And now, a photography blog?

I've finally decided to start writing something here, for a few reasons. I used to write my own web page, with lots of stuff including photos of rock art. Then I started posting photos on Flickr because (a) it was so much easier, and (b) it made it easy to link one photo (like a closeup) to another. For instance:



What I wasn't expecting was how much of a benefit the Flickr community was -- my web page always brought in a lot of email, but the more immediate feedback aspect of Flicker made it much more interactive. The web page felt much more one way, Flickr was more like a conversation. So that's the first reason for writing things down in blog form -- if I have something to write down, it would be nice to have some feedback.

And following from that, the second and more immediate reason is that I now have some stuff I want to write down, mostly for my own benefit. I've been learning a lot over the past two years about the whole digital darkroom end of things. Well, maybe learning isn't the right word, since it might indicate that I know what I'm doing . . . but my workflow has evolved a lot. I keep trying different things. And in the last week I've realized that it hasn't always been for the better, that apparently I've completely forgotten some of my old tricks which work better than my new ones. I used to process my images pretty much exclusively for the web, without a great deal of care, because I thought if I ever actually wanted to print anything I should go back to the original image and work a bit harder at it. That's probably true, but I'm now at the point where I want to try to make some prints, both for myself and for other people who have asked, and in two of the three first images I went after I found I was unable to reproduce aspects of my earlier work. In one case I think I've figured it out, in the other I'm still only left with guesses. So I better start writing this stuff down for myself. And I'll do it here so that you can tell me what I'm doing wrong.

I'll start with what I learned today: if I'm scanning slide film of rock art (ocher pictographs at least), when it comes to color, VueScan does a much better job than Nikon Scan.

I started reworking this photo yesterday, here's the version I posted on Flickr long ago:



I scanned that with my old scanner, a Nikon Coolscan IV. With that scanner there was a trade off between dynamic range and resolution -- I could do a single pass scan and see resolution limited only by film grain in many films, but also a lot of noise in the shadows. Or I could do a several pass scan to bump up the signal in the darker areas. But that scanner didn't support multiple pass scans, I used VueScan to do that. The scan head wouldn't always be perfectly realigned for each pass though, so multiple passes would soften the image a bit. On average, the more passes the softer the image but the more dynamic range as well, something in the 5-9 pass range seemed best. For posting web and even wallpaper sized images that was fine, at that resolution all I lost was grain noise. (My new scanner is actually a lot more work because of that.)

I also used VueScan for a second reason, one I had long forgotten due to the multiple pass thing that I need VueScan for anyway. Whatever version of NikonScan came with the IV in general did a pretty poor job with color and exposure on slide film. It was great with prints, but not slides. That's what drove me to get VueScan in the first place, since it gives you a lot more control over the color profiles that are used. But I'd forgotten that until today.

Now I have a Nikon Super Coolscan 5000 ED. (Seems like they need an additional "ultra" or "mega" in that name somewhere. "Plus" at least. "MegaPlus.") It supports up to 16x oversampling in a single pass, so I no longer have to choose between resolution and dynamic range -- with Provia film I get some 19 megapixel scans that look pretty ok at 100%. It also seems to have a better implementation of IR dust and scratch removal than the IV or VueScan, and unlike whatever version of Nikon Scan I started out with on the IV, it seemed to do an ok job with slides. So I'd taken to using the Nikon Software again.

Back to the image -- I need to reprocess it, and my old scan from the IV was pretty soft, so I decided to rescan it. One slight hitch I should mention here -- I recently upgraded my G5 mac to OS X 10.5 Leopard, and NikonScan is currently pretty buggy on Leopard. I never successfully scanned a slide at all until I reinstalled it, not it finishes but at can take 20 minutes or more for a single scan.

Anyway, I scanned the slide with a few different settings for various levels of grain reduction, compared the results (for this Velvia 100 slide, GEM=1 was a good compromise) and went off to working on the image. In fact I worked on it for a couple hours, and when I was close to happy with the result I compared it to my old image . . . ugh. While the resolution and contrast looked great, and overall the image had a lot more pop to it, the colors just really sucked in comparison. Enough that I preferred the old image -- it has wonderful reds that were just faded bruise purples in my new one. I was up until after midnight trying to figure out what I had done to the original, but no matter what I did, the reds were just not there in the new versions. So I saved what I had and went to bed.

This morning I wondered if some of the auto color routines in PhotoShop had changed from CS2 to CS3 (another recent upgrade), but a quick check showed that the results were indistinguishable. Then half way through the day today (probably in an attempt to avoid cleaning up the house -- Amy has been gone for 2 weeks) it occurred to me that maybe the difference was in the scan itself, either the hardware (I hoped not) or the software. I brought up the original scan from the IV, and in no time at all (because I used to spend no time at all, compared to the time I spent now) I had the colors. Step 2, rescan the slide with the new scanner but using VueScan instead of Nikon Scan. And there is was -- same wonderful reds, I could see it in the raw scan without any processing, and it only took about two steps of processing to fully reproduce the original result in terms of color. Mystery solved, lesson re-learned.

I guess this means I should also write down all my VueScan settings, but I really do need to clean up the house some. For now I'll just say that for media I usually use "Image" instead of slide film and "none" for color balance. With a good dynamic range multiple pass scan, that usually gives me something with reasonable but unsaturated colors and thus a lot to work with Photo Shop.

Now I just need another free afternoon to rework that image . . . because I really want to get a print of that.