has no direct relationship to the raw data and has a direct relationship on the JPEG IF you asked for that JPEG.
IF you expose for proper JPEG in camera processing, you will effectively underexpose for raw for the use of ETTR to minimize noise. The data from each is just vastly different. IF you expose for the best possible raw data, you’ll over expose the JPEG. Again, treating the two as if they are remotely close only results in one or the other being less than ideal in terms of the captured data. One’s got a gamma curve, the other contains half of all the data in the first stop of highlight tonality.
In the case of the 5D used in the article above, the JPEG Histogram and meter was exposed 1.5 stops farther and there was no highlight clipping in the raw data. Not even close, 1.5 stops! And you can get that ETTR to look much like the JPEG on the camera, in terms of exposure by building a preset (the problem of the original topic) that normalizes the Exposure slider for the ETTR for better data.
Author "Color Management for Photographers"
http://www.digitaldog.net/
One must understand what their sensor is capable of capturing when their ISO and meters are “calibrated” for the raw processor. While it would be great if the camera manufacturers would provide a switch to show us the linear encoded Histogram we no more need the LCD and Histogram to properly expose digital than we needed (and didn’t have) such “tools” to properly expose film! Again, its Photography 101! You also have to understand what the meter is telling you (if its not an incident but rather a reflective meter), the dynamic range of the capture device and how to place the exposure onto this sensor when the scene exceeds its dynamic range. Just like we do with film. An “ISO” 100 color neg (in quotes because until you actually run exposure and development tests, the value on the box is a starting point and will likely not produce ideal neg exposure) and “ISO” 400 transparency film cannot and should not be treated the same in terms of exposure and development which is what assuming a raw and JEPG are equal is assuming. And remember Expose for the shadows, develop for the highlights in film (neg)? Again due to the H&D curve we don’t have in raw capture, the idea behind ETTR is Expose for the highlights, develop for the highlights (and the shadows result in less noise).
So yes, meters that behave in terms of linear capture and Histograms that provide this data would make the process a lot easier. But all one has to do is decide if they want ideal low ISO and low noise raw capture, run some exposure tests and then understand how to use a light meter. 101 basic stuff.
Author "Color Management for Photographers"
http://www.digitaldog.net/
One question that I am trying to grasp is why is the effect of ETTR greater at low ISO than high ISO?
Because as you raise ISO, you increase noise as a result (its a gain) and its effectively the same as using a lower ISO without ETTR (under exposure).
Author "Color Management for Photographers"
http://www.digitaldog.net/
Funny, its how I took your posts.
Unless I tell you that you are an idiot (which would be rude), you should take that feeling you’ve generated as an internal issue! I didn’t say it, I don’t believe it, you apparently may. I can’t help you with self esteem issues.
I do not understand the sentence.
The JPEG is built by the camera from the raw just as if you asked it for solely a JPEG. Its a camera processing the raw, in this context to produce a thumbnail. FWIW, that thumbnail is discarded and regenerated right after you import into Lightroom, ACR and probably most if not all other raw converters. Its no more “correct” in terms of the data inside the raw as if you rendered the raw in a dozen different converters using a dozen different raw processing engines using dozens of different presets or defaults.
The camera produces a rendering from the raw. Its not right, its not wrong, its a rendering. The reason most of us shoot raw instead of JPEG is we want to control this process.
No, its not. Its initially a JPEG the camera built with the picture styles, then LR builds its own previews (you have options for how this is done) but at this point, the picture styles are out of the equation. The picture styles are proprietary metadata. LR doesn’t understand nor use them. The manufacturers raw converter may and often do use this proprietary metadata. LR and all non manufactures raw converters don’t.
Again no, that’s not true. The Adobe provided DNG profiles attempt to mimic the picture styles that may match the names of said Picture Style. But those renderings are unique to Adobe raw engine just as the manufacturer’s rendering using the proprietary metadata are.
The Adobe raw engine has absolutely no idea what the picture style is nor can it use that proprietary data. Do some research on this topic. IF you want an initial raw rendering that matches the in-camera JPEG, you have to be using the manufactures converter (and said converter has to have the ability to read, and use the proprietary metadata to produce a match). One of the Adobe built DNG profiles may get yo closer to a visual match to this rendering but YMMV.
Which results in under exposure in the raw data. Or you could apply the DNG profile, expose correctly for raw, run the Exposure slider down to account for the ETTR you properly used, and be fine. DNG profiles are applied in a differing and earlier processing in LR/ACR then the controls in Basic. That is, you can use ETTR, drive down Exposure to normalize the rendering AND apply a DNG profile and get a closer match to the picture style AND have less noise due to ETTR/Proper raw exposure. But again, the DNG profiles are not going to produce an exact match to the JPEG from the camera outside the manufacturers raw converter. The metadata is as useless to LR as a LR metadata file is to the manufacturer’s converter.
Its different no matter what DNG profile you apply!
Look, think of the JPEG rendering in camera as a preset as we have in ACR. You expose for the JPEG, the rendering produces a fixed rendering. Now suppose you expose properly for raw (ETTR). The camera could easily alter this “preset” in effect moving the “exposure” slider down 1.5 stops if you use my 5D sensor tests as an exaple. Your JPEG would look just fine! But your raw would be properly exposed too, with less noise. This is not what happens but it could.
You can’t separate exposure and development! Its no different with raw as it was with film.
ONLY if they understand the true exposure the sensor, producing the raw data can accept and they extrapolate what the incorrect Histogram, based upon the “preset” rendering in-camera applies to said exposure. In current products, its not taking the raw linear data and the rendering one would apply to normalize the rendering. They produce a rendering based on under exposure of linear raw data.
Yes because you don’t seem to understand some facts about exposure, raw data, rendering and testing exposure and the effect of development. You think that exposing for raw and JPEG are somehow equal or related and they are not. You think that the JPEG histogram has a relationship to the raw data using optimal exposure for raw which is not the case. You can easily test this yourself as I have. Bracket and shoot raw+JPEG. Use what your meter says is correct. Look at the various Histograms on the camera. Then look at the various Histograms in LR as you normalize the exposure (using the exposure slider). If you find, as I and others have, that you can add at least 1.5 stops more exposure (photons) to a raw exposure, then adjust the rendering using Exposure in LR, then see how much less noise in the shadows as you’d get if you just kept the “normal” JPEG exposure, I think you’ll ‘get it’. You’ll also see that the ETTR exposure will produce a vastly different looking camera Histogram from the “truth” (the ETTR histogram with proper rendering).
Author "Color Management for Photographers"
http://www.digitaldog.net/
I feel like I am speaking a different language here, not because you disagree with me, but because when you read my posts, you read whatever you want it to mean, and reply to stuff I didn't say. Also making sure to drive home the point of how wrong I am and how this is all "basic photography 101", "101 basic stuff", etc.
This has been your 12th reply to me like this in 8 days!
Attack me however you like in whatever way you like, as much as you want but just so you know, I would never saying anything like that to you!
MileHighAko, sorry for all the drama
This was not directed solely to you although you seem to feel the need to take personally.
IF this is your idea of attack, you have pretty thin skin but sorry. You made some technically incorrect statements, I simply corrected this.
Author "Color Management for Photographers"
http://www.digitaldog.net/
http://schewephoto.com/ETTR/index.html
Author "Color Management for Photographers"
http://www.digitaldog.net/
http://digitaldog.net/files/ETTR_Histograms.pdf
Author "Color Management for Photographers"
http://www.digitaldog.net/
These are great articles, Andrew. I read Michael's original articles about ETTR, and found them quite informative, and convincing, but mis-understood by many posters, who missed that the you are not advocating frying the highlights, but merely moving the whole data array to the right as long as the contrast range permits this shift to occur without frying the highlights.
One of the issues, is that the ISO value reported/suggested by the manufacturer is not actually correct for RAW files with this technique. Like you alluded to above, one can find their own ISO by using this technique to identify an optimum ISO for your specific camera - and it will be lower than that reported/suggested by the manufacturer, hence shifitng the data array to the right, increasing the exposure to the right, increasing the actual number of photons for each grey step in the image array.
Almost no one shoots Kodak grey step scales anymore. And that is too bad.
Learning to shoot, and to capture, the full array, from deep black, to blistering, but not blown white, is quite instructive to the shooter, on how accurate the exposure must be for an optimum file.
Moderator of the Technique Forum and Finishing School on Dgrin