Is CDAF really derived from in-cam JPG adjustments?

Started May 27, 2012 | Discussions thread
Timur Born
Senior MemberPosts: 3,751
Re: Correction to OP: Is CDAF really derived from in-cam JPG adjustments?
In reply to Detail Man, May 28, 2012

Basically there seem to be two forms of auto-focus used by both my Olympus E-M5 and Fujifilm X10.

1) Focus acquisition with dedicated sensor settings that are mostly independent of whatever JPG settings you set up. This happens with AF-S on both cameras, where the screen freezes for a short duration on X10 and changes exposure, likely analog gain, maybe saturation ( de_creases) and most importantly contrast (higher than +2) on E-M5.

Furthermore on the E-M5 this AF method is used with initial AF-C acquisition when you first half-press the shutter and very likely before every single frame of low speed burst mode (up to 4 frames). I sat for over one hour at a busy street photographing cars at different focal length and different JPG settings and color modes (including Vivid) and did not experience any differences.

The only settings that seem to make a difference in some situations are Gradations and Frame Rate.

2) Focus acquisition with current live view settings. This method may or may not be affected by current JPG settings, but some indications point to that it's not. Among other tests I managed to focus on a printed piece of white paper that was clipped to complete white in live view due to JPG settings. If these settings had an impact on auto-focus how would the camera focus on pure white without any contrast?

On the E-M5 this AF method is used when you keep the shutter half-pressed in AF-C (after the initial half-press). It seems to work quicker/more aggressive in this case with more back and forth hunting even after holding the camera still. It is also used when the "Full-time AF" function (custom menu -> AF) is active, but seems to work slower/smoother there. Last but not least a combination of both these behaviors (quick vs. smooth) seems to be used by video recording in combination with AF-C, with a tendency towards smoother. Again Frame Rate can have an impact on AF performance in those modes where "High" frame rate is allowed.

When AEL is not used you may see screen flickering with this mode as the E-M5's exposure metering tries to keep track of light conditions. This happens when the light changes abruptly in both stills (especially with fast frame rate) and video modes (especially with manual zooming). As a positive result both AF methods 1) and 2) seem to meter the same exposure.

On the X10 this AF mode is used in AF-C, but only while the shutter button is not touched at all. Once you half press the shutter the X10 reverts to using method 1) before taking the shot (and stops focusing when the shutter is held in half-press position). There is no flickering from auto-exposure on the X10, but AF methods 1) and 2) may come to different exposure metering results. This means that once you half-press the shutter (to invoke method 1) you may get different results than what live view displayed right before.

PS: As an interesting side-note I noticed that the E-M5 shows some kind of local "discoloration" on the focus DOF area in live view. You can literally see where the DOF focus area is on the screen, at least when aiming at green (trees, grass) and gray (stone floor) areas. I'm not sure if it's a form of lesser saturation or WB difference and have to check tomorrow if it also happens with the EVF and 12-50 (only checked the 45/1.8 yet). Happens with both AF methods and manual focusing.

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