Hope this doesn't hijack the thread, but I had the above question as well as changing the pixel dimensions and then converting to jpeg, for projection on a projector/computer combination that are profiled, but I have no way to softproof for. My original images are Tiffs, 16-bit color and ProPhoto RGB color space, and post-processed from Canon 5DIII CRW's. They get reduced to 1024 pixels on the longer side, 8-bit and sRGB. The sequence I had been using is:
1) Convert the color space to sRGB,
2) Downsample to 1024 x 688 or whatever the other dimension comes to),
3) Convert from 16-bit to 8-bit color,
4) Save as Jpeg.
Is there a better sequence to use, or does it not really matter that much?