Hey everyone,
I like to use the Harrison & Harrison black dot filters. My Canon 5d3 has a terrible time focusing with the filter(s). Even though I see focus confirmation in the VF, it focuses 6-8" in front of the focusing point. It does it with my Canon 28-70 2.8L and my Sigma 85 1.4 xxx lens. Without the filter, the focus is perfect on both lens. The focus, with the filter, works better in very bright light. Not in average room light.
Can anyone tell me scientifically-technically, what is happening? (yes, I know the black dots are interfering somehow, just don't know why...)
Hi Michael,
Sorry to hear you're having this issue, but it's interesting as well (and a good warning for others, something to consider).
I know about them but I have no personal experience with the particular filter type. According to the information available on the internet, these filters require aperture compensation of approximately 1 stop, regardless of the degree of diffusion. That by itself would not be enough to cause low light level issues for the focusing system of the 5D3, except for the darkest possible scenarios that the AF system is designed for.
So that probably leaves the pattern itself that somehow interferes with the phase detection principles. I would not have expected that to be an issue except for (very) short focal lengths, because the dots (are they round or have another shape?) would be more in focus than on long focal lengths due to DOF. To give an idea, on my fish-eye lens I can even see relatively well focused dust spots on the front lens element.
I know they are intended for adding various degrees of diffusion to an otherwise well focused image, somewhat similar to a Softar. A Softar uses dispersed lenslets to add defocus blur, which should not lose much light, but it adds both a white veil, and a black veil, to the image which therefore loses a lot of overall contrast as well. The Harrison & Harrison handmade dot texture screen is supposed to maintain better contrast, because mostly a dark diffusion is added to the image, with a much nicer effect than e.g. black netting.
The issue though, is that it is not just diffusion because the dots are (I presume) not significantly transparent. They just block certain rays from entering the lens and do not contribute to the focused image but do darken the image. In addition, there is probably some diffraction at the edges of the dots if they are sharp edged, and that will cause the very complex blur to be added.
Due to, what I presume to be, the complexity of the diffraction blur, it will probably have an overall Gaussian blur effect on the image (but combined with a well focused image). To see if this compound effect affects maximum resolution enough to throw-off the phase dectection capability would need to be analyzed based on a well focused (Live View / Contrast detect) image of
a test chart that will tell you if the maximum resolution is compromised. Just compare an image with, and one without the filter, and see if or how much the extinction resolution is changed. Maybe it is, and if so then it may be too much for the phase detect system to handle?
So, my best guess at the moment is that either the dots are too well focused (mainly a short focal length, short focus distance issue), or the extinction resolution of the optical system is degraded too much for the Phase detect system to get a solid lock on the subject.
Cheers,
Bart
