Shadow Boosting ICC Profiles
I ordered some Wolf Faust (link) profile targets for Velvia, Provia and Astia last week and got a chance to play with them over the weekend. The reason for ordering the targets was primarily to ascertain if there was a problem with me scanning Astia film (as it had been suggested I needed a correct target otherwise there would be potential colour casts) but to begin with I wanted to see what improvements I might get from using the Velvia and Provia targets. First of all you should know that I use EZColor for profile generation and Silverfast for scanning. The process for making a scan is
-
Set the scanner so that it is doing no adjustments and is applying no colour profiles.
-
Scan the target profile and save as a TIFF file (I’m sure you can use other formats but TIFF seems as close to RAW as I could manage).
-
Start EZColor and read in the TIFF and the data file which will create your icc profile
Now following this procedure, I scanned a picture that I knew was challenging and compared it with both my original scan from an EZColor IT8 target and with the original transparency. The colours were good, very good. However the shadows were very blocked up. I tried for some time to find a profile editor so I could tweak the curves but to no avail.
However, I then had a seemingly obvious idea (aren’t they all obvious by the time you’ve had them). The reason that the shadows were blocking up is that the IT8 target I had scanned created a profile that mapped rgb 10,10,10 to pure black. All I needed to do was to adjust my scanned IT8 to tell the scanner that only black should be mapped to black. I did this by overwriting the black patches on the TIFF I had scanned in with pure black. I tested this and lo and behold (and other cliches) I had a scan with great colours and full depth in the blacks. The blacks were noisy but all of the data was in there. Because of this I have the choose of clipping then again (as in the original icc) rolling them off to black with curves or using shadow highlight to recover some shadows before clipping. Overall they’re the best scans I’ve had out of the v750.
I did have one slight hiccup in that nice story. The first scans I got out of the velvia scan was of a bright saturated sunset and I original chose ‘relative colorimetric’ from the intent drop down in the scanner when making my transparency scan. The result was nice colours apart from the area around the sun which had turned into a great big fried egg. The area around the sun had ended up almost a pure yellowy orange. This puzzled me for a bit until I read about what the different intents actually do. To understand what they are for, we first need to understand a bit about ‘gamuts’. Basically, when you have an icc profile, it tells the computer what the maximum bounds of a colour are. For instance, the red channel will say “when we have a 255 in the red channel, the red is this colour” However, if you’ve got an original scan that has colours outside of the gamut, you need to have a strategy to know what to do with them. Perceptual intent says to compress all of the colours close the gamut edge so that their relationships are mostly preserved but colours are changed. The other is to make all colours accurate but clip the colours that exceed the gamut. Perceptual is the first (munging) and relative colorimetric is the second (clipping). So my solution was to just change to perceptual and ‘lo and behold’!
Finally create a profile for my provia pictures in the same way. My ‘lo and beholding’ ration was well and truly out though. The shadows had a significant magenta/brown cast. I played around with this for sometime but in the end gave the Velvia profile a go. The results surprised me in their veracity and colour accuracy in the shadows. So .. everything simple then. Just use my boosted icc profile for everything!
I’ve attached a link to it here and I’d be interested in how you find it (if you have an epson v750 that is)..
2 Responses to “Shadow Boosting ICC Profiles”