Zeroed the C35 on another couple guns today and I did need to break out the foil squares for the targets due to conditions. That said, everything went well. Today I added zeroes for a 16" lightweight 6.5 Creedmoor AR as well as a Ruger Precision Rimfire in 22 LR.
For the Creedmoor AR, despite the somewhat jarring vibration of that gun, there was absolutely no image cutout or any other issue. It doesn't have a ton of recoil, but that huge BCG and mid-length gas on a high pressure cartridge certainly cause it to move around.
For the RPRR 22 LR, I got it dialed in no problem and noticed a pretty interesting thing. With the 3-18×50 in 1.26" rings on a 30 MOA base (C35 is on zero MOA rail section on handguard), I could actually see "around" the clip-on if I got my eye in just the right (wrong) position. On the C35 there are some gaps to each side of the thermal/mount junction which let just enough image get through so I could switch from thermal image to substantially dimmed visible spectrum view just by moving my head placement. I've owned a lot of clip-on devices but have never experienced this phenomenon before. Based on the fact that I won't be using the C35 much during the day, this added "capability" is mostly useless and tells me that I am at the ragged edge of how low in the image I can look through with the day optic and still get a usable view.
An additional note is that, across the three platforms I've used it on, the difference in image location once zeroed barely changes at all. It wound up being 2 increments (~2 MOA) max spread. I have yet to move more than 2 MOA from dead center factory position (0,0 on X,Y axis) to obtain a good zero. This tells me that the image must be fairly close to true unity. I also measured an 8 mil height telephone pole with just the day optic and then again with the C35 to estimate how well my dope will translate to use with the C35. Within the pixel resolution of the C35, it appeared the 8 mil pole remained 8 mils. This, again, tells me that the image must be pretty close to unity. After all, if it wasn't approximately unity, drop and wind dope would be rendered largely useless. I will have to devise a better, more precise, experiment to get a solid verdict here but it looks good so far.
TL/DR: It still works and I still love it