SX130IS and UFRaw, green saturation issue...

Herra Tohtori

20

SX130IS and UFRaw, green saturation issue...

« on: 05 / June / 2011, 04:50:43 »

Advertisements

I recently got a PS SX130IS as a birthday present and of course jumped at the chance of installing CHDK on the memory card.

I'm very interested in using the raw image mode because its potential benefits compared to JPEG. I've fooled around a while in UFRaw interface, taken some shots in RAW and basically I'm trying to get the colour space to match the one in JPEG's produced by the camera, because they do have rather good colour quality, but I think I've hit a stonewall of sorts here.

I found pretty good settings otherwise - I'm using VNG four colour interpolation, denoise as required depending on ISO levels used for the shot, and manual WB by selecting a neutral grey/white (but not full bright) area from the image, which seems to be yielding good results. As colour profiles go, I've experimented with both standard sRGB and AdobeRGB1998 as output/display profiles, and because I don't have either colour matrix or ICC profile for the camera (don't think one exists yet anyway) I've been using "No Profile" option for the input colour profile.

So, what I'm doing is basically opening the JPEG on background, then tinker with exposure, gamma, saturation and white balance (by selecting different areas) to either match the colours in JPEG as closely as possibly, or otherwise get satisfactory results.

However, this approach seems to be somewhat limited by UFRaw's interface, since I can only adjust saturation value globally instead of for specific colours. This results in following issue:

(here's the camera's original JPEG output)

Like captioned text in image says - the global saturation value boosts the bluish shaded road as well as the green.

If I use Saturation 1.0 value, then the green is bland and unsaturated. My empirical comparisions using Eyeball Mk.1 calibration sensor system tells me that when saturation is set to 3.0, greens (vegetation) appear similar to how they look in JPEG, but rest of the image looks oversaturated, while if I set saturation to 1.0, most of the image looks fine except the greens.

Now, this is not necessarily a problem as such. Some would say the colours in the manually adjusted RAW file look more vivid; your mileage may vary. However, I would prefer it if I were able to match the colours with the JPEG as closely as possible in UFRaw (or other RAW handling program; UFRaw is just convenient because of its interfacing with GIMP). Of course I can manually adjust the saturation of different colours in GIMP after the initial RAW handling, but I'd much prefer to do it with UFRaw because GIMP as such only supports 8 bits per channel colour depth, and that results in cumulative quality loss per each colour operation, as you all probably know.

So... is there a RAW handling program that includes separate saturation values for different colours, or do I have to do this in GIMP after outputting the image from UFRaw?

EDIT:

Overall, though, I'm fairly happy with the quality gains using raw images. Here are a few sample details of the JPEG and raw results of same photo:

Ignoring the slight blurriness on the RAW shots (that's just a matter of finding good unsharp mask parametres), the image quality is noticeably better than on JPEGs. At least in my opinion. Still not sure how to deal with the colour space.

« Last Edit: 05 / June / 2011, 07:48:43 by Herra Tohtori »

Logged

vit40

276

Re: SX130IS and UFRaw, green saturation issue...

« Reply #1 on: 08 / June / 2011, 08:31:59 »

Check the thread: CHDK dng profile generator project

Logged

Herra Tohtori

20

Re: SX130IS and UFRaw, green saturation issue...

« Reply #2 on: 08 / June / 2011, 09:16:22 »

I did check it; however, I don't really see how it is supposed to be working.

If I take a picture of my display and send it, what is it compared to? The original colour target? If so, there is way too much margin of error from display calibration and colour space model to make it a truly viable method of creating an accurate colour matrix for the camera.

For example, my primary display is an LG Flatron IPS231, using AdobeRGB1998 ICC profile and set to custom colour temperature of Red 50, Green 50 and Blue 25.

My other display is a Samsung SyncMaster 226BW and despite my best attempts at getting it calibrated properly, the colour board reproduction has differences that can be easily discerned visually on a side-by-side comparison.

If, on the other hand, the profile is calculated from comparison between the RAW data and the camera's JPEG output, then of course the colour space conversion from RAW would produce colours nearly identical to that in JPEGs - which sort of nullifies one of the advantages of using raw images in the first place, namely better quality of colours.

In fact, I'm interested in ability to convert RAW in a way that produces close colour match to JPEG only as an academic exercise. I am not at all certain that the colours I have now are "worse" than on the JPEG's.

Also, as I posted earlier in SX130IS thread, I tried the DNG mode, but the bad pixel data generated by the camera seemed suspicious to me, as it stated the CCD had 2999 dead pixels (a fairly high and suspiciously non-random number). I'm not certain if I want to use this mode if there's some bug somewhere in the bad pixel reduction system...

I can post my images of the colour target displayed on my screens, but I couldn't really find an explanation on what is achieved by this from the DNG profile thread.

Logged

vit40

276

Re: SX130IS and UFRaw, green saturation issue...

« Reply #3 on: 08 / June / 2011, 10:04:50 »

Logged

Herra Tohtori

20

Re: SX130IS and UFRaw, green saturation issue...

« Reply #4 on: 08 / June / 2011, 10:44:50 »

Hrmn. Well, I just wondered about that since the thread asks for the JPEG from the camera, and that was kind of strange since if you only have JPEG picture of the colour target rendered by some random display, it's not going to exactly look the same on every display, you know what I mean?

So, I'm uploading both RAW and JPEG shots of the colour target on my display, we'll see what happens with it. Looking at the images, the JPEG has slight differences to how the image looks on the display, but that might be partially due to the -2/3 exposure setting, I'm not certain.

I got to say, though, that using UFRaw to set the saturation of the images works quite well, so in a pinch I can always use that...

Logged

vit40

276

Re: SX130IS and UFRaw, green saturation issue...

« Reply #5 on: 08 / June / 2011, 11:33:00 »

Monitor doesn't play any role in my procedure. It's well tested and it looks like Adobe is using the same aproach, because their "v3" profiles are almost as good as mine (you can find my comparison of Adobe v1,v2,v3 and my profile for a Nikon DSLR camera on ACR forum).

Link to SX130 profile is at the last page of CHDK profile generator thread. However, it's usable only with ACR or Lightroom, as other programs don't support dng profles (unles something changed last months)

Logged

Herra Tohtori

20

Re: SX130IS and UFRaw, green saturation issue...

« Reply #6 on: 08 / June / 2011, 12:21:47 »

Hm, now I feel a bit silly, as I was looking at the stickied New project: camera color profile calculation thread instead of your CHDK dng profile generator project, so I posted my image files there. Oh well.

I found and downloaded the SX130IS profile from the correct thread, now I just need to figure out how to use it...

Logged

vit40

276

Re: SX130IS and UFRaw, green saturation issue...

« Reply #7 on: 08 / June / 2011, 13:16:52 »

Don't worry ...

I suggest downloading a trial version of Lightroom. Copy the profile to the folder with other dng profiles (it depends on operating system), open the dng-s and select this profile on a calibration pane in the program. Then decide what to do next ...

« Last Edit: 08 / June / 2011, 13:25:56 by vit40 »

Logged