JTK,Even though I'm no longer working my time is valuable. I first got NIK when it was free. I do hate the bundle but I think it has no rivals.
If precise color match is desired recognize that the closest anybody will come to that is still a judgement call after post processing.
JTK,
I also snagged the NIK bundle when Google made it a free download. Which tool(s) out of the bundle do you use?
Out of curiosity, are you running a Windows 10, an earlier verison of Windows, or a MacOS system? I seem to recall that the free version had some compatibility issues.
Thanks! I tried using your inversion, although I did not use your scanning method as I don't have the time for that right now.
I tried inverting a photo both with default orange layer and with a layer that I previously sampled as a mid point in epson scan (this tends to help and I got it from NLP manual):
$ ./invertscan -fb 0.796,0.498,0.373 -dmax auto neg/img4775.tif ./img4775d.tif
Normalised values are:
max min
Red: 0.749874 0.293736
Green: 0.475074 0.129061
Blue: 0.299794 0.0478828
using calculated FB: 0.796 0.498 0.373
log() values are:
base shadow DMax
Red: 0.0990869 0.0259248 0.432956
Green: 0.302771 0.0204681 0.586434
Blue: 0.428291 0.0948859 0.891529
using auto calculated dmax
dmax: 0.891529
scale: 19123.3
1 image files updated
inverting: ./img4775d.tif
This left the image rather low contrast and quite blue.
$ ./invertscan -fb 0.525,0.529,0.514 -dmax auto neg/img4733.tif ./img4733b.tif
Normalised values are:
max min
Red: 0.54049 0.110246
Green: 0.537621 0.138949
Blue: 0.529473 0.0339055
using calculated FB: 0.525 0.529 0.514
log() values are:
base shadow DMax
Red: 0.279841 -0.0126284 0.677796
Green: 0.276544 -0.00702055 0.5806
Blue: 0.289037 -0.0128807 1.18069
using auto calculated dmax
dmax: 1.18069
scale: 12825.3
1 image files updated
inverting: ./img4733b.tif
This was somewhat more acceptable but rather purple.
Am I doing something wrong?
OhPurple is not a phototechnical color.
I hadn't realised Nik had been acquired by DXO.
I still use the free Google bundle sometimes. I doubt any revolutionary features have been added to the DXO version, and the Google version 1.2.11 is still available to download.
https://www.merriam-webster.com/dictionary/magentaNeither is magenta!
PE
I'm trying to find out how to say orange in phototechnical terms but I'm stuck at "phototechnical" not being a English word.
https://www.merriam-webster.com/dictionary/magenta
People who are skilled in photo lab color correction, especially color enlargement or slide duplication, or with careful use of Ektachrome are very familiar with "magenta." And "orange" is obviously magenta and yellow.
Thanks.Phil, Win 10. I like Viveza, SFEX, and HDR tone mapping. I rarely sharpen... I often go back to PS to tune color locally.
Why? What alternatives? What were your main criteria?I hate it that NIK went to DXO, which doesn't add value for me. Nonetheless I bought the DXO package after carefully considering and rejecting several NIK "alternatives"....
Google for "alternatives."Why? What alternatives? What were your main criteria?
One former member of APUG insisted that Magenta was not a color. His arguments became so emphatic, he was eventually banned. That is how I remember it anyhow.
PE
Adrian, color is defined as the presence of absence of certain wavelengths of light. Thus we have CMY and RGB or Subtractive and Additive systems. But he argued that M was not a color because it does not exist on the electromagnetic scale. It does by inference and observation. Best leaving sleeping dogs lie, or not beat dead horses.
PE
Hi everyone!
I'm working on a new inversion method for both DSLR scans and actual scans.
This method is different from CNMY and provides absolutely fantastic results, but at the moment this is an internal manual inversion that I do, and likely will be for a while.
I'm inviting anyone and everyone to send me DSLR scans so I can demonstrate.
It uses a custom raw linearisation method, custom film profiling, 3DLUTs and batch processing to scale it up and do whole rolls of scans at once.
Alternatively, once the profiling is done, the user can do their own inversions using just a LUT but this will only work on a per roll basis, or on the same film stock developed in the same batch.
This method is currently available as a subscription on my Patreon, with unlimited scans.
The truth is that any other solution is based on approximation and assumptions. Think of this method as the high end of scan conversions, with personalised color grading
That sounds similar to what I’m doing, except you have to send me your film to process and I give you floating point DNG files in return. I do it for C-41, black and white, and E-6. Not sure what you’re specifically doing internally, but I’ve not had a problem dealing with any reasonably processed film. In fact I just finished digitizing a fairly large collection of C-41 that was shot and processed back in the late 70s, early 80s and it came out great. I had to work out the contrast of each channel for each unique emulsion, but once I did, it just worked for the rest of the film shot in that emulsion.
That sounds interesting! Would you be willing to do a comparison?
You can PM me a few of your uninverted scans that I can use my method on, and then post here to compare with yours
Digital cameras do not "scan".
In video modes they do. If you are trying to be semantic about it.
A single instantaneous sensor readout won't give you oblong pixels (unlike certain Nikon scanners).
We use cookies and similar technologies for the following purposes:
Do you accept cookies and these technologies?
We use cookies and similar technologies for the following purposes:
Do you accept cookies and these technologies?