Ten RAW Images

High speed ftw! A quick $12.99 sale price later, and the LX2 is somewhere between 25-30% faster on RAW capture. Thank you OCZ and thanks Gary for the hint!

To my surprise, shooting time for ten "high quality" JPEG images remained about the same -- around 20 seconds -- indicating to me that for compressed pix, the limiting factor is the speed of the "Venus Engine" processor itself.

Addedum: yesterday saw the publication of dpreview's test of the Sigma DP-1 which is probably the closest competitor to the LX-2 and considered by many, in anticipation of it, as the compact streetphoto heir apparent. To my great surprise, the Sigma is significantly slower than the LX-2 -- the only got a little over 7 seconds per frame in RAW while the LX-2 turnaround was only 4.5 seconds (even faster than what I was getting). The Sigma does have a three-frame buffer, though it's not clear to me if you can use it in regular shooting mode. And a hotshoe for Martin Parr & Bruce Gilden fans.

Addendum #2: And on the heels of that, another review lamenting the DP-1's lack of speed, this time by Edward Taylor on The Online Photographer -- "My main complaint about the camera then was that it was painfully slow. It was and it is...." Hmmm.

LX2 Part 2

As I expected, I've gotten more used to the LX2.

In the clichéd and time-honored tradition of pointing Leicas at brick walls to prove that their lenses are top-notch, here's a closer sample of an in-camera-sepia JPEG. The right-hand area shows a detail from the picture on the left -- pixels at one-to-one size (if anything, the image here is degraded just because it's a web-compressed pic. It was also hand-held).

As long as I'm willing to put my thumb on the monitor, it's fine in the hand. During the past week I've been shooting with it at the ION Conference, using it as a notepad to keep track of presentation slides. In the hand for an hour at a time and I've gotten used to the idea. No hand strain. In JPEG, it's also plenty fast.

Tomorrow I'm taking Gary's advice and trying a much faster SD card for shooting in RAW mode. If it can get the differentials indicated on Rob Galbraith's benchmark site, there might be as much as a 4x acceleration, which will keep me quite happy (even a modest improvement might be enough).

LX2

After returning from China I gave myself a few weeks to see if Panasonic would announce a new LX3 at February's camera-business trade show. No dice, so I promptly ordered a new LX2 to replace the stolen LX1. Here are a few notes, comparing the two.

Only the LX1 camera, one card & one battery were pickpocketed -- my case & charger, 2nd battery & backup card were still in my luggage. Not surprisingly, everything fits perfectly, equipment-wise. Perhaps with time I'll also learn to adjust as smoothly.

The lens is the same killer Leica 28-105-ish. I have been impressed with the improved color renditions, and I don't mind the slightly different character of the in-camera JPEGs (I also don't buy the notion, proposed by some bloggers and others on photo.net, that the LX2's "Venus III" chip does pre-processing on its RAW data -- an assertion that makes no sense to me).

The higher ISOs (two extra stops, from 400 to 1600 as the top end) are a very welcome addition, regardless of how noisy the highest ones may be. A sharp grainy photo is better than an unreadable blur or no photo at all.

I might yet tape the back of the camera for grip as I did the LX1 -- but not the front. The LX2's front finger grip is much-improved in providing your middle finger purchase. Simple, and well-done -- but it needs to counter a downside change, hands-wise: to accommodate the larger LCD screen, the LX2's designers have cramped the rear-side controls. It's harder to shift your (my) thumb around and hit the correct thing in that smaller space.

If I want to hold the camera firmly, one-handed, I end up with my thumb on the screen (and away from the buttons). I am pretty sure that I only rarely used the LX1 one-handed, so this difference may feel like a bigger deal than it is.

Overall, the narrower thumb space means that my hand is significantly less comfortable when holding the camera "at rest," and that in turn creates the (false) impression that the LX2 is heavier.

The controls themselves have been subtly improved, mainly by Panasonic's UI designers realizing that people confuse the joystick with the directional buttons -- so, whenever sensible, you can use either control to accomplish the same task. This is good usability design, thinking about what the user wants to accomplish and letting them get there in the way most natural for them. A+

The biggest impression of difference between the LX1 and LX2, however, has been speed of operation -- or rather, the lack of it. The LX2 is slower between shots. I would have guessed about 25% slower, which matches change in the pixel count. dpreview's test shows a 32% slowdown on RAW writes, as well as a slowdown in burst mode. Those percentages feel like a lot, given the rhythm I'd already developed with the LX1. This is frustrating given that the camera's operations are otherwise responsive and the AF is even a tiny bit quicker. That extra second and a half of write time can be tough.

Token LX1 sighting: I saw Sylvia Plachy using one on Ovation's televison bio Close Up. She was switching back and forth during the same session between an LX1, a Leica M, a Rolleiflex, and a Hasselblad 500 of some sort (SWC?).

Long Ride

Almost time to say goodbye to China, now that I'm back in Beijing. Also time to say goodbye to:

• My Canon 50mm f/1.4, which spontaneously has decided it doesn't want to focus on anything closer than 5 feet, even in manual (at least until I can send it off to Canon)
• My Panasonic LX1, which was pick-pocketed in Xi'an (along with most of my pix of the Xi'an city walls and street musicians)
• My HTC Excalibur smartphone, with all my notes and action items from the entire trip (pick-pocketed just tonight inside the Hyatt)
• My faith in Cisco VPN (pretty-well worthless on this trip)

So that's maybe $1000 in theft losses and a major dose of work frustration. Net balance for the trip then? Love it. Hassles and troubles like those just transform a vacation into an adventure, and the experiences I've had will last a lifetime, unlike any sort of electronic gizmo (and I have paper backups of my notes).

If I can just keep my laptop and 5D working for two more days....

(Followup: I remind myself, a bit, of my old second (third) cousin who raced motorcycles and cars and kept soldiering on through the many hospitalizations as just part of the passion....)

---

Which doesn't begin to compare to what happened to Michael :(

Soundslides

Click to play

I've finally started to mess around with SoundSlides -- I've had a link set to it for some time, ever since seeing a quick guide on Martin Fuchs blog almost a year ago (ouch).

The link above is my first crude attempt with the demo, a little recap of Isaac's summer hockey season -- after ten minutes of using the program I promptly sent Joe Weiss a nice PayPal delivery (and ordered myself a new audio recorder (sorry Griffin, my iTalk is cute and tiny but not a general-purpose tool) and some replacements for my main mics, which have seen better days long ago)(more stuff to carry when shooting -- hoorah). I'll replace the prezo some time soon when I get my proper SoundSlides reg code and also re-record the audio with Isaac playing it (or something similar).

Obviously I've got an learning curve ahead of me but I'm excited -- SoundSlides is just far easier and more direct to use for these sort of presentations than anything I've see so far using regular Flash, or Flex, or even Premiere. All good programs, but SoundSlides is directed -- it does one thing and does it well. No wonder it's so popular for news shooters and wedding folks.

Hi-Fi Lo-Fi

It's common to tell digital photographers: "don't trust the camera LCD as a preview."

Why the heck not? A lot of the time, I happen to like the picture I see on the LCD. So I made myself an Adobe Camera RAW preset that, as best as I could eyeball, would match the tonal range of the LCD on the LX1.

It was a somewhat subjective process, not entirely perfectly scientific, but simple enough. I shot some Kodak grayscale charts, played them back on the camera LCD while simultaneously loading them in Adode Camera Raw, adjusting the corresponding RAW/DNG conversion on my laptop under Photoshop CS3. I could see where the blacks petered-out, and the overall relationships in tones between neighboring patches. So patch 1 was full-on, the grays died out arounf patch 14, the values were a little boosted around patch 5, etc. It made the picture that I liked.

Once I'd made such a preset fro RAW files, I also made a corresponding adjustment curve that would alter camera JPGs to also more-or-less match the results I was getting from ACR. It's easy to make such a curve with a three-layer photoshop file (I like RAW but some situations — particularly very fast repeat shooting — still require JPEG for this little bufferless compact camera).

To make a curve that matches a JPEG to the ACR result: First, open the JPEG. Next, add a Curves layer and close the Curves dialog (we'll come back to it). Now, open the RAW file in another window, Select-all, and paste it on top of the JPEG (which will make a new layer). Set the blend mode of this new layer to "Difference."

Now all you need to do is open that curves layer again and adjust it until the visible differences between mictures are the absolute minimum. If the picture is black, then both the bottom (JPEG) and top (RAW) layers are a match.

The less-than-wonderful surprise I got was: the pictures don't align. At first I thought it was sharpening, but actually they just don't line up. They are two or three pixels misaligned, apparently at a 45-degree angle. In fact it's not even an integer number of pixels — the pic above (a 100% blowup of the previous blog entry) shows the closes I could get, and shifting it in the opposite direction simply moves the various contour-outlines from one side of the face to the other.

The second surprise was that, despite the fact that these curves reduce the tonal range (that is, they step on constrast), the RAW pic holds detail quite a bit better than the JPEG. I'd expected that since the JPEG had more range than my desired pic, I wouldn't make much difference. But it does. The higher fidelity of RAW still matters even on a low-fidelity images.

---

As a minor aside, we noticed last night that the LX2 makes a guest appearance in Spiderman 3 — in a scene where a photographer loses his SLR, he wastes no time in dragging an LX2 out of his jacket pocket & just keeps on shooting.... (though I'd never recommend carrying the camera in your pocket with the lens and flash both already extended).

LX1: Ongoing Reports

Whaddaya know, the new issue of Consumer Reports has arrived and what do they pick as their favorite compact camera? The LX2, the latest variant of the LX1 (bigger LCD, higher res, higher ISO's, but the same lens, UI, and camera frame).

I couldn't help but get a smile out of their pic, though: what's she looking through?

The weird part is, I was just getting ready to blog about that topic anyway: not Consumer Reports, but viewing with the LX1 (or LX2, or Leica D-Lux2, or D-Lux3).

At this point I pretty-much know what the field of view is (especially given how much hipshooting I've done), just looking around. I can "see" the picture in front of me without any camera, so I've found it's comfortable enough to just pretend there's a viewfinder. No glued-on dots or minifinder, just hold the camera up near where my eye is (without covering the eye, as CR has bizarrely done), and go ahead & shoot. I can see the top of the lens to ensure me it's pointing straight, and a bit of the LCD glow gives me some vague notion about the luminance (which I usually ignore — can't turn it off, alas -- gaffer tape and a bit of black paper to the rescue?).

And then yesterday, I finally got to reading Sean Reid's Review of the D-Lux 3 and he's commenting on the lack of detail in any LCD screen. Second-Opinion co-reviewer Mitch Alland has his own VF/LCD take, where he say his use of the LCD is mostly about gross-scale framing, and that (somehow) he too uses his un-VF'd eye a fair bit (though not, I think, the way I'm showing here).

Nub is in the Details

This is a little game development (or general graphics) tip that I've been thinking about for the past couple of nights, with additional applications for photographic and other images.

It's really a simple observation, followed by some implementation tricks. The dumb observation is that people like noisy pictures. This has been well known, of course, in famous older papers like Rob Cook's 1985/86 paper Stochastic Sampling in Computer Graphics (PDF). And for many years photographers have been keen on using grain as a means to elicit a sense of sharpness that may be actually greater than what's really in the picture. This isn't really news, but in playing with noise I've found a really simple trick or two that have pretty broad uses.

First, we're going to make a texture.

• Start up Photoshop, and create a new image that's twice as wide as it is high — preferably both values will be powers of two. In this example, I'll use 512 by 256, RGB.

• Set the foreground color to middle gray (128/128/128 or #808080), select-all, then shift-F5 to get the fill dialog and fill with the Foreground color at 100%. You image is now gray.

• Select only the red channel by selecting it in the channels palette or hitting ctrl-1 (or cmd-1 on a Mac). You should see a plain gray rectangle and the select-all marching dots are still active.

• Select from the Menu bar: Filter -> Noise -> Add Noise.... and set it to 15% and "Gaussian." (Note that if you squint so the at the image is out-of-focus, and click the "Preview" checkbox on and off, the overall gray value doesn't change).

• Now select the Green channel.

• Apply noise in exactly the same way as before.

• Jump back to RGB display, and you should see something pretty similar to this:

• We're going to save this as a DDS texture. You'll need the NVIDIA Texture Tools DDS plugin if you haven't got it already (if you are a game artist, you almost surely do).

• When the dialog appears, select the "Use existing MIP Maps" option and save.

• Close the window.

• Open the resultant DDS file and click "Load MIP Maps" to see all the data.

What happened? The DDS exporter assumed the right half of the original image was filled with an image pyramid — the data in the black area was thrown away, since it's not used. We now have a MIP data set where all layers of the MIP stack have noise of roughly the same statistical characteristics for each and every level (rather than each level being a scaled-down vesion of the layer above it).

What this means is that if this texture is applied to a 3D model, the on-screen frequency of the noise will tend to be constant, regardless of scaling. The texture mapping hardware will always select a level appropriate to the current pixel's screen size, and that level will always have noise of about the frequency we see here on the screen.

Let's try it out on everyone's favorite teapot:

Whether near or far, you can see that the texture feature size — that is, the on-screen size of the noisy "blobs" — is about the same regardless of whether near or far.

I call this approach "constant frequency noise."

If you're experienced in game shading, you may already suspect why we added noise to the red and green channels, but left the blue channel alone.

Applying this as a tangent space normal maps, you can see that the nubbly detail maintains its on-scren size regardless of whether the pot is near or far. This is different from typical MIP map behavior (which would degenerate towards a smooth surface — see the example below), and while not "physically correct," gives a better consistent feeling to the nature of the rough but shiny material, regardless of rotation, scaling, or distance.

As long as we've got this noise texture, let's try something else. Here are two teapots, one with a standard color texture and one with our noise-mips added to the texture UV values:

In areas where textures are greatly over-extended, this method can be used as a variation of "detail texturing" so that instead of soft linear-interpolated blobs, we get little bits of noise.

Of course, in this application we don't want to offset the texture too much when we're not super-close, so rather than using constant-frequency noise, we can just make a square texture, rather than a rectangular one, and let the DDS exporter "Generate MIP maps" insteead — the lower maps will smooth to a center, zero value, so that at a distance there's no visible offset.

Here's an example. As you can see, it's identical at the largeest MIP but smooths quickly to middle gray at the smaller sizes. In this case, where we are using the channels as signed values, middle gray equals zero: no offset, no bump.

The noise-offset-as-faux-detail technique can be applied to still pictures, too. In Photoshop, you may be familiar with the idea of using the "add noise" filter to cover-up the defects of a picture that's out of focus or enlarged too much.

Let's take the thumbnail of this photo as a sample. If we blow up the thumbnail, it looks pretty soft (especially if using bilinear filtering, which is typical for realtime applications like games).

But if we "add noise" (here quite a lot — the same amount we used when creating the noise texture), the effect is "Grainy" rather than "Muddy," as we can see in these side-by-side views:

Alternatively, we can apply the same noise pattern offset method we used for 3D detail texturing to the 2D image.

• Make a red/green noise pattern as before, but the same image size as our soft enlarged image. Save if as a .PSD file — say, "imgOffsets.psd."

• Then go back to the soft enlargement, select "Filter->Distort->Displace..." In this example, the scale was set to 4, the wrap mode was on, and no stretching (since the images were the same size). When the file dialog appears, select our just-created "imgOffsets.psd" as the offset sourcefile.

This version creates an illusion of small detail even though there is none — and without altering the colors directly. Some of the obviously-aliased details, such as the highlight on the lower lip, really look far better this way.

Both approaches can be combined — offsets and also a smaller bit of color noise to enhance the apparent sharpness. Another option might be to make the offset image a smaller size and then stretch it in the offset filter, so that the "clumps" are larger than single pixels.

Understanding these uses of noise can extend well beyond Photoshop. An obvious application that would join together these techniques would be better, more sophisticted video scaling performed in real time on a GPU.

Here's a last combined-noise version with clumpy offsets, followed by a sharp version (the thumbnail was a scaled-down version of this) for comparison.

Styling

Both of my primary digital cameras now have the ability to save a full-sized JPEG image along with the corresponding RAW file. So I've taken to setting them both up the same way: with the JPEG stored as B&W, high contrast, and the color balance set to "AUTO" or "Daylight."

This buys me a couple of things besides the obvious one: B&W photos that are B&W out of the box. It gives me the option of later tweaking the B&W conversion from the RAW files, but also — and I've found this to be genuinely useful — a strong, contrasty image on the preview screen that reads well in bright or dark conditions — even if I'm shooting in color.

The punchy B&W preview is useful as a strong litmus test about the immediate readability of an image. Not that all pictures need a crisp graphic style, but when they do, the B&W preview shows it.

When I worked a lot on TV commercials we would keep a bad B&W TV set around for previewing. There was a $3000 13-inch color studio monitor next to it. The B&W set was placed there by our postproduction color timing expert, who always wanted to preview everything on it to be sure the results of his work were truly readable.

A few years later, I attended a talk by an art director from Pacific Data Images, talking about Shrek, and they followed a similar method: they would look at every shot in both color and B&W. "If it looks good in B&W, the color will only add to the shot." At Pixar, there were a few animators who would turn off the preview shading on their characters — the result wasn't B&W images, but while working they got the figures down to their essential outlines, so that the scenes on the film would play with inerrant clarity.

I've also found, and this is doubtless some failing on my part, that if I see the picture only in color — if I don't see it early in B&W — then I just never move it to B&W. In the world, while making photos, obviously what I'm seeing through the viewfinder is the full-color world... yet it doesn't have, for me, the same psychological effect, the pictureness that gives it permanence, until it's a picture. If it's a color picture, I just read it at that. It's easy to imagine a B&W picture from the world, but hard to imagine one from a color picture. I don't know why this is, but it seems true for me. I react to pictures as pictures. Which is probably as it should be.

There's a further technological spoon stirring all this up. Frustratingly, Adobe Camera Raw (ACR) simply refuses to honor the B&W-ness of RAW files from either the 5D or the LX-1. Canon encapsulates such data in "Picture Styles," which they expressly encrypt or otherwise mask from Photoshop's prying software. And I genuinely can't even tell if Panasonic included the color effect information in their RAW format (Besides, I habitually convert all my Panasonic RAWs to Adobe DNG format, which only consumes about half the disk space).

For a while, I took the strategy of trying to set up ACR defaults per-camera, so that everything would come through set to a high-contrast B&W. But that was counterproductive since I only some of my photos are really intended to be B&W. ACR wants to give an all-or-nothing solution. The only good way is to keep the JPEGs. Better yet, sometimes the JPEG can be the final picture anyway.

Another option, which I've been somewhat surprised at, has been to make less use of ACR and Bridge, programs that I've really liked and had hoped would be the axis of all my workflow. At least for the Canon files. Canon's Digital Photo Pro doesn't have the scriptability of Adobe Bridge or the full power of Photoshop, but it does have the simple ability to honor Canon Picture Styles, and its color algorithms closely match (should be: identical?) the intents of the original camera-sensor (and picture style) designers.

As for ACR, I still use it all the time, and I reset the defaults to keep doing what it likes to do: assign the brightness and color and everything else according to Adobe's own secret sauce knowledge. It does a good job. I know that I can always look back at the JPEG to get a glimpse of what I was really thinking when I snapped picture "IMG_6769" or "P2565811."

Right now I'm more or less happy with how this is working. It saddens and frustrates me, thinking of the ongoing difficulties revolving around all the work I went through to get to this point, and the ongoing future storage headaches. But things seem, at least for today, stable and predictable. Maybe I really will be able to keep the same cameras and workflow for a long time.

Sharp Distinctions

I'm something of a believer in half-baked photo tests. If test results aren't obvious except in highly-exacting circumstances, for equipment that's unlikely to be used in exacting circumstances, then: who needs them?

If results can be shown in ad hoc, half-baked test situations, then they're more worth examining. So here's a quick little comparison. I'm not looking at bokeh, or chromatic aberrations, or anything else. Just focus near the center.

I was concerned about some wobble in the focusing elements of my Canon 50mm ƒ/1.4 lens. I figured I should compare it to my corresponding rock-solid Contax-Zeiss 50, that I can mount to the 5D via a "Cantax" adapter. And since I was shooting at ISO 100 anyway, why not do a quick comparison against my compact LX1?

What you see above are pixel-to-pixel crops from the centers of three photos. The white-balance was set to "auto" for all, so the color shifts are not significant.

The results surprised me a bit. These shots are all made around ƒ/4, 1/250th of a second — handheld but leaning against a wall. I made several exposures & these are 'representative.'

The first surprise, to me, was how poorly the Zeiss lens did compared to the Canon. I had expected the opposite. While it's possible that my manual-focus skills aren't up to snuff when compared to the Canon AF (even with the special Canon 'S' manual-focus screen), I did check the entire frame, and found that there were indeed areas where the Zeiss focus was a bit better than here in the center, specifically in the corners. So either way, the Zeiss had less flatness, or maybe it was just less sharp at the center.

Or I can't focus. Either way, since in this case I only care about photographs that I myself will make, the net effect is the same: I get a sharper result with the Canon, at least for the 50mm. And that was a surprise (I'll have to test the 28mm lenses after the holiday).

I also found that the wobble seemed to have no effect on the effectiveness of the Canon lens (still disconcerting, and I may have to send it in. If I tilt the lens forward or back after locking focus, the elements move — and then I do shift focus. I can also feel a distinct "thunk" when the elements slide).

The third surprise was just how well the little compact PanaLeica held up. At ISO's above 100, or in terms of instant-on accessability, the big Canon dominates — but for good light, the LX1 (or LX2, which has the same Leica zoom lens) puts up a real competetive fight! Especially for a pocketable camera that costs one-fifth the Canon price (or even less: since the LX2 came out, I've seen real LX1 deals. I ran into one at Fry's last week, brand new for less than $300).

B&W Conversions

While my iTunes subscribes to it, I have to say that I'm not a huge fan of NAPP's "Photoshop TV" video podcast. I subscribe in the hope that some of the tips on the show will be useful. At the same time I dread having to wade through the hosts' gossipy and self-congratulatory prattle. It's better to watch on iTunes than the iPod, mostly because it's easier to fast-forward and skip those sections on the PC.

Another gripe: often the latest episode sometimes takes an hour to download on a broadband connection. Ugh.

This week, though, my pains were rewarded by a segment shot at the recent Photoshop World conference, featuring John Paul Caponigro and his recommendations and method for converting color images to black and white. His method was different from what I have been using and I like it a lot. If you're used to working in Photoshop adjustment layers, the pic above tells almost the whole story... with more details below.

What I've been doing (up to now, that is) is using a "channel mixer" layer with the 'monochrome' button checked to do my conversions, & quickly previewing all three channels (ctrl-1, ctrl-2, ctrl-3) before adding that layer. So does Caponigro. The tricky part for me has been getting the balance of the three R, G and B channels to look good (more effort than just hitting "grayscale" but a lot more control).

Caponigro's method depends on the "channel mixer," but he uses two layers where I had been using one — for a lot more ease, speed, and flexibility. In this expanded method, just leave the "channel mixer" at its monochrome default: that is, check the "monochrome" box and leave the sliders set to 100% red, 0% green, and 0% blue. The resulting monochrome pic with therefore contain only the red channel. Rather than tweak the channel balancxes in the mixer, insert a "Hue/Saturation" layer immediately below the channel mixer layer, and play with the input color balance by simply sliding "Hue" back and forth.

A big Duh! moment for me. This method is so much more fluid to use, lots of variations can be developed just by dragging around that single slider.

I think I'd seen an article describing his method before, but the video really made it come alive for me — and I realized it wasn't just useful for the landscapes typically used as examples.

The pic above shows a range of variations that can gained just from sliding "Hue" back and forth — some cartoonish, some pretty standard and some Just Right.

If I felt like further noodling, all the other "Hue Saturation" controls could have been used too, including tweaks across specific color ranges (raising the saturation of just the reds, for example, could lighten the darker side of the face — if I'd wanted that). For the sake of this example I've just stuck to a "straight" manipulation of only Hue.

The conversion happens independently from the two top Curve layers, which are optional. I use them (a) to give the overall picture a bit of a contrasty "S" characteristic tonal range that I like (that's the 'bwLimits' layer), and (b) to give the final B&W image some color tone by tweaking the blue channel only: boosting a wee bit in the shadows and pulling it back a wee bit in the highlights for a slight warm/cool effect (the "toneCurves" layer) (these layers would be the same regardless of the method I'd chosen for the basic B/W conversion from color).

In the pic at left you can see the "bwLimits" layer editor, which shows the "S" pretty clearly. It's not a strong "S" — but it's there just the same, and the high and low points of the histogram are deliberately brought-in a little to give a full range between pure black and pure white.

Below that we see the "toneCurves" editor, which is only applied to the blue channel to create that warm/cool effect (for a color image, as on the web. When printing B&W I currently prefer to use Roy Harrington's Quadtone RIP, which has its own warm/cool print style).

(Photoshop experts may point out that I could have done the work of both curves layers in a single layer. They're technically correct, though I prefer using two layers to do two different jobs)

In both cases, the 'curviness' of those curves applied is really very slight. The Hue/Saturation layer is where all the really bold action occurs.

This shot was just a quick grab from yesterday at the coffee shop. Iz was working with her laptop and didn't even realized that she'd set herself up with a cluster of papers on the tabletop supplying reflected fill beneath her face, which was already illuminated by a big soft-source picture window. The color image (shot quickly with the camera set to "P") doesn't really have the right feel, especially given the mixed-color nature of the light sources (the window was partially obscured by a tan sahde above her head, blue-gree anti-glare film near her face, but no film on the papers — combined with incandescent lamps from the other side), but the B&W starts to approach the luminous glow I saw when shooting. Other than the conversion, there's not a bit of dodging or other alteration required for this one.

ChartThrob

ChartThrob: A Tool for Printing Digital Negatives

A few months back at a PHIG meeting, I met Thomas Howard, and saw how he was using charts to hand-profile his process for making digital negatives for platinum-process contact printing. I figured this could be automated, so: I automated it, and made a program called ChartThrob. I let a few folks on APUG and now HybridPhoto try it out, and after a few unexpected glitches (who knew Photoshop could have fractional pixels?), it's ready for public abuse.

ChartThrob is a a JSX-format javascript for Adobe Photoshop. It runs under Photoshop CS2 and later versions, but won't run under Photoshop 7 or Photoshop CS (Sorry, that's the price of progress). It works for both Windows and Mac versions of Photoshop. It's free.

ChartThrob creates profiles for your process, for your printer, and lets you create consistently-beautiful digital negatives from your pictures — every single time.

Current Version

ChartThrob is currently at version 1.07 (14 Nov 2006). There have been no functional changes that would affect chart output or analysis since version 1.03 — only UI refinements and some localization tweaks. I'll be regularly adding updates to this Botzilla entry for changes to the code or changes to the FAQ..

Right-Mouse-Click and Select "Save Link As..." Here for the Current Version of ChartThrob.

Typically, "Save As..." this:
  C:\Program Files\Adobe\Adobe Photoshop CS3\Preset\Scripts\ChartThrob.jsx

To be notified whenever new updates appear, keep an eye on the PhotoRant RSS feed.

(Click here for more.)

Installing ChartThrob

To install it, click "Save As..." on this link. Save the file as "ChartThrob.jsx" in your Photoshop scripts directory, which will typically be something like "Adobe Photoshop CS2/Presets/Scripts" (the location within the Photoshop directory is the same for both Windows and Macintosh)

That's it! The next time you start Photoshop, ChartThrob will appear as an option under Photoshop's "File—>Scripts" menu.

Using ChartThrob

ChartThrob is really two scripts in one. First, it's a script for creating grayscale calibration charts. Second, it's a tool for automatically evaluating scanned prints of those charts and setting up appropriate profiles depending on the nature of your printing process.

The ChartThrob workflow has a few basic steps:

1. Create a Grayscale chart in ChartThrob.
2. Print a digtial negative from that chart (on Pictorico material etc).
3. Contact-print that negative onto your medium of choice (platinum, mimeo, silver-gelatine paper, sun prints, etc).
4. Scan the resultant positive print.
5. Crop the scan back to the original chart boundaries.
6. Analyze the chart in ChartThrob — the result will be a new Curves layer containing a Printing Curve, which you can use on the spot or save for repeated use later.
7. Apply that curve to any B&W images you like, before printing them to digital negatives. The curve will correct the original image grayscale values to neatly fit to the grayscale range of your chosen medium.

So let's begin! From any Photoshop session, you can start-up ChartThrob by selecting "File—>Scripts—>ChartThrob."

If you have no documents open and call ChartThrob, you should see a dialog box similar to the one above (if you have documents open, the dialog will be more complex, but will still contain this info) (The illustrations in this doc page show both Windows and Mac examples). Pressing "Help" will provide you with step-by-step instructions, or pressing "Build New Chart Now" will do exactly that — it will create a new document and start filling it with profiling information. Photoshop draws very quickly, but this will typically take several seconds — especially if you have the 'Numbers' option checked. The result will look like the picture below (with or without the numeric labels).

This is a positive chart — that is, you'll either have to invert it when you print it to a negative, or before (depending on your printer). The text at the bottom reads: "THIS IS A POSITIVE IMAGE WITH DARK TEXT ON WHITE." Keep that in mind, because ChartThrob creates and analyzes positive images.

You may want to resize the chart when printing, by default it's pretty large. You should be able to resize it according to your own printing habits. Then print to a (typically transparent) negative, and contact-print that negative according to whatever process suits your fancy: silver-gelative, old xerox, woodburytype, cyanotype, whatever, so that once again you have a positive print that looks like the original chart. Be sure that you have a solid, dependable printing process so that you can repeat your results later. The chart print doesn't need to be huge, just big enough to see the individual patches (platinum printers will probably be happy to hear that, considering they pay by the droplet...).

If you have a good grasp of your printing already, try to print so that the midtones are as properly-exposed as you think you can get them. The blacks and pure whites will work themselves out.

Also, be sure that your printing is uniform across the entire size of the chart — if the exposure varies from one side to the other, or from the center of your prints to the edges, there won't be any way for the calibrator to second-guess that. You'll just get junk.

Okay, so now you've made a positive print from the chart. Let it dry, and then scan it, amking sure you have a linear (gamma 1.0) scan with the full grayscale range (see the FAQ below on how to do this). Crop the scan back to the boundaries of the chart, and you'll have something perhaps like the image here.

With this new scanned print loaded, call ChartThrob again. The dialog box will still let you create a new chart if you want one, but now it also contains options for analysing a scanned printed chart.

If we hit <return> or press "Analyze," that's exactly what ChartThrob will do: analyze the scanned chart, adjusting for paper tone and process color and evaluating every patch. When done, it will display a brief report telling you everything's okay, and will add a new curves layer to your scanned chart document, called "Print Curve."

If we double-click the "Print Curve" in the Photoshop layers palette to view the resultant curve, it would look like the one shown here (we're just showing the curve rather than the whole dialog, to save web-page space).

The new curve layer is hidden, because the curve isn't meant for adjusting the scan itself — instead, it's for adjusting other B&W images so that they can be printed using the same process that you used to create the scan.

When a ChartThrob curve is applied to a B&W image, the image's original gray values will be remapped so that they will print to match the grayscale range of the target printing medium, as long as you're consistent in the print exposure and processing. So if you expose a silver-gelatin contact print for 30 seconds, then as long as you expose and process all subsequent prints the same amount, they should print consistently and the curves will adjust them perfectly to that tonal range.

You can apply the curve to other images either by saving & loading it as a Photoshop .csv file, or just drag the curves layer from the layers palette onto another picture if it's opened in Photoshop.

With the curve applied, the original image may look dull and washed out on the monitor, but those tones are what's needed to hit the darkest blacks and whitest whites that the particular printing process can handle — at least the tones that were in the printed chart. If the chart is strongly over or under exposed, ChartThrob will still make a curve, though it will tell you if the midtones seem to be strongly skewed.

Checking Your Results: If things have gone well, you can take your original chart (as created by ChartThrob), apply that correction curve to it, print again and you should get a full range of grays from the new corrected print.

The ChartThrob FAQ

Will ChartThrob work with 16-bit images? I keep giving it 16-bit scans but it reduces them to 8-bit.

Yes. You can absolutely apply a printing curve made by ChartThrob to 16-bit images. Just apply the curve in the usual ways. ChartThrob does all of its calibration work in 8-bit, but unless you have an extremely narrow and contrasty process (say, rubber stamps), that should be plenty of precision for the purpose. Applying the resultant curve to a 16-bit image will still give you a 16-bit image.

What about 32-bit images?

Do you have a 32-bit printer? Call me...

Can ChartThrob be ued to calibrate B&W images printed by bookprint services like Lulu and Blurb, or on my local newspaper's press?

If the printer is reasonably consistent from run to run, then absolutely yes.

How do I contact you to make a report or tell you my great idea for an enhancement or to send you money?

ChartThrob is free. You are, however, welcome to buy my photos. You can contact me here at PhotoRant as either 'kevin' or 'bjorke' at photorant.com.

How can I be notified of new versions of ChartThrob?

Use the RSS feed.

DoesChartThrob work with Photoshop CS3, now that it's out?

Yes.

I only have Photoshop 5 (or 7, or CS). Will you change ChartThrob to work with my old version of Photoshop?

Nope. ChartThrob uses UI tools that were introduced with CS2, and Photoshop pre-7 didn't have scripting at all.

Why doesn't ChartThrob set curve points for all of the patches? It only seems to set about a dozen points on the curve.

This is a limitation of Photoshop, but 14 points seems to be plenty for real-world printing.

What do all those extra check boxes do?

Try them out!

Can you make ChartThrob as an Action instead? I don't like having to call scripts from the 'File' menu, I'd rather see it in the actions palette and attach it to a hot key.

Actions can't do everything scripts can do, especially the stuff that ChartThrob does. You can create an Action that will start-up ChartThrob. That's easy, do it the normal way. The Action won't see the stuff you click in the ChartThrob dialogue boxes, though. That seems to be a limitation of the current Photoshop.

I resized the chart scan to be really, really small and now I'm getting junky values from ChartThrob.

The program averages-out an area of the scanned chart and samples this average. It does this to compensate for grain and paper texture. If the scan is too small to get a good average, the quality of the curve will suffer. In general, don't let your scanned-chart image get smaller than, say, 600 pixels across.

The defaut chart size is 4 inches at 300 dpi. I want 7.3 inches at 1440 dpi. Can you resize it?

You can use Photoshop's 'Image Size' command for now. I'm considering adding more print-size options, though they will make the dialog box more complicated.

Should I be setting stuff up using "Relative Colormetric" in the print dialog? What color space should I use?

It doesn't really matter what your print settings are, as long as they are always the same each time you make a printing negative. In general, try to avoid anything in your printing or scanning processes described as "automatic," since such functions may be changing stuff behind your back — and be sure that your scan is in the same color space (particularly w.r.t. gamma) as the B&W images you plan to print.
 
Be sure that your scan covers the full print range from darkest to lightest within the straight-line portion of the scan. Some scanners will "roll-off" the shadows and/or highlights. This sort of compression of the original print values will give ChartThrob a distorted input.

You want the full range when scanning. These dialogs from EpsonScan™
show that the Automatic exposure of a printed ChartThrob chart
distorts the grayscale values. Instead, manually set the max and min pointers
to beyond the brightest and darkest vales, & set the gamma to 1.00.
Note that the entire grayscale will thus be linear: the "shoulder" & "toe" buttons will be ignored.
 

I have a fax scanner and the quality is not so great. I'm losing light (or dark) tones in the printed chart! I can see them, but my scanner can't

Either get a better scanner, borrow a better scanner, or try creating a "camera scan" — take a glare-free, evenly-lit photo of the printed chart using a good-quality digital camera — this make also be a solution for materials where the dark tones are more reflective than the light tones, thus "fooling" the scanner and not scanning well.

Why isn't the generated chart a negative, since I'll be printing it as a negative? It hurts my hand to have to press <Ctrl-I> to invert the chart before printing.

Get a better grip on your printing! Some people print directly from a positive image to a negative and let their print driver do the inversion for them. Others need to print from an inverted image. ChartThrob can't know if you are a negative person or a positive person. So it looks on the bright side and assumes that everyone is positive.

Why does ChartThrob use the font 'Myriad Pro,' which I don't have? I have to click lots of stupid 'okay' dialog boxes when I generate a new chart.<

First: Upgrade your version of (free) ChartThrob and/or (not free but worth it) Photoshop. ChartThrob doesn't actually use "Myriad" — it only uses "Arial." I'm told it's "an Adobe Thing" which was also updated in Photoshop CS3. IFF you get those (annoying but harmless) "font not found" messages: Just upgrade to V1.06 or later. Or, if you're insanely determined to use an old version, install 'Myriad' and those dialog boxes will vanish forever.

Where can I find out more about ChartThrob, and compare results with other users?

See this discussion thread at HybridPhoto.com. In addition, check out this page by Michael Koch-Schulte, who has done additional work in understanding how to get the most out of the varying color sensitivities of different alt-process materials: A Quick Guide to Making Digital Negatives with RNP-Arrays and ChartThrob. Or just poke it into Google — it's been encouraging to see ChartThrob users have cropped up all over the net.

Sometimes ChartThrob seems to take a long time to run. How do I know that it's working at all?

Leave the 'Histogram,' 'History,' or 'Info' panels open, and you'll see just how fast and furious ChartThrob's drawing and analysis really is. It's edu-taining!

ChartThrob: posted 09:08 PM | Comments (2)