Actually I might just work on the program. These are 2 different versions of the same image and comparing those is more important than having the image perfect. They were both the same exposure and plotting the values in one vs the other could be useful. They both have exactly the same flaws. From the standpoint of comparing a red (or green, or blue) of x value against the same swatch in the other it should still work.
On 10/1/15, Alan Corey <alan01...@gmail.com> wrote: > Last in, first out. I have an HP 2025dn color laser printer which, > once I replaced the demo toner cartridges printed purple flowers as > blue. So I bought the Color Munki and calibrated the printer and my > monitors. The Color Munki software doesn't run under Unix, it's > Windows/Mac only, but most of my computers have multiple operating > systems and once I had the monitor icm files I can load them at boot > with xcalib. The printer's attached to a Windows machine and does > reasonably well except on some browns. > > I carefully did some photographs of my color grid on my monitor screen > with my Nikon D5200, intending to compare the raw to jpeg that way. > The first annoyance was having to defocus slightly to avoid moire. > > The show stopper for now is the backlighting in my monitor, a Dell > S2409 which I've been happy with for years. Notice the difference in > the gray area at left between the top and bottom of the images. > There's ccfl backlighting at the bottom, looks like I should invest in > another monitor. > > The second image is the raw file converted with dcraw -4 -T to a tiff, > then Gimp to a jpeg. The camera's set to save jpeg and raw from each > image. Metering is just averaging. > > I think I could write a program to sample and average the color > swatches, stopping at the white grid at the edge of each. I got up > looking forward to that, then hit the backlighting problem. > > On 10/1/15, Greg Troxel <g...@ir.bbn.com> wrote: >> >> Alan Corey <alan01...@gmail.com> writes: >> >>> I did a little playing with dcraw to tiff and ppm and I was shocked at >>> how different they are from the jpegs. I would think shooting color >>> ramps, measuring, and fitting the results to curves might be close to >>> what you're talking about. >> >> Keep in mind that different != wrong, in some ways. There is scene >> luminance, which raw should arguably be trying to record, and then there >> is developed digital file luminance, and then there is print luminance. >> Photography has a long history of transforming scene luminance (via >> negative density) to print luminance in a pleasing transformation vs a >> quantitatively accurate one. The Ansel Adams writings about the zone >> system and -/+ development is interesting. >> >> However, all that said, when you have a reasonable illuminant, and spot >> meter on a grey card, then you expect to get something that's mid-range >> in the camera-produced jpeg and also in the jpeg/tiff/whatever produced >> by raw conversion with +0 exposure compensation. And also the white >> card should be two stops higher (90 vs 18). If I had more Copious Spare >> Time, I'd focus on just that first. In my experience that's most of >> the issue. >> >>> This is something I was working on for printer calibration before I >>> gave up and bought a Color Munki. I'm not sure I finished a parser >>> for reading back the printed out versions from a scan. Linearity and >>> the range of actual colors would be an issue. I think I was looking >>> at the mixing that happens here because printers use cmyk and I was >>> trying to look at rgb, so whenever I'd ramp up one color others would >>> go along with it. Some of that might have been to create a value >>> change. >> >> There are multiple open source projects for calibration/profiling. >> lprof seems inactive but worth looking at, and Argyll. lcms2 can do >> cmyk/rgb converesions, I think. I have also seen IT8 targets >> available, plus color checkers are not that expensive. >> >> >> > > > -- > Credit is the root of all evil. - AB1JX > -- Credit is the root of all evil. - AB1JX ------------------------------------------------------------------------------ _______________________________________________ ufraw-devel mailing list ufraw-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ufraw-devel
