DoF planning with downsampled output

Esteemed colleagues, and in particular, Bart:

Imagine we have a photographic task, the sole "deliverable" of which must be an image 800 × 600 px.

My plan is to:

• Use a camera with a sensor 4000 × 3000 sensels in size

• Frame the shot to be suitable for the deliverable without cropping.

• Take the 4000 × 3000 px JPEG output.

• In post, take the entire frame and downsize it to 800 × 600 px. This is the deliverable image.

In the scene, I have principal objects at distances from 9 m to 11 m (based on my chosen camera position). I use a focal length that will provide the proper framing.

The planning challenge:

My objective with respect to DoF is an "outlook B" objective: the nearest and farthest of the principal objects should be imaged with a sharpness that is "not significantly degraded" from the sharpness for a object at the perfect focus distance.

The questions:

• At what distance should I focus the camera?

• What is the largest aperture that will allow me to meet my objective?

I assume that diffraction considerations are not in the picture.

************

My current approach would be as follows:

Background

We often find (in situations where downsampling is not involved) that using a COCDL of 1.0 times the image pixel pitch (on the sensor) in DoF calculations typically results in a situation that we can describe as: objects at the calculated field limits are imaged with a sharpness that is "not significantly degraded" from the sharpness for a object at the perfect focus distance.

So

Picking up on that, I choose a COCDL of 1/600 the picture height (on the sensor) for DoF calculations , from which (based on the needed near and far distances) will come the maximum usable aperture and then the desirable camera focus distance.

************

Now:

• Is their a fallacy in that approach?

• If so, what is it?

• If so, how can we proceed in a practical way to get a result that is comparably "valid" to the one I described for the "not-downsampled" situation.

Best regards,

Doug

My quick reply is that objects have different extents of smooth areas or detailed textures, so that your picture might demand different apertures for each content "class". In fact, sometimes you might need focus stacking to be satisfied with the detail.

Also, It's hardly ever needed to have focus so clear in street photography over such a large distance, from 9-11 meters. For people, gestures, style and interest dominate out decision-making. However, in landscape pictures with flowers, example in the foreground, fine grasses with white swaying blooms in the mid field and detailed trees in the distance, nothing but a tilt-shift lens can get what one must have in a single shot anyway.

So your question, perfectly valid as an engineering challenge, ignores esthetic sensibilities for subjects we'd want to photograph, except perhaps for forensic, espionage, engineering or insurance needs.

Asher

Hi, Asher,

Asher Kelman said:

My quick reply is that objects have different extents of smooth areas or detailed textures, so that your picture might demand different apertures for each content "class". In fact, sometimes you might need focus stacking to be satisfied with the detail.

Also, It's hardly ever needed to have focus so clear in street photography over such a large distance, from 9-11 meters. For people, gestures, style and interest dominate out decision-making. However, in landscape pictures with flowers, example in the foreground, fine grasses with white swaying blooms in the mid field and detailed trees in the distance, nothing but a tilt-shift lens can get what one must have in a single shot anyway.

So your question, perfectly valid as an engineering challenge, ignores esthetic sensibilities for subjects we'd want to photograph, except perhaps for forensic, espionage, engineering or insurance needs.

Click to expand...

Well said.

And of course it applies equally well in the matter of DoF "planning" when the deliverable is to be at actual camera resolution.

Note that my question was about an approach to adapting a certain widely-followed concept to a "different" context, and was not intended to endorse the concept as unilaterally, or even broadly, valid in its own right.

Best regards,

Doug