[stevebrot]

I was looking at some photos I recently took and noticed that in some of the shots the face was out of focus when viewed at 100%. When I was checking the focus to see where the focus point was, it seemed that the centre and bottom half of the image appeared much sharper, which is odd because I would think that everything is roughly on a plane to the camera.

As noted above, the plane of focus is not parallel to the subject with the camera pointed downward. I estimate the distance to be about 5m with calculated DOF of about a meter (45cm near, 55cm far)

Autofocus mode would be either AFS + single point or manual focus.

Assuming AF single point (center)*, the intended point of focus of IMGP5637 was at the top of the fly flap of her pants. Focus on that point is good with decent detail captured. Reasonable sharpness at full resolution viewing is maintained at the legs as far down as the knees, but decreasing paradoxically further down with the bricks/mortar continuing to show detail. The drop in sharpness moving up towards her head is much more abrupt and with nothing in reasonable focus (bricks included) much above the bottom hem of her top.** The area of acceptable focus is neither planar nor regular to the point of focus.

Assuming other lenses are functioning properly on the camera, it is safe to assume that the there may be a serious problem with this lens caused by a fault in element placement/alignment.


Steve

* Unfortunately the uploaded files have none of the makernotes that would have been helpful in determining what happened here. OOC TIFF, DNG, or PEF would have been preferable.

** This is counter to the camera angle. Expectation would be sharper bricks towards the top.

---

Last edited by stevebrot; 05-18-2018 at 12:57 PM.

[Qman]

It looks like it's back focused and the camera is tipped down. The bricks at the feet are roughly in focus and the focal plane is tipped away from the camera as you move up. At her head level the focal plane is behind the brick wall. I would suggest that you focus on her head and use more like f3.2.

[swanlefitte]

Could this be focused on eyes half shutter press and recomposed? It looks like focused behind low focused at belt middle focused in front up high? I am having trouble visualizing the plane shift to figure out if that would be a cause.

[dofmaster]

As noted above, the plane of focus is not parallel to the subject with the camera pointed downward. I estimate the distance to be about 5m with calculated DOF of about a meter (45cm near, 55cm far)




Assuming AF single point (center)*, the intended point of focus of IMGP5637 was at the top of the fly flap of her pants. Focus on that point is good with decent detail captured. Reasonable sharpness at full resolution viewing is maintained at the legs as far down as the knees, but decreasing paradoxically further down with the bricks/mortar continuing to show detail. The drop in sharpness moving up towards her head is much more abrupt and with nothing in reasonable focus (bricks included) much above the bottom hem of her top.** The area of acceptable focus is neither planar nor regular to the point of focus.

Assuming other lenses are functioning properly on the camera, it is safe to assume that the there may be a serious problem with this lens caused by a fault in element placement/alignment.


Steve

* Unfortunately the uploaded files have none of the makernotes that would have been helpful in determining what happened here. OOC TIFF, DNG, or PEF would have been preferable.

** This is counter to the camera angle. Expectation would be sharper bricks towards the top.

Thanks for your input. it's likely that I was spot focusing on her eyes/face, but obviously the midsection is the clearest area.

I dropped the DNGs in the drive folder.

Update your browser to use Google Drive - Google Drive Help

As for your suggestion about lens element, I have also noticed that it feels like there is more resistance than usual when zooming, mostly when zooming from 24mm.

---

Last edited by dofmaster; 05-18-2018 at 01:42 PM.

[stevebrot]

I dropped the DNGs in the drive folder.

Woo! Hoo! Smoking gun! From IMGP5637.DNG...

Code:

AF Point Selected               : Far Far Left; Single Point

Ignore all that stuff about defective lens I wrote above

I don't have a focus point overlay for the frame, but I suspect that the point of focus was a brick somewhere between her knees with the actual plane of focus being deep inside the chimney somewhere at her head level. Of course that assumes right side up for the portrait orientation.


Steve

[dofmaster]

Woo! Hoo! Smoking gun!

Code:

AF Point Selected               : Far Far Left; Single Point

Ignore all that stuff about defective lens I wrote above

I don't have a focus point overlay for the frame, but I suspect that the point of focus was a brick somewhere between her knees with the actual plane of focus being deep inside the chimney somewhere at her head level. Of course that assumes right side up for the portrait orientation.


Steve

I normally shoot right side down for portraits.

But regardless, does that hold for IMG5640 as well? Because there you can clearly see that the focus plane is far in front of her face, there's nothing behind her in focus in the top half of the frame.

[stevebrot]

But regardless, does that hold for IMG5640 as well? Because there you can clearly see that the focus plane is far in front of her face, there's nothing behind her in focus in the top half of the frame.

IMGP5640 is also set to use the same Far Far Left focus point. I honestly can't say if there is a plane of focus at the level of her face, though I do get your point and we are back, at least in part, to the lens. As suggested above, a test setup to adequately characterize the problem should account for variables such as camera orientation, intended point of focus, and whether focus was actually attained. To whit:

• A flat, smooth subject with areas of fine, high contrast detail (newsprint?) in the areas of concern including the center.* A bright/evenly lit interior wall would work. It is important that the subject be easy to focus on, flat, and even. Brick walls are not a good idea in that neither the surface nor the flatness is particularly consistent.**
• Test at 65mm and f/2.8, the same settings as the images at the start of this thread.
• Moderate distance, say 20x the focal length of the lens (1.3 meters for this case where the focal length was 65mm). The moderate distance is important because ease of focus is essential and most AF lenses have short-throw focus rings. In addition, out-of-focus is harder to detect as distance increases without an increase in magnification.
• Use a sturdy tripod, SR off with either mirror lockup or 2s delay. Electronic shutter is probably not required for this test
• "Zero" the camera to the target subject such that the sensor is parallel to the target surface. The in-camera levels help, but only account for the horizontal axes. The vertical axis may require an assistant with a tape measure and a ruler held against the lens mount face on the camera parallel to the floor.
• Use manual focus with magnified live view. We are not testing the AF system nor the focus screen calibration. Unless using the electronic shutter, switch out of live view for the actual exposure.

If the problem does not show at 1.3 meters, try again at the same distance as in the example photos on this thread. Both focus and alignment will be more difficult, however. The reason for all this pain has to do with how internal focus zoom lenses do their work with a system of sliding elements/groups that accomplish both zoom and focus by relative position of each element or group. As a result, poor performance due to alignment or position issues may depend on a specific combination of focal length and and subject distance and may not happen any other time.


Steve

* Usually that would include center, the four corners of the frame and a vertical and horizontal line through the center.
** To be honest, brick walls are only useful for evaluation of barrel/pincushion distortion and even then, much depends on the skill of the bricklayer.

[swanlefitte]

If it was the far far left at 65mm not 70. Was it focused at 70mm and backed off to 65 for framing? Seems even if the lens was bad but that was where focus was being set it should be better. So i would guess front focus issue.

[stevebrot]

If it was the far far left at 65mm not 70. Was it focused at 70mm and backed off to 65 for framing?

I dunno...the EXIF says 65 and focus was acquired. I would expect that the lens is parfocal enough to absorb that much shift. After all, the DOF was likely over a meter.

So i would guess front focus issue.

Explain why the bricks next to and between the model's legs are in focus. That is paradoxical since the camera was pointed downward. I don't think we can know for sure until the intended point of focus is confirmed and that is best done with fully manual focus.


Steve

[swanlefitte]

Explain why the bricks next to and between the model's legs are in focus. That is paradoxical since the camera was pointed downward. I don't think we can know for sure until the intended point of focus is confirmed and that is best done with fully manual focus.


Steve

The plane is tipped down but also rotated counterclockwise to the model so the right bottom moves back as the left top moves forward.

And a nice test shot or a series of panned or focus changed would definitely help.

[stevebrot]

The plane is tipped down but also rotated counterclockwise to the model so the right bottom moves back as the left top moves forward.

Good points (I think). I will have to research if reported pitch angle is effected by on camera rotation (reported as 90 CW). The actual plane of focus does not depend on camera orientation. It is always orthogonal to the lens axis. Based on the perspective of the shots the camera was being held fairly high with the lens axis pointed at the models belt line.


Steve

[rawr]

It would be a good option to shoot LiveView with face detection turned on, on a tripod or monopod, for these kind of very static subjects.

I'd also bump the aperture up a notch to f4 to help widen the plane of focus a tad, so that more than her nose or eyes can be in focus once the face is acquired.

[swanlefitte]

Steve,
photo middle is around the belt under her left shoulder. With only the left rotation the plain from that point on the belt keeps the head far behind the same plane. Now tip down and I guess at around 20 degrees somewhere on the head and center of picture are on the same plane. Tip a little more and the head is in front of the frame. I am guessing its a little more. I can't guess precise enough to know of course. My bet is on the head in front of this plane.

[stevebrot]

Steve,
photo middle is around the belt under her left shoulder. With only the left rotation the plain from that point on the belt keeps the head far behind the same plane. Now tip down and I guess at around 20 degrees somewhere on the head and center of picture are on the same plane. Tip a little more and the head is in front of the frame. I am guessing its a little more. I can't guess precise enough to know of course. My bet is on the head in front of this plane.

It depends on which photo, obviously. With the one where she is facing head on, center is at the very top of the flap on her pants fly. The plane of focus, regardless of where it is takes the angle of -8 degrees from vertical.* I did a little more research and found that the intended point of focus is at the model's chin, meaning that it precedes towards the photographer (or tripod) from that point down and recedes from the photographer from that point up. Calculated DOF is about a meter with 45 cm fore and 54 cm back.

What we would expect is for the chin and head to be in decent focus along with stuff a half meter behind and for vertical elements of the photo below her chin and behind the plane of focus to become increasingly soft working down. Instead, we see the point of focus and everything behind it as very soft and details of the model becoming increasingly sharp to about her waist and then softening a little working down with the chimney behind her becoming becoming increasingly sharp to about her knees.

Part of what we see might be a matter of posture and/or pose, and might be explained by 25 cm or so of front focus combined with 30cm or so of hip thrust. However, even if we can explain what we see with the model, it all falls apart with the sharpness of the chimney bricks. Astigmatism?

I dunno...

Steve

* I tested with my K-3 and found that the leveling sensors are not sensitive to roll. If the camera is pointed down, the sensors report negative pitch.

[madbrain]

Looks like wrong AF point is the smoking gun.

Also, very few lenses are at their sharpest widen open. Stopping them down a bit usually improves things This is partially because of larger DOF, but not only.. Since you were shooting in daylight, stopping down was likely a good possibility.
Tripod would not be required in such lighting conditions. Just maybe slightly higher ISO, or slower shutter speed. The K-1 stabilizer is excellent. You may just want to take several identical shots if using something very slow (like slower than 1/5) since it's difficult and time consuming to tell if the pics are at their sharpest by viewing them on the LCD, as opposed to later on on a large computer monitor. Using high FPS drive mode and holding the shutter button only once can be helpful in getting sharper shots at slow shutter speeds.