I have been trying to copy the development done in camera by my cameras. I was fairly succesfull at times but generally surprised that they are actually doing a lot more than I expected.
Now to facilitate this process I came up with the following:
In RT develop a raw with neutral setting (just adjust lens distortion etc so it matches the jpeg)
Split out of camera jpeg and RT-jpeg into either RGB channels or better LAB channels.
Subtract channels of RT-jpeg from camera-jpeg.
Create a histogram for each channel.
Use curves adjusted by these histograms on the channels of the neutral file. this should then get me the camera-jpeg.
Now my question, has anyone done this?
Are there programs that do this?
I understand that this should be possible in Gmic (@David_Tschumperle?)
My aim is to just save raw in the future and to have the out of camera processing as a starting point for further processing. Right now I am saving both as I do like the out of camera jpeg expecially of the EOS500d, which just creates wonderful warm images.
There must be something in the air!
I read this not too long ago about reverse engineering RGB curves. In Spanish, but there is a link to Google translate.
There’s also some code linked in the article.
That’s from Gui (Guillermo Luijk)… I used and sometimes still do his tonehacker software. He also fathered the pioneer zero noise - http://www.guillermoluijk.com/software/index.htm Specially tonehacker has an interesting concept, some years ago there was this guy developing a simillar idea, a bit also like the AIs nowadays, take this input, analyze it and apply it over there , it was something about automatic adjustments, remember we talked and he was interested in Luijk’s work and (at the time) the newcommer Oleg Sharonov’s powerfull 3DLutCreator which also has inbuilt a “ripping curves” mode
