D800 - focus under incandescent light

Started Feb 24, 2013 | Discussions thread
mdg137
Regular MemberPosts: 154Gear list
Like?
Re: This is common to many DSLRs....
In reply to Truman Prevatt, Feb 27, 2013

Truman Prevatt wrote:

The focal length of a given lens varies with the wavelength. It is pretty clear for anyone ever doing IR photography where you focus using visible (which is basically how you focus) light and the lens has a red dot so you can offset the focus for IR. You can run this experiment for yourself.

http://academia.hixie.ch/bath/focal/home.html

The reason is simple the refractive index decreases with with increasing wavelength hence blue and UV light focuses farther than does red. That is a physical fact. It has a name, chromatic aberration. Lens designers work their buns off to eliminate it but it is a physical fact and it cannot be totally eliminated because at the end of the day red light has a greater index of refraction than blue in the same glass.

In a strong white light source, e.g. daylight it's not an issue since the wide spectral response of white light will drown it out. However, when you have sources other than while light the effect will be magnified. To build the proper AF algorithm, they algorithm would need to know the spectral distribution of the reflected light. That is a difficult task.

Here is the spectrum of various light sources.

http://www1.assumption.edu/users/bniece/spectra/lightsources.html

As can be seen the spectrum of incandescent light source is very low on the blue end and very high on the red end with not a lot of green. Depending on the IR filter used, there is significant IR radiation in incandescent light source (after all it is generated by heating a filament well duh).

If the autofoucs algorithm had this spectrum, it could focus fine. However, at the end of the day most light is not just single source especially indoors. First there is artificial lighting which could be a combination of halogen, incandescent, florescent (whose spectrum tends to be concentrated in discrete frequencies see source above) and the different types of bulbs (daylight , cool white, warm white, etc.) the frequencies are different and some filtered and reflected sunlight (which take on the spectrum of the reflected surface). Another words the lighting indoors is quite complex, a combination of mostly red, components with strong spectral components and components of strong reflected white light.

To make matters more interesting, different areas in the scene will be illuminated by different mixtures of the light sources so have light of different spectral responses illuminating them. The effective focal length of the lens will be slightly different for different objects in the scene depending on how they are lit.

It is pretty simple physics. If the AF algorithm knew the exact spectral composition of the light in your AF area - could compensate for it just fine. That capability is several years down the road - probably after I a pushing up daisies.

At the end of the day it is up to the photographer to understand how phase matching autofocus works, how the spectral composition of light changes (particularly indoors in complex scenes) and understand how to use the tools at his disposal. In some cases, contrast detection or even manual focus using live view will be the preferred focusing option - the one that will give repeatable consistent results.

At the end of the day the photographer is the brains behind the camera. The camera is a tool not a robot to replace the brain of the photographer. Sure there may be some QC issues but from a lot of what I have seen, some more appreciation of the limitations of the automation in the cameras and approaching the camera as a tool rather than an automatic picture taking robot would produce better results.

Joe Porto wrote:

I first noticed color temperature effecting focus with my D7000, and later found my D3200 to do the same. This is NOT a D800 defect. It is more the nature of phase detection. I'm sure it's very complicated, but the simplest example is how a prism can divide various color temperatures to come out at different angles....phase detection also works by converging light from different angles.

Perhaps certain cameras can be more or less sensitive depending on the strength of their IR and UV filters.

-- hide signature --

Truman
www.pbase.com/tprevatt

I dont think any of us dispute the reality of the physics involved, but rather still ask the question, "If this is an issue beyond solving, then why has Canon managed to at least mostly mitigate it?"

-- hide signature --
Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow