My 60x/1.40 is faulty and how to fix a tilted focus
The last time I tried to buy a high end objective was a deception.
Olympus UPlanApo 40x/1.00 Oil Iris ∞/- objective
It was an Olympus UPlanApo 40x/1.00, an incredible objective and a very good candidat to make fast panoramas with a high level of details.
Unfortunately, after receiving it the image quality was just garbage for no obvious reason.
After some research and help from the microbehunter forum, we deduced that the objective was totally faulty.
The UplanApo 40x/1.0 in action.
My 100x/1.25 for comparison.
Something was very off with the focus.
PlanApo objectives are very precise multi-lens systems, and any small misalignment can cause serious problems to the image quality.
It probably got misaligned during shipping?
Sadly, it’s almost impossible to fix.
Fortunately, I was able to return it and get a refund.
What about the 60x/1.40 objective ?
October 2024 update: the tilted focus is not caused by the objective but by a slight misalignment of the microscope stage.
This was very hard to see on other objectives (the 100x oil have a too small field of view and low magnification objectives have a large depth of field). The radial aberrations are caused by two factors: the 0.5x reduction lens I’m using, and the fact that the 60x is off-centered (it’s still faulty, but less than expected).
At a first glance everything was looking perfect.
Only some slight out of focus areas, probably caused by the sample not properly leveled, or a missaligned camera.
So, I designed a camera mount to get everything straight and rigidly held, but also a kinematic XY stage to get a perfect leveling for the die.
The problem is still here.
If I rotate the camera, the out of focus area is still at the same area on the chip (not the same area on the screen).
Also, everything is fine with all of my other objectives.
The bottom of the image is clearly out of focus.
Fortunately, in my case an uneven focus is not something that cannot be corrected.
To create my multifocus panoramas I’m taking a lot of different focus points per tile.
So, I added a new feature in my pipeline to use a depth map to mix different images to restore an even focus over the full field of view.
The hand drawn depth map to fix this 60x objective.
The depth map was made by trial and error.
Here the result of a corrected single frame.
Still not perfect on the bottom, but probably good enough for stitching with at least 30% overlap.
Tradeoff
Obviously, not everything is perfect. Recombining multiple focus requires more focus points, and so more images.
This means longer capture times. In my case probably about 5 sec / tile, which means about 1h30 of time lost for 1000 tiles.
It’s not nothing, but when you consider how much time this objective saves me, it’s not a big deal.
Conclusion
In both cases, the objectives were advertised as fully functional.
This may be true, maybe damage occurred during transport, maybe not. Actually, the packages were quite secure.
After talking with people from microbehunter, my understanding of the situation is that the first objective may have been disassembled and reassembled without taking into account the original position of the lenses.
The 60x objective shows a fingerprint trace in the rear lens, but it’s unlikely to place a finger here because of the objective rear aperture… ?
Anyway, maybe the price was “low” for a reason. It’s hard to tell.
At least the 60x was worth it. And it seems possible to fix it somehow.