Software for drawing characteristic curves?

Discussion in 'Black and White' started by juke, Jan 15, 2009.

  1. Hi,
    I have used plain old paper & pencil approach for drawing film characteristic curves and other sensitometry related figures.
    Now I have finally will to try drawing with a computer. But with what software?
    I have tried Excel, but it is not very suitable for task. The line smoothing (fitting) is not very good and it seems to be really complicated to use. No easy tools for measuring contrast indexes, zones, etc.
    I know that there is BTZS software, but that is not exactly what I need.
    What is your preferred software or are you still using paper and drawing curves by hand?
  2. AutoCAD is good, and you can use excel x/y values when inputting points on a "polyline" (pline). Very pricey though. AutoCAD light could do it also, I think it's under a grand. Maybe someone you know has the program and can help?
    Then you can output jpeg or pdf snapshots.
  3. Thanks, I give GNUplot a chance!
  4. I find that the curves showing the absolute densities are difficult to read. My solution is to establish my own standard densities for each zone and graphically show only the differences from these standard values. The scale I use is in f-stops, which makes the graphs easy to interpret and easy to read because of the scale used.
    I enter the values using a normal spread sheet application (called RagTime, but that is not important) and use a bar chart to show the result. I find that better than a curve, as I have only 9 distinct measurements. Why use line smoothing? Isn't that cheating, telling you something that is just assumed?
    There is no reason why Excel should not be suitable for your needs. Perhaps you need to reassess and simplify your requirements.
    Write me ( almqvist ett ) if you want me to email you a sample printout of my chart.
  5. I don't see why Excel should give you any problem either. I've tried a few other spreadsheet programs like Lotus and Star Office, and Excel's chart options are by far the most simple to use. Other spreadsheets swap the axes unexpectedly, or you have to setup the data series differently from the defaults. I've not tried the GNU spreadsheet, but my experience with the GIMP makes me suspect that the Windows port is probably loaded with bugs.
    Anyhow, how much smoothing should a simple scatter plot need? There's a built in smoothing option in Excel's chart wizard, or you can add a trend line, and if that's not smooth enough you can add a column in the spreadsheet to calculate the moving average or whatever other fudge you like.
    Also the chart isn't the place to be measuring contrast indexes from. You should go back to the original numerical data to do that. CI = (X1-X2)/(Y1-Y2), which Excel can calculate in a trice. Delta X over delta Y gives you the instantaneous slope of the curve, which can be plotted in another column, and then the true average gradient can easily be calculated using the AVERAGE function. Bar Gamma can be found by taking the relevant increase in density and exposure values and dividing them.
    What's tricky is going the other way round. Trying to extract accurate numerical data from published characteristic curves!
    Don't get too hung up on the exact exposure value which gives 0.1D above base+fog. No practical measurement system is ever going to land you with the exact figure, except by pure luck. A value between 0.09 to 0.11 or thereabouts is plenty close enough. The difference between batches of film or the accuracy with which you can time your development will more than swamp such a small error.
  6. Hi,
    I did more testing with spreadsheet program. GNUplot requires too much work in windows environment which lacks all scripting tools for manipulating and generating data.
    It seems that spreadsheet is enough for showing general shape of the curve. I am not going to deep with sensitometry. Another option could be use of spreadsheet for plotting coarse curve and then use drawing program with bezier curves for drawing smooth curve (which is quite closely my approach on paper).
    The main things in my measurements are the speed point, zone VIII (and thus contrast). All other information is just for comparing quickly charasteristics of different films and/or developers and developer dilutions.
    Which shape does toe have, is there shoulder on the curve?
    That is information what I like to see from the curves. That is also the reason why curve smoothing is necessary. The measurements will always have error marginal, there can be also error within one measurement which is originated either from exposure, development or measuring phase. For such errors, the smooth curve is good.
    The spreadsheet that I have used has one problem with smoother curves (trendline), it tends to emphasize curvature too much.
    For example, here is one curve family as an attachment. As you can see, the N+1½ curve goes belove the N+1 curve at lower densities. However this is not the truth, in reality it is upper but trendline drawing algorithm has troubles with N+1 curve (as can be seen from it's shape).
    About these curves: The developer I used is staining and densities are read with blue light. I have also made test exposures for finding speedpoint for my materials with this stain color. The speedpoint is quite accurate 0.09D above base+fog and really good visual VIII can obtained from density 1.35D above base+fog.

Share This Page