I am anxious to see what improvements are made in S2 conversions. Too bad I will have to wait until this evening to find out.
I've been testing the S2 this week. I also had shot with the camera briefly several months ago. There are definitely positive improvements in file output via the latest version of Lightroom. I'm very much liking the newer files.
Can you post some examples showing the improvements?
Jack, as I still have the camera for a few days, just let me know what you'd like me to post and I'll try to help out.
Others are much better at analyzing and quantifying results than I am. In deference to Michael Jackson "I'm a shooter not a tester."
The colors I was getting in my earlier go-around just seemed a little out of whack ... they are better this time around. How much? I don't know ... but the files are starting to excite me.
As an fyi ... http://labs.adobe.com/wiki/index.php/Lightroom_3.2
The new version adds an Adobe Standard profile for the S2, with tweaks for color and reduced color noise at ISO 640 and 1250.
Would be great to simply see one of your files with human skin in it processed with the old LR version and the same image processed with the new version. Crops of the same area with some detail would be helpful too.
I don't have the older version of Lightroom on my computer. But here is a quick hand-held shot I just grabbed five minutes ago using my daughter as a guinea pig. It's processed using the latest version of the software. Sharpening is turned off on the first two samples. I've added a bit of sharpening via Lightroom in the third shot. The skin tone and color is looking very good to my eyes.
Thanks for posting these. It does appear they've tamed some of the red skin issues.
Anyone else have any opinion on the new profiles?
looking at the background of the portrait my feeling is that the white balance gives the entire image a bit greenish/yellowish touch. That, of course, makes the skin tones much nicer here...
almost neutral, you are right.
However it looks a bit greenish here (my monitor is okay).
Either way... if there is an improvement that's a nice thing for S2 users!
Guys, it's tough to judge unless you've shot a color checker under a known lightsource.
Some of the model captures (of Tori and Dennise) in Jack and Guy's S2 review were processed in Lightroom and the results were posted in the review.
Maybe the easiest way to see if there are any differences and improvements would be to reprocess a couple of those files ?
According to post # 20 in the review the S2 DNG files can be found here:
Mortar is not really neutral grey, but a slight bit greenish (at least over here). That may explain what Victor noted.
I'll convert one now, update to 3.2 then covert the same photo with the same settings and see what happens... Give me a minute...
It's a big improvement, it seems. These crops certainly show a more neutral colour, with better handling of reds and skin tones. What is evident also in the shot with the purplish woolen jersey, the new version seems to be doing better at pulling a little more detail out of of the files....
Last edited by tjv; 7th January 2012 at 00:23.
Very nice contibution; this does demonstarte the value of tweeking the profiles.
The open question is the need for a dedicated raw developer and what it would do. If the current system has a good algorithm for the Bayer Matix conversion (that is, we don't need to have a 4x4 and 8x8 etc higher order correction, that would be VERY slow) then one wonders whether profiles will close the gap.
Guy has suggested that the processor does more than color balance, and that a good raw processor also can take out anomolies of the sensor (aliasing, artifacts etc.)
What strikes me here is that the profile allso contributes to detail, by moving the color space to a point where neighboring pixels show more detail (that may have been "Bayered out")
an ultimate raw proc would treat the four inputs almost like a color space hologram, and have an extremely detailed algorithm to reconstruct. THEN it would need to filter the know artifacts of the conversions. Looks like LR has a nice compromise, but not one that digs out wha'sin the S2 files.
I must say that this could be a nice niche business; I would immediately buy a raw processor in the uder $100 range, even for casual use; ONE good print would pay for it.
So, are ther forum members who are adept?
Victor, maybe this triumvirate would be able to do it:
Has anyone tried Raw Photo Processor (RPP) on the S2 files ?
- - - - -
@ Jack: I don't know if the latest ACR build has the same profile as LR (not using Adobe's programs).
Btw. I didn't mean to say that you should do the work with reprocessing, you've already done a huge effort making the review together with Guy.
I was just thinking of all those already using LR as their preferred processor.
tjv, sorry but when I copy your posted samples to my pc and quickly toggle between them I'm not able to see any differences whatsoever.
Could it be that you've maybe mixed up some of the processed versions from 'before' and 'after' the LR upgrade, so that both posted version pairs are 'before samples' (or both are 'after samples') ?
They also have exactly the same crop and exactly the same resulting file size which might indicate they are actually identical files ?
The photos seem to have the correct names, perhaps they have lost something in the upload process? When I open them in Photoshop I see a difference in rendering, most noticeably in the skin tones / reds. I used 3.0, made three variants, one full frame and two crops, then exported. I then installed 3.2, imported my 3.0 catalogue and reexported, making sure the files had the same settings. I used compression 12 and sRGB with no pixel resizing. Perhaps I'm missing something?
I'll try again...
I'm totally lost now, but here are the same crops reconverted. Man, maybe my eyes are just playing tricks on me and seeing what they want to see
I W/B'd off the same point. The files attached here are the 3.2 versions, so compare them to the above ones.
Last edited by tjv; 7th January 2012 at 00:23.
ACR 5.7 (in CS4) vs. ACR 6.2 RC (in CS5).
Same WB settings on the image in both softwares (5500K / +36).
Color Fringing correction off.
Processed to 16bit TIF (ProPhoto) and assembled on layers.
Upper layer set to layer mode "difference".
No (zero!) difference with Color NR set to zero (attachement 1).
Little difference with Color NR set to Camera RAW defaults (attachment 2).
IOW: there are no tweaked color profiles (at least not comparing the ACR versions... which should be the same in LR).
Just a different handling of color noise (ACR5.7 in CS4 doesn't feature the "details" tool in Color NR).
I wanted to wait until I got to my color-managed desktop before commenting, as I personally saw zero difference on my laptop over the weekend. Result? I'm still seeing zero difference in these images. Reds are still over-saturated and too purple, yellows are still over saturated, and so skin still looks red-orange. I do however see a slight difference in detail, but while the sweater fabric looks slightly better in the 3.2 file, the skin quality looks the same amount worse, so I'd suspect a difference in base sharpening settings. At least this is what it appears to my eyes -- I certainly respect other's may see things differently...
All the three are exactly the same. If you see slight differences comparing the images side by side then the differences come from the different viewing angle on the monitor.
If you put the files on layers and check them in layer mode "difference" they match exactly.
You can also load them in a color grapher (e.g. Color Think)...
attachment 1: only one of the three images is displayed
attachment 2: all the three images are displayed (and the graph does not change at all).
I didn't bother to overlay them in CS, so I'll take your word for it -- as I said, what I was seeing was slight. Plus, given no differences in color, it seems unlikely they would have made any sharpening changes either.
If I were an S2 shooter, I'd build my own C1 capture profile set...
Possibly the difference is leveld out in these sRGB JPGs here as the changes most likely just affect high saturated, dark tonal values that are simply clipped in the sRGB JPGs (just a guess...)
agree. Or simply use RAW Developer...If I were an S2 shooter, I'd build my own C1 capture profile set...