I suppose I might as well throw this up here as well, just on the off-chance that any developer-type friends have seen the image manipulation symptoms before. We're doing something that involves converting PDF pages to images on the fly using the PDFBox library, but certain images that look perfectly all right embedded in the PDF come out a bit... wonky. For example:

I think that I'm quite close to identifying what's wrong, and after analyzing the differences in how two libraries handle the colour models and spaces of PNG-encoded images, ( I have managed to produce this. )

I think that I'm quite close to identifying what's wrong, and after analyzing the differences in how two libraries handle the colour models and spaces of PNG-encoded images, ( I have managed to produce this. )