ChartThrob V1.01

[bjorke]

Click "Save Link As...." here to get a new version of ChartThrob.

What it is: a Photoshop script for CS2+ -- install it in your CS2/Presets/Scripts directory and restart Photoshop. It will appear in the File->Scripts menu.

What it does: It creates (positive) grayscale charts. If you print these charts via digital contact printing (or any other process), you can scan the results, run ChartThrob a second time to analyze the scan, and it will automatically create a Photoshop "Curves" adjustment profile that you can apply to digital positives before printing to ensure that they get the full range of available grays from your wet process.

Caveats: your process must be consistent between prints, and the tones should not vary wildly depending on location on the print (that is, 50% gray should always be 50% gray whether it's in the center, corner, bottom, top, etc)

Feedback most appreciated!

 

[bjorke]

BTW, here are the changes from the previous version:

1. In some cases, slight numeric errors (eight bits can only hold so much!) would cause some colors to be calculated as slightly brighter than Photoshop's permitted color range, which would cause a script error. These errors are now correctly avoided. This occured principally in circumstances where the print had been made on very dark or unevenly-toned paper, or the original chart had been accidentally printed as a negative (it's meant to be a positive).
2. The chart text and 'Help' dialog box have been modified to make it clear which images and prints are positives, and which are negatives
3. The ChartThrob main dialog box was upgraded for clarity
4. A DPI setting for new charts was added, though it's a bit superfluous -- you can always resize the chart as you see fit from the Photoshop 'image size' dialog, because what matters are the color values of the patches -- not the pixel counts
5. When analyzing scanned prints of charts, ChartThrob now lets you know where in the grayscale range it found the dmin and dmax of your scan. Generally, if these are at the extremes of the chart range, then you may be served by adjusting your exposure to get a wider range of tones

 

[David A. Goldfarb]

Sounds very cool. I'll be interested to hear how it works for those who are printing from digital negs.

 

[tom_micklin]

Will this work in Photoshop CS, or is it just for CS2?
Thanks,
Tom

 

[bjorke]

I have updated to Version 1.02 - same link as above.

ChartThrob uses the 'jsx' script rather than plain 'js' -- so CS2 is required, since some of the UI elements are not present in CS (or PS-7).

The latest revs correct some weirdness at the upper or lower ends of the scale (particularly when the scanned exposure range goes past either end of the scale due to underexposure or overexposure), improved layouts for the dialogs and more constructive online help.

kb

BTW, it's funnest to run the script if the Histogram panel is exposed.

 

[David A. Goldfarb]

I don't make digital negs or own any kind of photo quality printer, but just for the sake of seeing how this script works, I ran it using my HP5L, and heck, ya know, it does what it's supposed to do. Of course the result in this particular case still looks like something off a crappy desktop laser printer, but with the right process, I can see how this would be a very handy thing. Well done, Kevin!

 

[ann]

ok, i take david's word, but how does one download this to even test it out.

all i see is code when i click on the "here".

 

[bjorke]

ok, i take david's word, but how does one download this to even test it out.

all i see is code when i click on the "here".

I've made a wee page for it: ChartThrob Instructions and FAQ.

Basically, right-click on the ChartThrob code link and choose "Save As..." or "Save Link As..." and save it as ChartThrob.jsx.

Or follow the more detailed instructions I just mentioned.

cheers,
KB

 

[Digidurst]

Hey, just curious to know... is there a particular color space you suggest for this? What about printer settings? Thanks!

 

[sanfrancisconudes]

Very cool! I printed a test negative and will try it out on new cyanotype first...

 

[bjorke]

Hey, just curious to know... is there a particular color space you suggest for this? What about printer settings? Thanks!

As described in the FAQ, it really doesn't matter -- as long as your workflow is consistent and repeatable. Do try to avoid any setting that might "automatically fix things" and adjust the brightness, but otherwise you can use any color space you like. As long as it's consistent (your scan and print should be in the same space).

 

[mkochsch]

Scanner Settings for ChartThrob

As described in the FAQ, it really doesn't matter -- as long as your workflow is consistent and repeatable. Do try to avoid any setting that might "automatically fix things" and adjust the brightness, but otherwise you can use any color space you like. As long as it's consistent (your scan and print should be in the same space).

Kevin,
Where should the black and white points be set for the scan. For example, using EpsonScan should I turn off the "Auto" setting and use an unadjusted scan as my input for ChartThrob. Or, should I turn off and setting and adjust the Black Point/White Point in CS2 and then feed that into ChartThrob? Thx.

~m

 

[bjorke]

Where should the black and white points be set for the scan.

Good question & I should put it in the FAQ. What's important is that the scan gives a good coverage of the full range of the chart. ChartThrob finds the dmin and dmax itself and set the printing curve relative to that range, but only if the dmin/dmax are there in the scan. If your highs are blown out or the darks blocked-up, ChartThrob won't know that there's more shadow or highlight detail -- it'll just react as if your printing process is very contrasty.

I recommend that you set your points manually so that they can be consistent and predictable -- in Epson scan, iirc, you can do this by clicking the little histogram button and inputing numbers directly. If you can sample the scan preview for the lightest and darkest areas, use those values with a little bit of cheating -- that is, let the black number go slightly lower than the black sample and the white slightly more than the white sample -- this will give you a little slack in case your process or the paper color is prone to a little variation.

Then again, since the chart DOES have a full range of tones and is evenly distributed (if printed at a 'standard' exposure, where 50% gray is near the middle), many scanners should do well on "auto." Again, what's important is that the full range of printed tones is represented.

 

[BillSchwab]

Well I'll be ....

Kevin,

I have to say I was prepared to be disappointed in this script simply for the fact I spent many hours building curves with a reflection and transmission densitometer. I'm pleasantly surprised so far. The true test is in the burner as we speak. The curve that was generated was surprisingly close to what I had spent hours building and the resulting negative reads very similar to my control done with my curve. We'll see how it looks when done.

Thanks!

Bill

Addition: Well, I've run the print and perhaps I spoke too soon. A little hot in comparrison to my control. Too much contrast blowing out the subtle midrange I am used to. It got pretty close though. I'm also assuming variables in my workflow have had an impact. Bt the way... the end result is a pure palladium print.

 

[bjorke]

A little hot in comparison to my control. Too much contrast blowing out the subtle midrange I am used to. It got pretty close though. I'm also assuming variables in my workflow have had an impact. By the way... the end result is a pure palladium print.

Interesting! And I'm very keen to figure out what's different and to correct accordingly. My initial guess is that there are differences in the values you get from your densitometer and your scanner. This can be due either to [1] gamma differences between print, scan, or comparison image; [2] shoulder-and-toe rolloff in the scanner curve; or [3] inadequacies in the scanner's ability to discern values at the high and/or low ends (in other words, your densitometer may be more accurate than your scanner).

#3 is a hardware limitation, but the first two may be entirely correctable.

Did you test by making a regular image print, or by printing the chart with both your original hand-made curve and the automated one?

What would be really great would be if you could send me some detail info (bjorke at botzilla dot com): the scanned chart (a jpg is fine), the curve you've been using previously (as a .csv file), and, if you're up to it, a few of sample values from the printed chart made with your densitometer (so that I can see if there are differences in the scanned gamma). With that info, I'll bet I could get at what the issue is pretty quickly.

Thanks Bill!

 

[mkochsch]

Setting Black and White Points

I think this step is critical Kevin. Letting people do nothing or letting default software settings be the norm probably won't work. In my experience I've always set the points manually, in Photoshop. This varied somewhat depending on what method I was using to make my paper graphs. One way I would set the BP/WP to cut the headroom off the scanner, the other way I would target the first black and white steps on the graph for a given emulsion.

I'm guessing (I'm going to try tonight hopefully) that setting the points manually in Photoshop is the best method (as long as the scanner software controls have been set to off). Setting the Black Point to near 100 per cent and White Point near to the 0 square is the correct method. e.g. Even if pure black is printed from 85-100% and white is printed from 0-12% does ChartThrob take this into account and set it's own internal BP/WP to 85% and 12% (for example)?

 

[BillSchwab]

Kevin,

I've got to believe that it is my scanner and the way I scanned the image. I need to do more tests before I give-up. The ramp it created was really nice, just a little dense for the purpose I am using.

I basically dropped color control altogether in the image and in the print driver... much the same as with Burkholder's method. The scan was done without any auto adjustment and I simply pinched the histogram to the high end and low end like I saw you describe in another post. Like I said, it came really close which makes me believe it was pilot error somewhere. I'll do a negative of the corrected chart a bit later and post you a scan of it and the one done with my curve.

Thanks, Bill

 

[bjorke]

Even if pure black is printed from 85-100% and white is printed from 0-12% does ChartThrob take this into account and set it's own internal BP/WP to 85% and 12% (for example)?

What it does is march through the grayscale samples (you can see the areas it chooses if you click "outlines" when analyzing the scan). It finds the darkest one and the brightest one and uses that range as the dmax/dmin limits of the resulting curve. The points along the curve are determined by comparing the expected gray values to the scanned values, normalized against the calculated dmax/dmin as found in the scan. ChartThrob can then know that for any given value, say 20%, that the print value that would return 20% between the print dmax and dmin is between scanned chart patches 32 and 33 (for example) and will return a value appropriate to that, so that then the curve is applied to another image, that 20% gray in the original image will be remapped to 32.7% (or whatever, depending on the scan).

The whole scheme depends on linearity of the inputs, if the scanner is "curving" the data then that's going to skew the results.

 

[bjorke]

Thanks Bill. What would be most-valuable would be the items I mentioned above -- a jpg of the scan you used to calculate the erroneous curve, your handmade curve for comparison, and if you can a couple of densitometer samples from the printed chart (full-black, full-white, and 50% gray would be optimal, and let me know where you found them in the chart)

 

[ryberg]

ChartThrob works!

I've completed curves for cyano and VDB--both the charthrob curves are better than the ones I had hand constructed. Yes, it requires CS2 but you can download a 30 day trial, make your curves and save them for use in earlier versions.
Charles Portland OR

 

[dogzbum]

Well, this is superb. My only gripe is that I had to click through 675 double clicks due to a font substitution issue. (OK, I exagerate, it was only 673 double clicks).

Is there any possibility of using a more common font (Myriad Roman is not on my system) or is this just a 'dogzbum' problem? Be honest, many of them are and my shoulders are broad.

My 2 cents? Arial, Century Gothic, Tahoma, Courier.

However despite needing to buy a new mouse I now have one pristine digineg in my hot sweaty hand and will at some stage cyanotype or vandyke the little critter.

People donating their time and effort to make free tools like this restores my faith in humans.

Ta,
dogzbum

 

[dogzbum]

Well, this is superb. My only gripe is that I had to click through 675 double clicks due to a font substitution issue. (OK, I exagerate, it was only 673 double clicks). ..........

What a lot of whining. :rolleyes:
A quick Google and I've now got the font installed.
It may have been at the following url... (but it was not me who told you).

hypertextterminalprotocolcolonforwardslashforwardslashwwwdothawaiiclubdotinfoforwardslash<CAPS>f<CAPSOFF>ontsforwardslash

(I enjoy analog photography more than digital).

dogzbum

 

[bjorke]

I'm guessing that the font complaints are about the WEB PAGE, not the program -- the program uses Arial. The web page actually specifies Lucida and calls Arial as a fallback, you can examine the site-styles.css if you have a web CSS-analysis tool installed into your browser (there's a great one for Firefox that will let you completely dissect ANY web page on the fly...). My guess is that it's an older browser....? Or... what? In fact I don't find Myriad specified anywhere.

As for scanning. I've added a bit to the FAQ showing how scans may introduce errors -- a default scan on my scanner using EpsonScan, for example, sets a non-linear gamma and introduces nonlinearities at the shoulder and toe of the curve. The FAQ now shows how to correct this problem.

 

[sanfrancisconudes]

I got font errors too - when you use the program to generate the test chart the labels in particular generate the error. I only got it a few times presumably because I only got it on the text below and didn't have the numbers turned on.

I've made New Cyanotype and Vandyke charts. I'm hoping to get curves tomorrow and make actual images Friday...

 

[mkochsch]

ChartThrob Neat Trick

Oh here's a neat trick. Use a "pre-selection" colour for your particular process and standard print time as you might do with one of those "other" digital negative systems and overlay the chartthrob grid using a colour layer in "screen" "mode" in the new layer dialog. You now should be looking at the ChartThrob step wedge as a nice green or orange pyro coloured negative...
Holy cow! You're now using colourised negatives! And all the advantages they give! ie. tuning the 0% square to be just dense enough to produce the white you want on the page for your highlight white point...more to follow...I think I can hear the sound of coffee being spit out...