# Focus Desk

> EQUIVALENT FOCAL LENGTH IS NOT EQUIVALENT. Matching a frame across two sensor sizes needs a
> different lens, and the different lens has a different depth of field at the same stop. A 50 mm
> T2.8 on full frame frames like a 35.9 mm on Super 35 - and to match the depth as well it has to
> be T2, not T2.8. Five lanes over one shoot, plus a free browser-side engine computing depth from
> the real equations and reporting it as the distance an actor can move.

URL: https://focus-desk.skillsafe.ai/
API: https://focus-desk.skillsafe.ai/api.html

Derived from dexhunter/seedance2-skill (https://github.com/dexhunter/seedance2-skill), MIT: a skill
for writing effective prompts for multimodal AI video generation. Not affiliated with or endorsed
by dexhunter.

## The one ratio

```text
same frame   ->  focal length x (destination diagonal / source diagonal)
same depth   ->  the stop moves by THE SAME RATIO
```

A SMALLER FORMAT NEEDS A WIDER STOP to match the depth. Say the direction slowly; it is the half of
the sentence nobody says and it is easy to say backwards. Read the other way round, which is how it
is usually met: a T2.8 lens on Super 35 is only a T3.9 in full-frame terms, which is why fast Super
35 glass is less fast than it sounds.

AND IT RUNS OUT. Matching a full-frame T1.3 on Micro Four Thirds would need a T0.7, which is not a
lens. That shot cannot be matched at all - only approached.

## The circle of confusion is a delivery decision

EVERY DEPTH-OF-FIELD NUMBER RESTS ON AN ASSUMPTION ABOUT HOW THE PICTURE WILL BE WATCHED, AND
ALMOST EVERY CALCULATOR HIDES IT.

A circle of confusion is the largest blur that still reads as sharp, and how large that is depends
entirely on how big the picture will be and how close somebody sits to it. It is NOT a property of
the camera, the lens or the sensor.

The default in almost every tool is the traditional 1/1500 rule, derived from an 8x10 print at
arm's length - and almost nobody delivers an 8x10 print. A cinema screen is twice as strict, so a
shot the calculator calls a comfortable 130 mm has 65 mm on the screen it was made for, and 65 mm
is under the point where it needs somebody whose whole job is the focus.

NOTHING ABOUT THE CAMERA CHANGED. THE ASSUMPTION DID. It is also why footage that was sharp in 2014
is soft now: it was judged on an HD screen and is being watched at 4K.

## The equations

```text
H     =  f^2 / (N c) + f            the hyperfocal distance
near  =  s (H - f) / (H + s - 2f)
far   =  s (H - f) / (H - s)        and INFINITE once s >= H
```

The common approximation `2 N c s^2 / f^2` is fine at portrait distances and wrong close up, and
close up is exactly where a focus puller needs the number.

REPORT THE TOTAL IN MILLIMETRES. What decides whether a shot is a setup or a focus pull is how far
the actor can move, and that is a distance in millimetres rather than a pair of limits in metres.

THE ONE-THIRD/TWO-THIRDS RULE IS TRUE AT EXACTLY ONE DISTANCE for any given lens and stop. Close up
the split approaches half and half; far away almost all the depth is behind the subject.

PAST THE HYPERFOCAL THE FAR LIMIT IS INFINITY, not a large number. Reporting a very large figure
there looks like a measurement and is not one.

## What this page cannot do

Everything here is GEOMETRIC OPTICS over numbers you supply. It is exact about the model it uses and
the model is not the whole of a lens.

No diffraction - stopping down past about T11 on a small format loses sharpness these equations will
happily promise. No focus breathing, no focus shift, no field curvature. And a lens marked 50 mm is
rarely exactly 50 mm.

So give the figures and say what they are: a well-founded starting point that a focus puller with a
tape measure will correct. NEVER PRESENT A DEPTH OF FIELD AS A MEASUREMENT OF A REAL LENS.

## The sheet

```text
SHOOT
format  | s35
viewing | cinema
compare | ff

SHOTS
s1 | 32 | 4 | 3   | the two-shot
s2 | 85 | 2 | 1.5 | the single
```

A shot row is `id | focal length in mm | stop | distance | description`. Distances are metres unless
a unit says otherwise - `1.5`, `1200mm`, `50cm` and `4ft` all work. Stops may be written `2`, `T2`
or `f/2`. `viewing` is required and has no default. `compare` is optional.

A MISSING NUMBER STAYS MISSING. A shot without all three of focal length, stop and distance is in no
calculation and the page says so.

## The lanes

- `plan` - picks the format and the stops before anything is shot
- `check` - says what depth every shot actually has, and which finding matters
- `depth` - how far the actor can move, and which shots need a dedicated puller
- `equivalence` - matching another camera, both halves of the word
- `deliver` - sorts findings into as planned, change the setup, change the kit

## What it does not do

It measures nothing and focuses nothing. Every figure is geometric optics over numbers you supply,
so it is a starting point a tape measure will correct.