The first is a crawl-walk-run problem. How do you debug? I am using VS2005 and PS v8 aka CS. I have successfully built some of the filters, told PS where they were, and they do indeed come up and work. But when I try to debug from VS, having told it to use PS as the executable to use the filter dll, PS starts, and then an error message comes up--"An error has been detected with a required application library and the product cannot continue. Please reinstall the application"--Uh what do I reinstall, PS or what. I have seen similar issues writing dll's for Matlab--it was a matter of what directory the debug dll was put in.
The second question regards what I think is a screwy idea. A restricted class of images will come in with a V4 ICC input profile attached. It would have AtoB0 and AtoB1 tags of LutAtoBType of the particular form M-Matrix-B. The filter record structure I see among other things contains this profile. It is proposed the filter would modify the B-curve to get its effect and the image would not be modified as it is in PS. The visual effect would only be seen in a small image in the filter dialog box. The filter effect on the image would turn up in the printing step, as the PS color engine combined the modified input profile with the printer profile. While this could conceivably work, there seem to me to be a number of reasons it probably won't. The most obvious one would be that PS doesn't copy and use the modified profile when the filter dialog is closed. Then there is the question of stepping on memory if the replacement B-curves are larger than the placeholders. etc. etc. I don't see any advantage to doing things this way, but I am urged if not pressured to consider it. So is this a unlikely as I think it is?
Mike