Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: ACR has given me the blues - comments please  (Read 12969 times)
xpatUSA
Sr. Member
****
Offline Offline

Posts: 293



WWW
« on: July 13, 2012, 09:42:19 AM »
ReplyReply

My first post here, greetings to All,

I normally shoot watches for the web at my bench, Sigma SD10 with Sigma 70mm macro lens, usually low-res. However, on the way to do that, I took a snap of a sunflower at 2 pm in bright sunlight. The resulting image has had me puzzled for about a week. ACR appears to have added some blues in the highlight areas of the flower petals:



I use ACR 5.4 with PS Elements 6 and am quite happy with those for my purposes. I also use Sigma Photo Pro 5.1 for quick jobs, and have been playing with DCRaw lately. The image above is a screen crop showing two files open in PSE6. At left you see the image from ACR, all sliders zero, no sharpening or noise reduction, "embedded" camera profile, with it's blue histogram moved over. At right you see the image from SPP, again all sliders were at zero and it's histogram showed no blues above level 0. Various runs of DCRaw did not "add" blues either. At first, I thought they were specular reflections but then they would appear with any RAW converter, IMHO.

I'm looking for suggestions as to where these blues came from. I'm not really looking for advice on how to fix it in ACR, and please don't blame the Foveon ;-)

The raw X3F file and it's side-car file are up on my site here, if anyone wants to play with them:

http://kronometric.org/images/post/phot/acr/

Thanks,

Ted
« Last Edit: July 13, 2012, 10:55:27 AM by xpatUSA » Logged

best regards,

Ted
MichaelWorley
Jr. Member
**
Offline Offline

Posts: 67


« Reply #1 on: July 13, 2012, 12:31:11 PM »
ReplyReply

Ted,

   Most images consist of Red, Green and Blue (RGB).

   I just sampled the yellowest section of an Arizona poppy, and the RGB values were R233, G218, B40. The 40 Blue value had a greater amplitude than the Red and Green, and did not extend beyond the left quarter of the histogram.   But the image did not look blue in any way.

   The 233, 218, 40 were simply the values necessary to make the yellow at the location I sampled.

   You might use the eyedropper tool and open the info pallete and click in various places in your image to see the color values that are there.

Mike
Logged
xpatUSA
Sr. Member
****
Offline Offline

Posts: 293



WWW
« Reply #2 on: July 13, 2012, 02:45:26 PM »
ReplyReply

Thanks Mike,

I went pixel peeping and it seems that ACR handled the X3F raw data for this image differently and perhaps more honestly than the others. In ACR the red channel is clipped in that area and the blue is as high as 130 - e.g. RGB = 255, 221, 130. Outside the highlighted area, starting from just a few pixels away, the blue content is mostly zero RGB = e.g. RGB = 240,188, 0 (more clipping?). So, with your help, I have deduced that the RAW file itself was probably clipped but I don't have the software to check on that . . . yet ;-)

Ted
Logged

best regards,

Ted
MichaelWorley
Jr. Member
**
Offline Offline

Posts: 67


« Reply #3 on: July 13, 2012, 04:42:18 PM »
ReplyReply

Ted,

   I hadn't thought about clipping. Having spent the last 33 years in the red rock country of Arizona, it had been drilled into me as second nature to check the histogram on the LCD. (A lot easier when they became three color.) After a few thousand blown red channels, I learned to look at and rely on the LCD. And, to be honest, some of the poppies photographed on the same day as the one I checked had blown channels. Apparently, a lesson learned about one color didn't quite click when it came to another. A lot of funny-looking poppies photographed that day. Not the one I checked, of course; the blown ones having been long deleted.

Mike
Logged
xpatUSA
Sr. Member
****
Offline Offline

Posts: 293



WWW
« Reply #4 on: July 13, 2012, 09:28:37 PM »
ReplyReply

The plot thickens . . if I save that same raw file (using Sigma PhotoPro 5.1) as a ProPhoto 8-bit TIFF,  then open it in PhotoShop Elements 6 (PSE6) there are no signs of in-camera clipping. There are no 255's and no zeros of any kind in the entire image. The lowest blue in the region of interest is 32 and the highest red is 231. The whole image histogram as shown in PSE is smooth with no tall peaks at either end that you get with clipping. I should mention here that PSE6 does actually recognize and display ProPhoto TIFF images.

So, thinking that the petal colors are probably all out-of-gamut with respect to sRGB, I opened a ProPhoto image in PSE6, reduced the saturation to move the colors into the sRGB gamut and then converted the color space to sRGB. After a few tries at that and a bit of tweaking in the final image I got this:



The highlight color clipping seen in the OP is gone. However, a lot of trouble to go to, seeing that I'm not a flower close-up shooter.

Especially when Sigma Photo Pro does the same thing in one click ("auto" button). Yep:



A bit of conversion clipping exists and the hue is a little too yellow, but not an objectionable image for one click, IMHO.


I'm back to thinking that, although ACR 5.4 may well have converted the image successfully into it's RIMM 16-bit working space, the image's journey from there to a) ACR's histogram, b) ACR's color-picker and c) onwards to PSE6's ROMM working space was less than smooth. What do y'all reckon?


Comments are still welcome.

best regards,

Ted
« Last Edit: July 14, 2012, 06:21:00 AM by xpatUSA » Logged

best regards,

Ted
Nigel Johnson
Full Member
***
Offline Offline

Posts: 124


« Reply #5 on: July 14, 2012, 06:17:58 AM »
ReplyReply

Ted

Both ACR and SPP will apply their own rendering to the raw file, apply their own tone curves, carry out their own highlight and shadow adjustments, and possibly highlight and shadow restoration (attempting to create acceptable colour when one or more of the channels is clipped in the raw file). It is possible that SPP is recognising that red is being clipped and reducing the level of all the channels resulting in no clipping in the highlights, in doing this it is probably also taking account of adjacent non-clipped colours and Sigma's understanding of the spectral sensitivities of the sensor.

Regards
Nigel
Logged
xpatUSA
Sr. Member
****
Offline Offline

Posts: 293



WWW
« Reply #6 on: July 14, 2012, 06:35:40 AM »
ReplyReply

Yes Nigel, I fully agree. Please see my edited post above for a sort of confirmation of your view. Also, I'm not ruling out the possibility that the problem areas are specular reflections (quite large amounts of blue channel) which ACR perhaps deals with more harshly than SPP. That too is a problem because once they are in ACR's working space, only gross slider postioning will get rid of them and that really messes up the rest of the image.

I persist with ACR/PSE for my Sigma SD10 shots only because SPP doesn't have a crop function and I do like to work in one application for my watch web pics rather than hop around between applications. But perhaps SPP is better for my purposes and just follow up with PSE, or maybe IrfanView, for cropping and finishing touches. Like this, with Irfanview, good enough for Govermint work, eh?



Thanks for your comment,

Ted
« Last Edit: July 14, 2012, 12:37:15 PM by xpatUSA » Logged

best regards,

Ted
Peter_DL
Sr. Member
****
Offline Offline

Posts: 421


« Reply #7 on: July 15, 2012, 05:26:23 AM »
ReplyReply


Not sure if this meets your point:

http://www.luminous-landscape.com/forum/index.php?topic=60711.0

Peter

--
Logged
xpatUSA
Sr. Member
****
Offline Offline

Posts: 293



WWW
« Reply #8 on: July 15, 2012, 10:26:16 AM »
ReplyReply


Yes, thank you for a most relevant link, even down your example of a yellow flower in sunlight. This past week, I have learned much in this forum about gamut clipping which appears to be quite normal when rendering in sRGB from "what the camera saw". And my version of ACR (in combination with PSE6) doesn't show gamut warnings, won't allow color space selection nor will it allow selection of rendering intent and I haven't found a gamut warning function in either ACR 5.4 or PSE6 (not that I'm really complaining, it means I have to be more diligent in processing). Adobe is highly adept at masking off functions in their ACR plugins - I'm sure that even my version would do much, much more if it were plugged in regular PS or CS . . .

There is a dilemma here, similar to selecting a satellite TV plan - do I want to buy 500 channels mostly of no interest to me, just so I can watch National Geographic? We get that same dilemma in Photography a lot, don't you think? Do I really want to buy a D700 for the sensor and never shoot movies or use Live View or use the umpteen-point auto exposure? Similarly, should one buy PSE10 thereby getting the latest ACR but then ignore PSE10's additional bloatware?

Ted
« Last Edit: July 16, 2012, 07:20:38 AM by xpatUSA » Logged

best regards,

Ted
xpatUSA
Sr. Member
****
Offline Offline

Posts: 293



WWW
« Reply #9 on: July 16, 2012, 08:17:28 AM »
ReplyReply

The saga continues . . .

ACR rendered the original X3F flower shot on-screen like this . .



Out of curiosity, I used Adobe's RAW to DNG Converter, and opened the DNG file in ACR. No blues!



The only point of interest here being the marked difference between ACR and the RAW to DNG converter for conversions of the same X3F file.

Ted
« Last Edit: July 16, 2012, 09:26:38 AM by xpatUSA » Logged

best regards,

Ted
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad