The adjustments available in the UI on DSLRs are 1/3 ev, and on the Canon DSLRs the adjustments the camera uses on automatic settings are 1/8 ev. I found this out simply by superposing EXIF data on timelapse tests:
(A single link shows up as a double video, sorry... don't know what's going on there)
This video shows exactly why I want smaller ev adjustment steps on DSLRs. The 1/32 ev steps on the compact camera result in smooth looking timelapse while the 1/8 ev steps on the DSLR cause quite visible and distracting steps. So I'm curious exactly what in the firmware determines the size of these steps (in order to realize whether it would be an easy fix), and my first idea was that there was simply a table of allowed exposure times. But now I understand it as if the step size is a predefined unit and then the calculated shutter speed can only be an integer multiple of that unit - is that correct?
I've presented the issue in the DSLR hacking forum, hoping that somebody might take a look at it, but odds are probably better if I can be specific (and correct!) about what it would take to do this hack. I have no hacking experience myself and know it would take a lot of time to get into, so I'm hoping somebody who has the skills already would consider doing it. The exposure stepping you see in above video is probably the biggest and most discussed problem among timelapse shooters worldwide, and it would be a *big* deal if a fix was developed. Personally I've promised a $100 reward to whoever can come up with a fix I can apply to my own cameras. But if nobody has the time/motivation to dig into this, then I'll probably try to look into it myself if I can find the time. In which case it would also be very valuable to know exactly what it is I need to do.