Are pixels really blown in ACR?

[loosecanon]

Remember your jpeg will be affected by the picture style settings. If you ask the camera to produce contrasty punchy images then it is more likely the jpeg will lose details in the highlights and shadows.

The reason why we shoot in raw is because we have the full dynamic range the camera can record at that ISO setting, between 11 and 13 stops rather than the 8 of a jpeg. However most jpegs show even less dynamic range as many images would appear flat without this processing.

[dje]

Lon Frank and Glenn

I just opened the image in ACR (v7 which comes with PS CS6 and is the same engine as LR4) and found no clipped highlights !!

Dave

[Glenn NK]

Glenn I wasn't aware of this but have just tried it on my 600D and found that it does indeed improve the accuracy of the histogram in relation to blown highlights. I've actually got it set to -4.

Thanks for sharing

Dave

Dave:

You are quite welcome.

Another opinion: Post No.2 by tzalman (a knowledgeable photographer imo)

http://photography-on-the.net/forum/...+picture+style

And yes, some people use minus four.

Glenn

[FrankMi]

Lon Frank and Glenn

I just opened the image in ACR (v7 which comes with PS CS6 and is the same engine as LR4) and found no clipped highlights !!

Dave

The fact that we can remove the clipping of both the shadows and highlights in CS5 clearly indicates that the clipping is not 'real' so maybe CS6/LR4 ACR is presenting a more accurate picture of the values?

[Lon Howard]

The fact that we can remove the clipping of both the shadows and highlights in CS5 clearly indicates that the clipping is not 'real' so maybe CS6/LR4 ACR is presenting a more accurate picture of the values?

Thanks Frank, Dave and really everyone. I do know one thing that's real, and it's that I'll be downloading the latest update to ACR very soon. Also, it's REALly important to understand how your own equipment works, regardless of what it is.

[dje]

Thanks Frank, Dave and really everyone. I do know one thing that's real, and it's that I'll be downloading the latest update to ACR very soon. Also, it's REALly important to understand how your own equipment works, regardless of what it is.

Lon I think you can only get ACR v7 if you have CS6 rather than CS5.

Dave

[Lon Howard]

Lon I think you can only get ACR v7 if you have CS6 rather than CS5.

Dave

Aaah yes. Thanks for the reminder, Dave; I hadn't thought of that. Well, I guess if I take the Starbucks home and grind it there, it will taste just as good. I can do that for a little while.

[Bobobird]

There is an ACR update at v6.7 for PS5.

[kris]

Lon Frank and Glenn

I just opened the image in ACR (v7 which comes with PS CS6 and is the same engine as LR4) and found no clipped highlights !!

Dave

I have both CS5 (ACR V6) and CS6 (ACR V7). The latter produces better results, and you really see the difference, especially in recovering the clipped lights, both shadow and highlights.

[Glenn NK]

I have both CS5 (ACR V6) and CS6 (ACR V7). The latter produces better results, and you really see the difference, especially in recovering the clipped lights, both shadow and highlights.

Yes. I'm reprocessing my digital images - LR4 has the same improvements over LR4 - and it is noticeable.

Glenn

[kris]

I loaded the .CR2 file into ACR6.7 (CS5) and ACR7.2 (CS6).

Here are the snapshots of what you see as you open it without any corrections.

ACR6.7


ACR7.2


A.

[Lon Howard]

ACR7.2 appears to do your raw processing for you. Nice ... thanks kris!

[dje]

I've been doing a little more research into this subject and there are those around on a couple of forums who believe ACR V7/LR4 has some sort of algorithm built in to automatically recover some blown highlights from the raw file. I also found a little program called RAWDigger which produces a detailed histogram of the raw data. Here is a screen dump of the histogram of Lon's RAW image. It does appear that the green channels have some blown highlights. Notwithstanding this, I'm sure the arguments advanced above about inaccurate in-camera jpeg histograms for RAW shooting are still valid.

Dave

[Glenn NK]

I loaded the .CR2 file into ACR6.7 (CS5) and ACR7.2 (CS6).

Here are the snapshots of what you see as you open it without any corrections.

A.

Andrea:

Thanks for doing this for us - it surely demonstrates the difference in the two processes.

I kept reading that LR4 had improved the algorithms over LR3, and your test proves this (equivalent to ACR 7.2 over ACR 6.7).

Perhaps Lon will sleep better tonight - I will.

And I will keep updating the process in older images.

Glenn

[Glenn NK]

I've been doing a little more research into this subject and there are those around on a couple of forums who believe ACR V7/LR4 has some sort of algorithm built in to automatically recover some blown highlights from the raw file. I also found a little program called RAWDigger which produces a detailed histogram of the raw data. Here is a screen dump of the histogram of Lon's RAW image. It does appear that the green channels have some blown highlights. Notwithstanding this, I'm sure the arguments advanced above about inaccurate in-camera jpeg histograms for RAW shooting are still valid.

Dave

Dave:

Thank you for showing this. I was so interested that I DL'd Rawdigger. I see that it has a limited time to expiry - what does it cost to purchase/license?

From your comment about the blown G channel in Lon's image, I'm inferring that a value higher than 16,000 is blown - is that correct?

Wouldn't it be a very useful and major development if something like Rawdigger could be part of the firmware of a DSLR, rather than showing the JPEG image in the LCD screen as is presently the case?

Glenn

[dje]

Dave:

Thank you for showing this. I was so interested that I DL'd Rawdigger. I see that it has a limited time to expiry - what does it cost to purchase/license?

From your comment about the blown G channel in Lon's image, I'm inferring that a value higher than 16,000 is blown - is that correct?

Wouldn't it be a very useful and major development if something like Rawdigger could be part of the firmware of a DSLR, rather than showing the JPEG image in the LCD screen as is presently the case?

Glenn

Glenn I'm not sure what the deal is with cost, I've only just downloaded it today myself. Must look into this further.

As for the blown greens, I'm just going on the fact that there are spikes at the RH side. If you download the csv file you can see how many pixels there are for each value. This image has about 21,000 at the max value of 14,737. I suspect the saturation value varies from one sensor type to another.

Dave

[Glenn NK]

Glenn I'm not sure what the deal is with cost, I've only just downloaded it today myself. Must look into this further.

As for the blown greens, I'm just going on the fact that there are spikes at the RH side. If you download the csv file you can see how many pixels there are for each value. This image has about 21,000 at the max value of 14,737. I suspect the saturation value varies from one sensor type to another.

Dave

Dave, I've looked at a few files of my own, and can't figure out where the "blown limit" is - at either end. Looking at a file where the blacks are clipped (according to LR4), there is no indication in RD where this point is either.

So far I haven't located a manual or help file for the program.

Glenn

Ooops - found the manual with the onscreen icon.

[dje]

Dave, I've looked at a few files of my own, and can't figure out where the "blown limit" is - at either end. Looking at a file where the blacks are clipped (according to LR4), there is no indication in RD where this point is either.

So far I haven't located a manual or help file for the program.

Glenn

Ooops - found the manual with the onscreen icon.

Glenn my education is "continuing" on this subject !!

I'm not sure about black clipping yet, mainly been looking at highlights. My assumption is that the white clipping or saturation point is where there is a spike on the RHS. This is based on the assumption that when a sensor pixel goes into saturation, the sensor electronics simply assigns the value equal to pure white for that sensor. This results in an unusually high number of pixels with that value because it includes all those pixels where the saturation point is exceeded (hence the spike). Exactly what that value is though is another matter. In the case of an 8 bit jpeg, that value is "normalised" to 255. In the case of the raw file data, I think the value will vary depending on the sensor type (and maybe the camera type also). Must look into it further.

Dave

[kris]

Dear Loan and Glenn

here is a short guide by Thom Hogan about RAW files.

http://www.bythom.com/qadraw.htm

It is some years old, but it gives some insights on what your camera is doing.

Just one more comment. Programs like RawDigger are made to analyze the behavior of different sensors/camera in a Lab. While this is an interesting subject, on a technical ground, IMHO it does not help much when you take pictures. I find more useful to learn how the camera responds under different conditions and how the developing software is handling the data. It is much faster than trying to recover in PP. I have both a D300 and a D800. Under the same conditions I have to use them differently.

Cheers
Andrea

[Bobobird]

...I have both a D300 and a D800. Under the same conditions I have to use them differently.

I am not following. The "facilities" may differ from camera to camera but surely, the process of composition and exposure would be the same regardless.