Another newbie question on scanning

[jonathan_lipkin]

Hello all --

Haven't shot film in about two decades, but just ran out and got a toyo field and have started shooting 4x5. Been working digitally for a few years, so I feel like I have a pretty good grasp of workflow, color management and so forth.

Except something's wrong.

Just got my first batch of ektar 100 back from the lab, set up my epson v700 and fired up silverfast. I scanned at 1200ppi, 48 bit, and used the correct negafix settings. When I open the files in PS, they have SFprofT (PerfectionV700).icc as the profile, and look terrible. I convert to ProPhoto and have been trying to correct, but the initial scan is so far off in terms of color balance that I simply can't get a decent image. The image has a severe green/cyan cast and is terribly blue in the shadows.

What am I doing wrong? It looks like the scanner is tagging the image with the correct icc profile - I've even tried assigning various Epson profiles but they look no better.

Thanks in advance for any advice.

-jl

[jonathan_lipkin]

Quick update: I scanned something last night using Epson Scan and the results were significantly better. Still a bit blue in the shadows, but that may be inherent to the film.

[neil poulsen]

I get substantially better results from NegaFix than I do from Epson Scan.

Try going to options and setting the internal profile as Adobe RGB 1998. That's what I get as an imbedded profile from my fairly recent version of Silverfast Ai.

[jonathan_lipkin]

Thanks, Neil. Looking at the options in Silverfast, it looks like Adobe RGB was already set as the internal profile. But isn't that a working space, not a profile? From what I understand a profile describes the characteristics of the scanner, how it translates colors of the neg into numbers. Then, when you open the file in Photoshop, it is 'tagged' with a profile for the scanner (or camera) you've used. That is, there is an icc profile which describes the characteristics of the scanner. You then convert into a working space like Adobe RGB or ProPhoto.

I've also read that it's difficult if not impossible to profile negative film to scan, as characteristics of neg change with density. So, creating a custom profile by scanning an it-8 target and running it through silverfast's profile creator would not be helpful.

Can I ask your scanning procedure? Do you do a lot of color correction in the software, or just scan at 48 bits and do the correction in Photoshop?

[jonathan_lipkin]

Just made a scan in SF. Using the 48 bit option, the color and tone correction tools disappear. The only option is to scan, which I did and got the following image (inverted, which I de-inverted in PS):



I tried to correct in PS, but it is so far off I can't get a reasonable color balance.

[neil poulsen]

Hmm. From my Silverfast options, the profile embedded with the image is the same as the Internal Profile. So, I'm wondering at what point SFprofT became the embedded profile? Did you assign it upon opening the image? What's your default workspace in Photoshop? Since Adobe RGB '98 is the Silverfast Internal Profile, I suspect your problem began when SFprofT became the assigned profile. When I scan using SilverFast, the images opens in Photoshop with the Internal Profile embedded with the image. I checked this out prior to writing this post.

In and of itself, NegaFix is not a profile type operation. Although, I can well imagine that it's designed to provide correct color for the given Internal Profile. (e.g. Adobe RGB '98.) It's based on information that LaserScan obtains from film manufacturers. (I spoke with LaserSoft about this some time ago.) For example, one "problem" with assigning scanner profiles to scanned images is that conversion clips the image's color gamut to that of the target used to create the profile. For scanned positive transparencies and reflective images, it's worth this trade-off, given the improved accuracy with which color is represented in the scanned image. I don't believe that NegaFix clips the image in this way, other than to the gamut of the workspace itself.

A given workspace is defined by a profile that bears the name of the workspace. But, the color gamat defined by a workspace profile has particular characteristics. For example, given an RGB triplet like (96, 96, 96), if the coordinates are the same number, the color will be a shade of gray. That's not necessarily the case for a scanner profile; hence the reason for converting the image to a valid workspace profile. Otherwise, color adjustments can go much awry. The workspace profile also sets the gamma for the image.

As another thought, does your version of SilverFast NegaFix have an actual setting for your film? If not, and if you're selecting something that seems close, this could also be the problem. Or, the NagaFix option might not work well for this particular film?

[jonathan_lipkin]

Interesting. The images open in PS with the SF profile, I then manually convert to Adobe RGB 1998 or ProPhoto. Looks like the default working had been set to sRGB, but when I changed to PP, got the same result (opened with SF tagged). I've tried assigning a few different profiles (various SF, Prophoto, Adobe RGB, Epsons, etc) but the color cast changes very little. I've posted this at the Silverfast forums and will see what they say.

As I understand it, the image is scanned into the SF color space, then needs to be converted to a working space. Looks like your workflow somehow automatically converts to Adobe RGB.

>>I suspect your problem began when SFprofT became the assigned profile

I also suspect there's some sort of profile issue. But wouldn't assigning it a different profile (the correct) fix this? IE if it is in color space x and PS thinks it's in color space y, it will display with a strong color cast. But I should be able to correct that by converting it into the correct color space, no?

>>A given workspace is defined by a profile that bears the name of the workspace.
But, the color gamat defined by a workspace profile has particular characteristics. For example, given an RGB triplet like (96, 96, 96), if the coordinates are the same number, the color will be a shade of gray. That's not necessarily the case for a scanner profile; hence the reason for converting the image to a valid workspace profile. Otherwise, color adjustments can go much awry. The workspace profile also sets the gamma for the image.

Yes, that makes sense. For a while I had been trying to color correct images in the SF color space, and things looked _really_ weird...

[jonathan_lipkin]

DOH!!!!!

I had the software set to 48bit HDR, which is meant for their HDR software. Scanning into 16 bit works fine. The regular 48 bit option is grayed, probably because this is a limited edition version of the software.

Thanks for all the help.