Another C-AF test with the E-M5

Started May 15, 2012 | Discussions thread
Senior MemberPosts: 4,900Gear list
Re: Another C-AF test with the E-M5
In reply to Detail Man, May 16, 2012

Detail Man wrote:

So you are saying that (some sort of stated opinions exist) that Panasonic as well as Olympus camera models AF systems in general (whether in a Continuous AF mode or not) may be affected by in-camera JPG user-settings ? Such a proposition sounds entirely counter to what (it seems to me) would constitute sensible as well as reliable digital camera system design practices.

It would be hard to implement CDAF algorithm on Bayer sensor output without doing demosaicing first, and sensible implementation of demosaicing is just a single multiply-add operation matrix operation between a matrix of sensor outputs and a matrix encoding mosaic pattern, WB settings, distortion correction settings, CA correction settings, vignetting correction, noise filter, exposure curve, probably sharpening too (although for the latter I don't know the easy way to encode it to a matrix for a simple multiply-add operation). This operation is needed for AF, live view, video and JPEG, but is not related to JPEG encoding itself.

If we would want to decouple demosaicing from the rest for AF purposes only, we would need TWO similar matrix operations with two different matrices. That approach would double hardware resources needed on the processor and potentially half battery life of the camera.

How have (or could) people evaluate this hypothesis on a truly objective basis ???

Shoot focusing on lighter grey/darker grey pattern, then decrease the contrast to the point where the shades are not distinctive in LV and try to focus on it again.

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow