Lapser
Thanks for the download, however, for some reason (most probably me), when I load your build my S95 goes unstable, even before I try and use your calls.
Should I ‘just’ replace the files you sent me, or delete all the previous CHDK files and only use your files?
BTW the reason I am interested in your histogram calls is that I want an auto bracketing script that ‘guarantees’ to eliminate any overexposed areas.
My current script (based on one by msl) uses the get_histo_range, but the granularity/accuracy is not sufficient, ie 0 to 100% returned.
Once I get your calls working, my thinking is as follows:
• Manual mode, manual focus, spot meter on the darkest area that I am interested in seeing some structure, place this at 0EV (or wherever I wish), freeze AF;
• Use your count=get_histo_range(2046,2047,0) to get the actual sample count in the end most ‘overexposed’ bin. In other words, I feel all I need to do is ensure this final bin is addressed, then I have ‘guaranteed’ I have no overexposed pixels in the image, ie I have maximized the data content of the brackets for further 32-bit processing.
• Loop and change the exposure of the next shot until the end bin count is less than a ‘sensible’ number.
Using the above algorithm/workflow means I don’t need to worry about the ‘underexposed’ end of the brackets, only the ‘overexposed’ end. The rationale being that I don’t have to complicate the algorithm addressing the black level end of the histogram, which for my 12b S95 is at 127.
Looking at RAWdigger data, I am confident that if I sample the count in the end most bright bin and ensure it is less than a sensible threshold, that should be OK; say less than 50 or 100 to account for sensor statistics.
I would be grateful for any thoughts you (or others) may have: first to allow me to get your build running and then any thoughts on ‘my algorithm’.
Cheers
Garry
