Ad
Ad
Ad
Pages: « 1 [2] 3 »   Bottom of Page
Print
Author Topic: Adobe RGB vs sRGB  (Read 12030 times)
Robert Boire
Full Member
***
Offline Offline

Posts: 164


WWW
« Reply #20 on: September 30, 2010, 12:23:49 PM »
ReplyReply

What difference are we talking about here? Between jpeg and RAW, or between sRGB and Adobe RGB, or...?

I was talking about the difference between jpeg and RAW with an Adobe RGB color space. Again comparing the two against each other on the computer monitor and the jpeg on the camera LCD, I did not see much of a differnce.

On the other hand my sample was small and referring to
Now read the article here on Expose to the Right, then actually test that. What you’ll see is that ETTR with proper “normalization” settings in the converter now produces a preview that matches the JPEG yet you “over exposed” (some would suggest properly expose) the raw data. And the matching JPEG with the same raw exposure? Its way blown out. There’s a really big disconnect between the raw data and the JPEG when you capture the raw data for raw (not JPEG) exposure.


I admit that several of the shots I was looking at had very subdued lighting and little contrast and no highlights to speak of, so I will do my homwork and read the ETTR article.
Logged

bjanes
Sr. Member
****
Offline Offline

Posts: 2882



« Reply #21 on: October 01, 2010, 12:30:32 PM »
ReplyReply

Not familiar with Rawnalize but would not a green clip indicate saturation clipping and not full tonal clipping until all three channels move towards a true clip?

Rawnalyze is a freeware program that looks at the raw data directly without demosaicing, white balance, or other processing. Unfortunately, the author has passed away and I don't know if the program is currently available on line. The use of the tool is briefly discussed in this Libraw aricle which deals with white balance.

The examples I showed deal with clipping in the green channel with intact red and blue channels. Of course, if white balance is applied, the channels would clip more or less equally. As you know, the green channel usually clips first with most cameras and most illuminants.  Personally, I try to avoid clipping in any channel. Although highlight recovery is useful, the colors may not accurate.

Bill
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9317



WWW
« Reply #22 on: October 01, 2010, 12:35:58 PM »
ReplyReply

The examples I showed deal with clipping in the green channel with intact red and blue channels. Of course, if white balance is applied, the channels would clip more or less equally. As you know, the green channel usually clips first with most cameras and most illuminants.  Personally, I try to avoid clipping in any channel. Although highlight recovery is useful, the colors may not accurate

Understood. But in the context of the “clippies” we see on the LCD versus the actual tone clipping of the raw data, would we not have to increase the exposure of the raw a bit more to produce an apples to apples comparison?

My original point is, 1-1.5 stops past initial clipping seen on my LCD provides a raw that, at least in LR and ACR can be adjusted such that there is no clipping (getting back to “Pardon my ignorance, but isn't "a huge stretch" a huge stretch“?).
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Daniel Browning
Full Member
***
Offline Offline

Posts: 142


« Reply #23 on: October 03, 2010, 10:09:48 PM »
ReplyReply

Unfortunately, the author has passed away and I don't know if the program is currently available on line.

It is available here:

http://dave-anderson-photo.com/blog/2010/08/23/gabor-rawnalyze-author-rip/
Logged

--Daniel
elied
Sr. Member
****
Offline Offline

Posts: 276


« Reply #24 on: October 04, 2010, 07:48:40 AM »
ReplyReply

I was talking about the difference between jpeg and RAW with an Adobe RGB color space. Again comparing the two against each other on the computer monitor and the jpeg on the camera LCD, I did not see much of a differnce.

You fail to say how you were viewing the RAW, because in fact you cannot actually view a RAW. A RAW file is not an image file. It is a collection of data that will provide the starting point for a series of operations that will in the end generate (convert to) a color image. So when you open a RAW in an image viewer one of two things has to happen - either the application does an on-the-fly generation from the RAW of a display image (which may or may not be done the same way as your camera does it when it generates a jpg) or the application is doing it the easy way and simply displaying the embedded jpg inserted into the RAW file by the camera. Obviously, in the second case the two camera-generated jpgs will be identical. So you really need to be more explicit about just what you are viewing and how.
Logged

Roll over Ed Weston,
Tell Ansel Adams the news
digitaldog
Sr. Member
****
Offline Offline

Posts: 9317



WWW
« Reply #25 on: October 04, 2010, 09:21:29 AM »
ReplyReply


My god, I had no idea we lost panopeeper who was a regular presence here. Very sad.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Mark Paulson
Jr. Member
**
Offline Offline

Posts: 89


« Reply #26 on: October 04, 2010, 11:00:33 AM »
ReplyReply

The camera histogram is pretty reliable on this camera.

I have experimented with UniWB and one thing I discovered is that the Nikon picture control and some of the other setting can greatly affect the histogram. I do not use UniWB anymore , but I found that zeroing out and turning off all of the thing on the D3 that have an effect on exposure allows me to get maximum exposure without clipping the RAW data. I generally can get 0.5 to 1.0 stop more exposure.
Logged
JeffKohn
Sr. Member
****
Offline Offline

Posts: 1671



WWW
« Reply #27 on: October 04, 2010, 03:26:44 PM »
ReplyReply

I have experimented with UniWB and one thing I discovered is that the Nikon picture control and some of the other setting can greatly affect the histogram. I do not use UniWB anymore , but I found that zeroing out and turning off all of the thing on the D3 that have an effect on exposure allows me to get maximum exposure without clipping the RAW data. I generally can get 0.5 to 1.0 stop more exposure.
Yes, agreed. I've found that for all the Nikon DSLR's I've used, using the "neutral" picture control with a flat tone-curve, Adobe RGB colorspace, and UniWB will result in the camera histogram being a pretty accurate representation of the raw data (within about 1/3 stop or so).

It's true that if you crank up the various in-camera settings so that the preview has more "pop", the histogram won't be as accurate. But I'm not sure why you would ever do that if you're shooting raw and care about histogram accuracy.
Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 9317



WWW
« Reply #28 on: October 04, 2010, 03:29:35 PM »
ReplyReply

Yes, agreed. I've found that for all the Nikon DSLR's I've used, using the "neutral" picture control with a flat tone-curve, Adobe RGB colorspace, and UniWB will result in the camera histogram being a pretty accurate representation of the raw data (within about 1/3 stop or so).

Accurate for JPEG or accurate for the linear raw data, (exposing for the best raw data ETTR)? That’s the $64K question.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
bjanes
Sr. Member
****
Offline Offline

Posts: 2882



« Reply #29 on: October 04, 2010, 08:03:59 PM »
ReplyReply

Yes, agreed. I've found that for all the Nikon DSLR's I've used, using the "neutral" picture control with a flat tone-curve, Adobe RGB colorspace, and UniWB will result in the camera histogram being a pretty accurate representation of the raw data (within about 1/3 stop or so).

It's true that if you crank up the various in-camera settings so that the preview has more "pop", the histogram won't be as accurate. But I'm not sure why you would ever do that if you're shooting raw and care about histogram accuracy.

I keep near UniWB supplied by Iliah Borg in Bank 4 on my D3. Usually the green blows first with normal WB, but with high saturation red or blue subjects, the red or blue channels may show saturation clipping after white balance and one can use UniWB to determine if the clipping was caused by white balance or WB multipliers > 1.0 in those channels or by clipping in the raw channels. However, I don't use UniWB that much.

True UniWB didn't work with previous versions of ACR, and I don't know if that limitation is present with the current version.

Regards,

Bill

Logged
Mark Paulson
Jr. Member
**
Offline Offline

Posts: 89


« Reply #30 on: October 05, 2010, 11:16:03 AM »
ReplyReply

Accurate for JPEG or accurate for the linear raw data, (exposing for the best raw data ETTR)? That’s the $64K question.

Accurate for RAW not JPEG. If you actually use UniWB to set the for WB then you will hardly ever blow any of the data. If you don't use it and zero out everything else. you can still get pretty close without blowing a channel by backing off a little from the right.
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9317



WWW
« Reply #31 on: October 05, 2010, 02:20:02 PM »
ReplyReply

Accurate for RAW not JPEG. If you actually use UniWB to set the for WB then you will hardly ever blow any of the data. If you don't use it and zero out everything else. you can still get pretty close without blowing a channel by backing off a little from the right.

I don’t understand how the UniWB has anything to do with exposure. WB for raw is just a metadata suggestion anyway. Can you clarify the workflow?
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
JeffKohn
Sr. Member
****
Offline Offline

Posts: 1671



WWW
« Reply #32 on: October 05, 2010, 03:15:19 PM »
ReplyReply

I don’t understand how the UniWB has anything to do with exposure. WB for raw is just a metadata suggestion anyway. Can you clarify the workflow?
The point of UniWB is that it makes the in-camera histogram more accurate because no WB coefficients are applied to the R/B channels. So you can see the actual exposure for those channels rather than being mislead by WB adjustments that may indicate a channel is clipped when it really isn't.

With UniWB and neutral/flat picture controls, I find the in-camera histogram to correlate quite well with the histogram in ACR. It's still gamma-corrected, of course; but it you don't need a linear histogram to judge exposure.
« Last Edit: October 05, 2010, 03:17:00 PM by JeffKohn » Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 9317



WWW
« Reply #33 on: October 05, 2010, 03:19:37 PM »
ReplyReply

The point of UniWB is that it makes the in-camera histogram more accurate because no WB coefficients are applied to the R/B channels.
Accurate for what? ETTR? The proper exposure for the JPEG? I’m confused.

Quote
So you can the actual exposure for those channels rather than being mislead by WB adjustments that may indicate a channel is clipped when it really isn't.

But how does this have anything to do with the raw data? I can take an incident reading, open up a good stop plus, shoot raw and completely bring in the clipping seen with the default raw rendering (or what the camera would obviously show as clipping of the JPEG) due to ETTR (all that actual data in the highlights that didn’t really clip). I could see how this tool would produce a better idea of clipping and exposure for JPEG. I still don’t see how a histogram and an exposure based on a gamma corrected JPEG correlates to the raw data based on a rendering setting (of the Exposure slider) it can’t possibly know exists.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
JeffKohn
Sr. Member
****
Offline Offline

Posts: 1671



WWW
« Reply #34 on: October 05, 2010, 04:05:34 PM »
ReplyReply

Andrew I'm not really sure how else to explain it, I think you're over-complicating things. The fact that the camera histogram is not linear doesn't really matter. The histogram in ACR isn't linear either.

All I'm saying is that with UniWB, flat tone curve, Adobe RGB colorspace, and other "neutral" in-camera settings, the in-camera histogram will be a pretty close match for what I would see opening the RAW file in ACR with default settings. The in-camera histogram will still be slightly pessimistic, showing clipping about 1/2 stop before the raw data is actually clipped.  But for ETTR I would rather have the highlights a hair below clipping as opposed to having even one channel clip.

Bill's point was that the green channel is almost always the first one to clip anyway, so UniWB isn't all that necessary. That's mostly true in the general case; but if shooting a subject where the red or blue channel is strongly saturated (flowers, for instance), UniWB can be useful. It's also useful if you're using a magenta filter to 'cut' the green channel and equalize it relative to the red and blue channels. Since I never shoot JPEG I tend to just leave UniWB on all the time.

Logged

JeffKohn
Sr. Member
****
Offline Offline

Posts: 1671



WWW
« Reply #35 on: October 05, 2010, 04:07:18 PM »
ReplyReply

Quote
True UniWB didn't work with previous versions of ACR, and I don't know if that limitation is present with the current version.
You may be right, to be honest I can't remember if the WB setting I got from Iliah was true UniWB or "near". The difference is negligible as far as judging exposure goes.
Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 9317



WWW
« Reply #36 on: October 05, 2010, 04:28:12 PM »
ReplyReply

The fact that the camera histogram is not linear doesn't really matter. The histogram in ACR isn't linear either.
No and neither is the data at this point. But more importantly, the ACR histogram shows you the raw data based on the current rendering settings. The histogram on the LCD shows you the JPEG rendering. If I expose such that the JPEG shows severe clipping, the clipping totally disappears in ACR, from the raw when I update the rendering for exposure (as does the clipping in the ACR Histogram).

I still can’t see how anything you do in front of or with the camera that affects the JPEG reporting on the LCD tell you the facts about the raw.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
JeffKohn
Sr. Member
****
Offline Offline

Posts: 1671



WWW
« Reply #37 on: October 05, 2010, 05:09:53 PM »
ReplyReply

Quote
If I expose such that the JPEG shows severe clipping, the clipping totally disappears in ACR, from the raw when I update the rendering for exposure (as does the clipping in the ACR Histogram).
I guess we're just talking past each other at this point. I don't know what you mean by "when I update the rendering for exposure". Are you saying that your images typically show clipping in ACR until you use a negative adjustment on the exposure slider? The fact that you can eliminate the clipping indicators in ACR with a negative exposure adjustment does not mean that your raw data wasn't clipped.

Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 9317



WWW
« Reply #38 on: October 05, 2010, 05:39:25 PM »
ReplyReply

I don't know what you mean by "when I update the rendering for exposure". Are you saying that your images typically show clipping in ACR until you use a negative adjustment on the exposure slider?

Yes, exactly. Yet there is no clipping based on this ETTR exposure to the raw data. There is clipping if I shot JPEG (or raw+JPEG) on said JPEG. IOW, the optimum exposure to avoid clipping is significantly different for a JPEG than a raw and hence I can’t see how working with controls on a camera that tells us something about the JPEG can correlate to the raw.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
bjanes
Sr. Member
****
Offline Offline

Posts: 2882



« Reply #39 on: October 05, 2010, 06:06:47 PM »
ReplyReply

Yes, exactly. Yet there is no clipping based on this ETTR exposure to the raw data. There is clipping if I shot JPEG (or raw+JPEG) on said JPEG. IOW, the optimum exposure to avoid clipping is significantly different for a JPEG than a raw and hence I can’t see how working with controls on a camera that tells us something about the JPEG can correlate to the raw.

Andrew and Jeff,

I think that this example can clarify the issue. This image was shot with the Nikon D3 under daylight using the standard picture control (neutral) and Adobe RGB (the widest space on this camera). My initial shot showed clipping on the camera RGB histogram in the green and red channels, so I decreased exposure until there was no green clipping. I didn't have UniWb at the time.

The camera histogram and the raw histogram as shown by Rawnalize are shown.  The camera histogram shows clipping in the red and green channels. The green might not clip in ProPhotoRGB, but that is not an option with this camera. The raw histogram shows that the red channel is more than 1/3 stop below clipping. However, the red multiplier for white balance for this channel is 1.6992. When this is applied, the red will be clipped. If I were using UniWB, I could have seen that the red channel was short of clipping and inferred that the clipping took place with white balance.

 

« Last Edit: October 05, 2010, 06:09:01 PM by bjanes » Logged
Pages: « 1 [2] 3 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad