Re: Sensor and lens tilt and how to DIY adjust your equipment
realVivek wrote:
The APO Lanthar 50/2 (and the 110/2.5) are sold with a infinity range and with no hard infinity stop like most lenses. There could be a reason for that. It is possible that the infinity mark shifts with temperature due to all the APD elements in those lenses.
Thermal deformation and camera flange distance setup is the reason why you would rather not have an infinity stop in most cases. I've had plenty of problems with infinity hard stop lenses calibrated at the factory to stop at infinity, especially painful was with the Loxia 25, failing entirely to focus at infinity, rather at some 100-120 feet away, before my camera was shimmed, so in the factory state. If you peruse Roger Cicala's flange distance article, you'll see how much the Sony cameras flange distance varies. About 100 microns of variation, excluding the outliers. This is not insignificant. Typically the lens infinity focus stop is calibrated to within 10 microns precision, so there is little meaning to that when the cameras vary by an order of magnitude more.
You'd want your lenses to focus at least a little past infinity to avoid this problem.
It is unclear to me if the lens tilt was detected on camera with a skewed mount or with a “perfect” mount.
If you are referring to my V50 that I adjusted, the lens tilt was determined only after the camera was straightened. It is pointless to try and detect lens tilt on any arbitrary body, because you cannot assume your camera is straight.
Also, generally speaking, the mount, if you are referring to the front mount, is not skewed unless it is damaged (the sensor is the one that is skewed relative to it) and the bayonets are CNC machined and thus reasonably consistent until one wants to exceed the manufacturer's tolerance for parallelism, like we were forced to, since we needed optimal performance at f/2.8 of the most demanding subject matter there is. Very few things in the world are perfect. There isn't a "perfect" mount, as mentioned in the article, even the specially made $8-10k mounts for OLAF at Lensrentals aren't machined to under a single micron precision in 40mm of runout.
For perspective, Canon's old spec for flange distance variation is 60 microns. I believe sensor parallelism can be routinely achieved to under 30 microns, with possible outliers. Some assembly methods can be even better, but we are not privy to how the factory chooses to do this. Bayonet CNC machining variation should not exceed 30 microns and could be a lot less. Lenses can go way above 200 microns for tilt and still be considered well-within 'factory spec'.
Sep 06, 2020 at 08:05 PM
Previous versions of inksandpaper's message #15339434 « Sensor and lens tilt and how to DIY adjust your equipment »
