Thread: In Camera Settings: do they ONLY affect the in camera jpg?

Tim Grey (http://www.timgrey.com/), another Photoshop Guru like Robert O'Toole, and someone working with Artie at several of Artie's workshops, has a daily newsletter entitled Ask Tim Grey. If is free! You can sign up for the daily email in a box provided on the front of his home page.

I receive the Ask Tim Grey everyday and, trust me , there is lots of good stuff - especially for us students.

Today, the following appeared; I do believe my reproducing it here will cause a problem because the information is disseminated for free, and I am not providing it in a commercial endeavor - only my personal endeavor to continue to learn and improve.

PS: I sent Tim an email directing his attention to this thread.

Today's Question: I have been told that it is important to control how your JPEG settings are set in the camera. Since the histogram is showing a JPEG conversion of the RAW data, it makes a difference what JPEG settings I have in my setup. For example, if I have a heightened contrast setting in the JPEG I am using it will distort the ends of the histogram in a way which doesn't reflect the RAW capture. I have chosen to make my JPEG settings completely neutral which means that I have adjusted the camera's neutral JPEG setting to further reduce contrast. Ostensibly, I am getting a histogram which more closely reflects my RAW capture. This helps me to "expose to the right" and not get unrealistic overexposure warnings. Is this correct?

Tim's Answer: Yes, indeed, you are absolutely correct. The vast majority of in-camera settings don't affect a RAW capture at all. However, they do affect JPEG captures, and by extension they affect the preview you see on the LCD (at least with the cameras I've tested for this purpose). In other words, by adjusting the color space on your camera you can influence the histogram displayed for captures. Of course, this creates a bit of an issue related to your priorities for previews. If you want the JPEG image to look as good as possible (especially as it relates to color saturation), you may want to set the color space in the camera to sRGB. If, on the other hand, you want the preview (and histogram) to better reflect the potential of a RAW capture, you should set the color space to Adobe RGB. But more to the point of your question, if you adjust the settings in the camera that affect JPEG captures but do not affect RAW captures, in most cases those settings will affect the histogram and preview of RAW captures, and so will enable you to better evaluate results achieved with RAW capture. Specifically, I suggest using the Adobe RGB color space, and setting all image adjustment controls to their neutral value. You can also experiment to determine if a further reduction in the Contrast setting might produce a more accurate histogram relative to what is possible. To do so, you'll want to compare the histogram of a RAW capture on the back of your camera to the histogram shown when you adjust the RAW conversion settings to those that produce a maximum tonal range without any clipping. If the histogram on the camera shows clipping while the histogram for the RAW conversion does not, then you should reduce contrast further on the camera if you want to produce a more accurate histogram relative to the potential of your RAW captures.

While I have know that the in camera settings do not directly impact the RAW settings, I had previously attempted to make the preview "hot"; perhaps it should be "not" for the reasons suggested by Tim.

So, teach US all/share with US all - what do you do, what are your settings, and why?

Thanks in advance for participating in what should be another interesting discussion.

Jay,

"I have chosen to make my JPEG settings completely neutral which means that I have adjusted the camera's neutral JPEG setting to further reduce contrast. Ostensibly, I am getting a histogram which more closely reflects my RAW capture. This helps me to "expose to the right" and not get unrealistic overexposure warnings."

Tim's advice as always is right on the money. And, I could not agree more with Tim's last statement.

I would further recommend lowering in camera contrast by -3/-4 depending on the camera to better reflect the dynamic range of the sensor. Knowing the full capabilities of your post-production software will allow you to envision your image in finality, before depressing the shutter.

Best,

Chas in sunny FL

Originally Posted by Charles Glatzer

Jay,

I have chosen to make my JPEG settings completely neutral which means that I have adjusted the camera's neutral JPEG setting to further reduce contrast. Ostensibly, I am getting a histogram which more closely reflects my RAW capture. This helps me to "expose to the right" and not get unrealistic overexposure warnings. Is this correct?

A couple years ago there was a discussion similar to this here somewhere. I believe it was recommended to set contrast at -4 and sat. at -1. I have been using these settings since with good results. Chas, since I believe no one consistently produces properly exposed images more then you can you concur with my settings or am I off the mark ?

Jay the advice is sound and I use the same settings as Chas on my Canon cameras. I want the histogram to reflect the RAW data so I can be sure that I capture all the data I need when exposing the image. I will then decide in post processing how I want my image to look when it comes to color saturation and contrast.

Originally Posted by Mike Tracy

A couple years ago there was a discussion similar to this here somewhere. I believe it was recommended to set contrast at -4 and sat. at -1. I have been using these settings since with good results. Chas, since I believe no one consistently produces properly exposed images more then you can you concur with my settings or am I off the mark ?

Mike,

Yes, I have been advocating this for quite a few years now.

If you are using a Canon 1D MIV I use -4 contrast, with the IDs MIII
-3 contrast. Sat is at default for both. And, thanks for the kind words, much appreciated. This is in conjunction with LR3.

Best,

Chas

Chas ,

What settings you will recommend for Nikon users as I know you have handled both systems very nicely

TFS Jay

Jay,

After trying various settings, including adobe rgb, I've gone back to default jpeg settings on the 1DIV and 5DII. First, raw data are linear, so no jpeg accurately represents the raw data. The raw conversion depends on a lot of settings, including the shape of the tone curve (the "standard" tone curve is not a gamma function--it a a constantly variable form of a gamma function).

I usually shoot raw + jpeg and when in a hurry, I review the jpegs. I find the standard settings come closest to a final image. Reviewing jpegs on a laptop is very fast, typically a fraction of a second, compared to many seconds to convert each raw file. An srgb jpeg is enough to evaluate focus and basics of lighting, composition, and exposure. If a jpeg shows merit, I'll flag it for raw conversion later.

The histogram and saturated pixels are generated from a sub-sampling of the image, so I find that judging the histogram on the camera is only a broad guide. If there are small details that are saturated, the histogram and blinky pixels may not show them, regardless of camera settings.

Often in sunlight, even overcast days, it can be difficult to judge the end of the right side of the plot to see how close one is to the edge. And you can't see the toe to the histogram which may be low on the plot (you sometimes can't even see that in photoshop). I have custom-written image processing software (scientific software) that will show these low levels (you can scale the histogram up and down to see the details). This convinced me that the camera histogram is simply not adequate and is only a rough guide.

Roger

Originally Posted by Harshad Barve

Chas ,

What settings you will recommend for Nikon users as I know you have handled both systems very nicely

TFS Jay

Which Nikon?

Some bodies only have a more or less contrast setting, set it to less.

D3 bodies... Picture Control under Shooting Menu I would set to Neutral with -3 contrast, setting all other parameters at 0. Although, you could raise the sharpening just to see the images sharper on LCD. I have not had a D3 in hand enough times to really test this, but, I think going to -3 contrast may only show a .3 stop increase in overall exposure headroom from 0 setting when viewing the in-camera histogram.

Best,

Chas

Thanks Chas , Much appreciated , will try this on my D300

I should have mentioned with Canon cameras I always use the "User Defined" Picture Style when modifying contrast settings. I fully agree...the in-camera histogram is most beneficial when combined with firm knowledge of your RAW conversion software of choice.

I and others have asked Canon countless times to either make the histogram box background lighter, or add a light demarkation line around the box so that we can more easily view highlight and shadow ending values in bright ambient light. Our request has thus far fallen on deaf ears. Suffice to say many have taken to putting a very thin piece of white adhesive tape on the LCD to judge the edges in field. I would really like to see the in-camera histogram represent a zoomed portion of the image as in CS, as this would allow better scrutiny of extreme shadow and highlight values.

Chas

moving to digital forum

Originally Posted by Stuart Hill

On another forum. It was discussed for Nikon users to use a linear tone curve and a custom w/b (looked very green) to more accurately reflect the sensors range. Maybe Roger or someone else could mention if this is worth setting up.

kind regards.
Stu

Stu,
I would need more information. For example, in photoshop/ACR one sees as default a linear tone curve, but that is not the actual tone curve. Photoshop applies a "standard" tone curve which is a variable gamma function that was defined for vidicon cameras back in the 1950s or before. Amazing but it still works very well. So in ACR, the tone curve is a straight line and when you change that line, that is a multiplied with the standard tone curve. So what I suspect is that the reference to linear is really the standard tone curve. A true linear output does not produce pleasing images in most cases. But a true linear tone curve can help recover highlights. ACR will not output a true linear tone curve (there is no way that I know of to select linear output). I use dcraw for linear output when I need it.

Roger

Thom Hogan spoke about using UniWB settings quite a while ago. Thom and I have been together on a number of occasions. Shooting side by side my metering methods produced the exact same results as did his using the UniWB method. My conclusion...the UniWB method adds yet another step in post processing that is unnecessary.


Chas

Originally Posted by Roger Clark

ACR will not output a true linear tone curve (there is no way that I know of to select linear output). I use dcraw for linear output when I need it.

Roger

From an earlier thread a while back about linear conversions with ACR I posted the steps to do it - "You'll have to save it in DNG format. Click the Save Image button, select .dng for file extension. In the compatibility drop down list, select custom and there's a linear option in the window that pops up."

Is this not a true linear but is the standard vidicon one you mentioned instead?

Originally Posted by Chris Ober

From an earlier thread a while back about linear conversions with ACR I posted the steps to do it - "You'll have to save it in DNG format. Click the Save Image button, select .dng for file extension. In the compatibility drop down list, select custom and there's a linear option in the window that pops up."

Is this not a true linear but is the standard vidicon one you mentioned instead?

I haven't tried this, but I would think the dng format is just the raw data. In that case it would be linear. Or does the dng conversion include the bayer interpolation to RGB?

If still raw, one would still need to do the interpolation to RGB to get a color image that one could save as a tif or other standard image format. It is the conversion to RGB where the tone curve is usually applied. I can test this when I get a chance (new system not yet up; old system limping). I'll do the conversion like you outline above and use dcraw to make a linear tif and see if the results agree.

Roger

I did a conversion with Breezebrowser and then ACR and the ACR is including the interpolation. Apparently ACR's linear isn't a true linear but more like Breezebrowser's combined conversion.

Found a bit of good information on linear DNGs here http://www.barrypearson.co.uk/articles/dng/linear.htm

Originally Posted by Chris Ober

From an earlier thread a while back about linear conversions with ACR I posted the steps to do it - "You'll have to save it in DNG format. Click the Save Image button, select .dng for file extension. In the compatibility drop down list, select custom and there's a linear option in the window that pops up."

Is this not a true linear but is the standard vidicon one you mentioned instead?

Chris,
I checked this out. I converted a test image with linear using ImagesPlus (100% sure it is true linear). Then did the dng conversion in acr as you specified. None of my programs that will read 16-bit images that I know can show me the true data can read the dng file. So I tried loading it in photoshop and photoshop calls acr. ACR then applies the tone curve if you want to save it as a tif.

A true linear tone curve will produce an image that appears very contrasty with only the highlights visible. Mid-tones and lower will be very dark.

regarding the web site:
http://www.barrypearson.co.uk/articles/dng/linear.htm
I don't think he really knows what is going on. When I loaded the "linear" dng file into acr, the tone curve is applied, so it is really no different than loading the raw image into acr in the first place. Why bother with dng "linear" files?

Some of what is on the above page probably is not correct, like acr reading a tif and making a linear dng file. ACR would have to do an inverse tone curve, and would need to know the tone curve that was applied to make the tif. There would be nothing "linear" about it, at least as far as image data are concerned.

I did not try reading the dng file with dcraw--I must try that.

Roger

His bit about the tiff to dng, jpg to dng confused me too and then I found that at least in 2007 versions of Lightroom, it was possible. http://blogs.adobe.com/jnack/2007/03...eg_to_dng.html I havent' checked if it still available now.

From an article on DNG workflow here http://photoshopnews.com/2005/05/23/...rkflow-part-i/
there is a comment from Bruce Fraser on linear DNG an the reason to use it. Not much... :)

"Bruce Fraser sent the following comment regarding Linear DNG: “The only reason to use Linear DNG is to feed the file to a DNG reader that can’t understand the particular flavor of DNG that DNG Converter creates for the specific camera, or to act as an interchange format-e.g., lens correx with DxO can write out a linear DNG that ACR can read. This is typically NOT an option people want, not because of size considerations, but because it’s no longer really raw-it’s half-baked. All the operations that take place during demosaicing are set in stone and can’t be redone."

Adobe is re-defining the original accepted meaning of what a linear file is?

"In Camera Settings: do they ONLY affect the in camera JPEG?"

I just attended a day long Nikon School course where they discussed this very subject. The short answer is "no" if you are using Nikon's Capture NX 2 for raw conversion and "mostly yes" if you are using anything else. Capture applies every one of the in camera settings when converting to a TIFF, JPEG, etc. Other programs may apply one or two settings, but generally very few because they must reverse engineer the raw file format.

I don't know if Canon has a similar converter.