Long story short, it appears to me that ETZ21 Pro firmware version V1.38 has improved the image quality of adapting wide angle lenses to Nikon Z, such as the Voigtlander 21mm f1.4 and Sigma 14mm f1.4. Has anyone had the same experience or any idea how a firmware update can affect image quality?
Back story. I was testing my Sigma 14mm for astro photography on my Z8. I was trying to compare the sigma on my a7rIV vs Z8 + Megadap vs Z8 + Megadap shimmed with tape, and Z8 + FotodioX Pro Fusion. To my surprise, I found the Z8 + Megadap to be performing the closest to my a7rIV, which was the complete opposite experience I had with my Voigtlander 21mm f1.4 in the past. After some further testing, I started to suspect it was the firmware version, and found even my Voigtlander 21mm f1.4 is performing better on the Megadap adapter now.
I have spent quite a bit of time adapting Nikon F lenses to Sony E and Sony E lenses to Nikon Z, and this is the first time it appears that adapter firmware may be affecting image quality.
You'd really need some controlled environment test shots with both versions of the firmware, otherwise it's all speculative and potentially confirmation bias.
I'd be curious if the adapter could report the lens as a different lens between the two firmwares, leading to different lens corrections in whatever post processing app you use.
There is no possible way for firmware to affect image quality on a manual focus lens. The CV 21/1.4 doesn’t do well on the Megadap (non +) because it’s too thin.
RoamingScott wrote:
I'd be curious if the adapter could report the lens as a different lens between the two firmwares, leading to different lens corrections in whatever post processing app you use.
I was going by the camera produced JPEGs to try and remove post processing from the equation. That does bring up a good point, though, that I should check the meta data to see if anything looks different.
I am also wondering if IBIS can potentially hurt image quality if the lens specifications are not transferred correctly. From what I can find online it appears Nikon Z IBIS does not affect the Z axis of the sensor. But I can't find anything on whether yaw or pitch can affect the Z axis of the sensor. We are definitely talking about very small adjustments affecting image quality, as a single layer of clear packaging tape is enough to show a big difference in the past, with the Voigtlander 21mm.
From the hardware side, I have two options I can think of. I can un shim my other Megadap adapter so I can compare the two firmware versions side by side, or I can update the shimmed adapter to v1.38. Not sure which approach would be a better test.
The Zf IBIS can absolutely affect the Z axis if you have VR tied to the focus point, that's how it works, and it's why corner performance is severely impacted if your focus point is anywhere but dead center.
It's why I tell everyone to turn that off and forget it ever existed.
RoamingScott wrote:
The Zf IBIS can absolutely affect the Z axis if you have VR tied to the focus point, that's how it works, and it's why corner performance is severely impacted if your focus point is anywhere but dead center.
It's why I tell everyone to turn that off and forget it ever existed.
Interesting, sounds like I need to add testing with IBIS disabled to the list of tests.
Jman13 wrote:
There is no possible way for firmware to affect image quality on a manual focus lens. The CV 21/1.4 doesn’t do well on the Megadap (non +) because it’s too thin.
That was my understanding too, and was my experience with the 21mm. For the first year I had the Z8, I didn't use the 21mm with it because I was so disappointed in the results. My first night of testing with the Sigma 14mm left me totally confused with the results I was getting. I swapped the adapters around many times and kept getting the same results. My second night of testing focused on the two Megadap adapters, and again, I was surprised by how well the non shimmed adapter was performing compared to the shimmed adapter. I started looking for differences, and the only one I could obviously find was the FW version. After going down some Google rabbit holes last night, I decided to give the 21mm a try this morning on the non shimmed and shimmed adapter, and ended up being surprised that they were now performing extremely similarly. Which led me to post this topic to see if anyone has some insights or ideas.
@RoamingScott believes IBIS may be playing a factor, so I will definitely be testing with IBIS on and off the next clear night I get and will report back on the results.
----------------------
EDIT: I went back and checked my timeline, I posted this back in the beginning of June when I got the Sigma 14mm and 20mm:
I have not gotten a chance to try them for Astro yet, but I got to put them through some initial testing for landscape use. So far, I am quite happy with their performance on my Z8. I don't feel I am seeing a noticeable loss in IQ with my Z8 vs my a7riv. My gut is the Sigma lenses play much nicer with adapter tolerances than Voigtlander lenses since they can be focused passed infinity.
Which was the same outing I tried updating the Megadap adapter to 1.38 off the recordation, it improved AF and posted this:
I gave the Megadap v 1.38 firmware a try this weekend, and it worked much nicer with my 200-600 + TC 1.4 on my Z8.
It's not a confirmation, but the timeline adds up that the firmware may be playing a factor 🤔
RoamingScott wrote:
The Zf IBIS can absolutely affect the Z axis if you have VR tied to the focus point, that's how it works, and it's why corner performance is severely impacted if your focus point is anywhere but dead center.
It's why I tell everyone to turn that off and forget it ever existed.
The IBIS module does not and physically cannot move in the Z axis (i.e. towards or away from the focal plane). There is in fact no IBIS unit in any camera that moves in the Z axis and attempting to do so would be a disaster. The IBIS module can shift the sensor up/down, side-to-side, or rotate the sensor. That's it. It remains tightly constrained within the focal plane and perfectly perpendicular to the optical axis at all times.
Claims of IBIS moving in a "z-axis" or outside of the focal plane is a frequent misunderstanding often repeated across the internet whenever a new IBIS feature is introduced or claims of a new "axis" of correction. All new IBIS features and "axes" are about sensing and correcting for some particular camera body motion (e.g. yaw, pitch, shift) by projecting that motion onto the three degrees of freedom (aka axes) that every IBIS unit has. Namely shifting the sensor in two dimensions and rotating the sensor. No matter what IBIS feature is advertised the end result is only these three "axes" being controlled and the sensor remains in the focal plane at all times.
Sometimes additional metadata is necessary for the correction to be applied. In all cases the focal length is needed and this produces the standard "three-axis" IBIS (correcting for camera body pitch, yaw, and rotation). So called "five-axis" IBIS (additionally corrects for camera body translation up/down and left/right) requires knowledge of focus distance as well, but again the sensor itself only translates in two dimensions and rotates all within the focal plane - only three degrees of freedom.
The VR tied to focus point accounts for the fact that for wide angle lenses the IBIS correction is only "correct" for the center region of the frame. This is because of the effects of the rectilinear projection, especially on a wide angle lens. Linking VR to focus point makes the camera set the IBIS motion to account for this projection as you move away from the center. The side effect of this is that the opposite edge/corner is even more poorly corrected than it would be otherwise. But it has zero to do with the Z-axis at all.
As many have discovered the VR link to focus point is often not a trade off that is useful to them. Just because we focus on a particular spot doesn't mean we are willing to make the opposite edge of the image even more motion blurred to try to achieve perfection at the point of focus. Landscape is an obvious example where we may choose a focus point just where we know there is the best and most reliable contrast and lighting for focus but we actually hope for the entire image to be sharp.
So yes, VR tied to focus point turned off is probably a good default setting for everyone unless they have a very specific goal. But no, it has nothing to do with the "Z-axis" and no IBIS module has any capability to move outside the focal plane. They are all carefully designed to constrain the sensor within the factory calibrated focal plane and all motions occur within this plane.
My bad, I'm using Z in terms of sensor rotation, which classically is not the Z axis of forward and backwards. It's easy to understand why the point on the sensor opposite the focal point would be smeary if you're stabilizing for rotation around a far edge and tying the VR to one edge of the sensor.
In my head, X and Y are simple side to side and up and down movements of the sensor.
kwalsh wrote:
Sometimes additional metadata is necessary for the correction to be applied. In all cases the focal length is needed and this produces the standard "three-axis" IBIS (correcting for camera body pitch, yaw, and rotation). So called "five-axis" IBIS (additionally corrects for camera body translation up/down and left/right) requires knowledge of focus distance as well, but again the sensor itself only translates in two dimensions and rotates all within the focal plane - only three degrees of freedom.
This is exactly the type of information I was looking for. A lot of diagrams make it appear that for 5 axis stabilization, the sensor can pitch and yaw, but I was thinking, wouldn't that affect focus and AF.
RoamingScott wrote:
My bad, I'm using Z in terms of sensor rotation, which classically is not the Z axis of forward and backwards. It's easy to understand why the point on the sensor opposite the focal point would be smeary if you're stabilizing for rotation around a far edge and tying the VR to one edge of the sensor.
In my head, X and Y are simple side to side and up and down movements of the sensor.
Ah, makes perfect sense! Yeah, terminology is really confusing without figures to go along with it and conventions often aren't as conventional as we'd hope. If I'd thought about it more I probably would have realized what you meant by Z-axis. Well, anyway, at least other folks reading won't get confused now! Thanks for taking the time to clarify.
Also, I'm not sure the effect is actually that it is trying to rotate around the focus point, which is to say that they are changing how they implement correction for camera body rotation (roll) based on the focus point - though that would be a fun thing to test.
What Nikon has been illustrating instead is that they are altering how they correct for camera body pitch and yaw by taking into account the rectilinear projection of the lens and how the computed IBIS shifts (not rotations) need to be changed to account for that projection when the subject is not centered. This would also have the effect of making IBIS corrections worse on the opposite side of the image from where the focus point was.
Of course often times what the camera manufacturers illustrate in their videos is not the entire story as to what is happening in the camera, so at least for me there is some uncertainty as to exactly what all they alter when the VR is linked to the focus point. Regardless, there are multiple mechanisms that would make other parts of the image worse off than keeping IBIS optimized for the center of the image.
So the mystery continues. I was able to produce the same results, but no insight into what might be causing it. Last night I tested my Megadap adapter with the older firmware again. I tried turning off IBIS, vignette correction, distortion correction, and diffraction compensation with no noticeable change in corner performance. I swapped to my adapter with v1.38, and the corners improved.
I am thinking that at this point, my best bet is to update the adapter to firmware v1.38 and see if it makes any difference. I will need to update it anyway if I want to use it with my 200-600. As v1.38 noticeably improves the AF performance of the 200-600. If updating the firmware improves corner performance, then there must be some magic going on in the firmware, even if it does not make logical sense. If it does not improve corner performance, then maybe I just got lucky and something mysteriously happened, like my adapter stretched out from use.
I updated my second adapter to v1.38 about a week ago, and didn't notice enough of a difference to make any conclusions.
But it's been bothering me a bit, why do I feel like my Megadap adapters are performing differently than when I went deep down the adapter rabbit hole last winter? Sure, the new firmware may have improved AF, MF (focus by wire), and IBIS performance, which will all help IQ a bit. But I feel like what I am seeing is more than just a bit different. Which then got me thinking...
Is it possible that thermal contraction/expansion is at play? Metal does contract in the cold and expand in the heat, and we are talking about less than a mm of distance affecting IQ. I couldn't find anyone talking about temperature affecting lens adapters, but I was able to find people online saying that temperature changes were enough to affect their lenses focus in a single night. So, in theory, I would think temperature could affect lens adapters too.
