Break/Fix: Unexpectedly Soft Images and Wrong Reciprocity Corrections

Seven negatives, each spoiled on purpose, each with its exposure-log row intact. Your job is to name the fault from the evidence rather than from a guess, and the method that gets you there is worth more than any of the seven answers.

The situation

A student comes back from an afternoon with the modular camera and eleven exposed sheets. Six are usable. Five are not, and they are not wrong in the same way: two are soft, one is thin, one is flat and grey, and one has a bright wedge across a corner. The camera passed its commissioning tests three weeks ago, and nothing has been rebuilt since.

Every one of the five faults on this page can be produced deliberately in an afternoon, and you should produce them, because a fault you have made yourself is one you will recognise for the rest of your life. The faults are:

1. A pinhole with a burr left on it.
2. The same nominal hole in a plate too thick for it.
3. A subject inside the near limit.
4. A sheet that was not flat in the back.
5. A shiny interior, made by lining one wall with kitchen foil.
6. A shutter leak, made by lifting the blade a fraction of a millimetre out of its channel.
7. An exposure corrected with the wrong material’s reciprocity data, and its three relatives — the correction applied twice, the extension factor forgotten, and a daylight paper speed used indoors.

⚠Level A, and what the hazards actually are

Nothing on this page introduces a new substance. Re-processing uses the same working-strength paper developer, citric acid stop at 1+19 and ammonium thiosulfate rapid fixer at 1+4 as every other practical page in this part, with the same controls: nitrile gloves, eye protection, one pair of tongs per tray never swapped, and ordinary room ventilation rather than extraction, because none of the three baths generates a vapour that would need it.

The one manufactured hazard is making the faulty pinholes, which uses the needle, block and shim precautions of the pinhole lab: the plate held against a block rather than in the hand, eye protection while a needle is being pushed through metal, and offcuts collected in a closed tin.

Used fixer is silver-bearing and is collected into the labelled container Part II established; local regulation governs the rest, and it differs between authorities even within the United Kingdom.

📓These are drawings, not photographs, and the course says so

The Visual Diagnostic Atlas this page feeds is specified to carry only licensed images, and this course has not yet produced and photographed the fault set. Every signature below is therefore drawn from the physics rather than reproduced from a negative, and the drawings say so.

That is a real limitation and it cuts one way: a drawing can show you where to look and what shape to expect, and it cannot show you the texture of a real fault. The remedy is the one this page is built around — make the faults yourself, keep the negatives, and photograph them. A student’s own fault set is a better teacher than anybody’s illustration, and the atlas entries below are written so that yours can be dropped straight into them.

Alternative route

This is a diagnosis exercise, and diagnosis is the part that does not need a darkroom. Every one of the seven cases below is worked here with its evidence, its confirming test and its fix, so the reasoning can be run in full against the negatives and the logs you already have.

No darkroom. Split the page in two. Four of the seven faults are diagnosed on the bench, not in the tray, and need only a dark cupboard at night rather than a processing space: the burred hole under a loupe and by its laser pattern, the plate thickness by measurement and arithmetic, the shiny interior by a torch inside the closed body, and the shutter leak by the same torch from outside. Reciprocity and its three relatives are diagnosed from the exposure log alone and need no camera at all. Only the sheet that was not flat and the subject inside the near limit want a processed negative, and both can be deferred to a borrowed session.

No deliberately spoiled sheets. Work the cases on negatives you already have. Most readers arrive at this page with at least two of these faults already committed and unrecognised in the commissioning batch, which is the better material anyway: a fault you produced on purpose comes with its own answer, and a fault you have to find does not.

No laser pointer. The burr shows under a reticle loupe or a flatbed scan at 2400 dpi optical, both of which the pinhole-making session already sets up. The laser is the fastest test, not the only one.

What cannot be substituted is the corrected log. Whatever you diagnose and however you diagnose it, the record of the wrong number struck through beside the right one is the artefact this page exists to produce.

The symptoms

Five kinds of thing go wrong with a pinhole negative, and they are worth naming before any of the seven cases, because the naming is most of the diagnosis.

Softness — the picture is there but no edge is crisp. It has at least seven distinct causes on this page alone, and they are separated by how the softness looks and by where on the frame it is.

Smear — a specific direction is soft and the perpendicular direction is not. That is movement, and it is never optics.

Density wrong — the whole scale sits too high or too low. That is exposure, or processing, and the rebate or the strip under the paper stops tells you which.

Contrast wrong — the scale is compressed, most often at the bottom, so the shadows are grey and lifeless. That is non-image light: flare, a leak, safelight fog, or an exhausted developer.

A local mark — a wedge, a band, a patch. That is a leak, a reflection, or handling.

!The method: log first, negative second

The instinct is to hold the negative up to the light and start theorising. Resist it for two minutes, because the log will usually have told you the answer already, and a negative read with a hypothesis in your head is a negative that will agree with you.

Read in this order.

1. The log row. Which plate, which focal distance, which exposure, what the light was, what moved. A soft negative made with the 0.15 mm plate at 120 mm is diagnosed before you look at it.
2. The unexposed control, if the session had one. It separates everything that happened outside the camera — the box, the safelight, the developer — from everything that happened inside it. No control, no clean separation, which is why every session in this part carries one.
3. The parts of the sheet that were not exposed: the strip under each paper stop, or a film’s rebate. Their density is base plus fog and it is independent of the exposure entirely.
4. The whole frame, for pattern. Is the fault everywhere, at one edge, in one corner, on one object?
5. One high-contrast edge under a loupe, in two directions. Only now, and only for softness.
6. Then form a hypothesis, and only then look for the evidence that would kill it.

Steps 1 to 3 cost two minutes and settle most cases. Step 5 is what people do first, and it is the step that most often misleads, because every kind of softness looks like every other kind at low magnification.

From 'the negative is soft' to seven causes, with the test at each branch

1. 1Same in every direction? — the single question that separates optics from movement — look at two perpendicular edges under a loupe
2. 2No: movement — camera (all smeared alike) · shutter jar (faint whole-frame ghost) · subject (only some things smeared)
3. 3Yes: optics, whole frame — oversized hole (plain flat blur) · undersized hole (glow with a skirt) · burr (lopsided, same direction everywhere)
4. 4Yes: optics, part of the frame — near limit (one object) · thick plate (hard-edged dark corners) · buckled sheet (denser sharper centre, bowed lines)

Drawn from the physics; the course has not yet photographed the fault set. Every terminal branch carries a confirming test that can be run in an afternoon, and the tree is read top down, never guessed at from the bottom.

Evidence to collect

Before any of the seven cases is opened, gather these. Every one is cheap and several of them are free.

From the log. Plate letter and its measured diameter; measured focal distance and the frame stack; effective f-number; the material and the effective ISO used; the metered light and where the number came from; the computed time, the reciprocity correction and the sheet it was taken from; the time actually given; what moved; and the processing.

From the sheet itself. Whether the fault crosses the strips that were under the paper stops. This single observation separates two whole families: light that came through the pinhole stops dead at the edge of the image circle, so flare, a hot spot and the picture itself all leave the covered strip clean; light that came in any other way, or that reached the sheet outside the camera, does not.

From the control sheet. If it is clean, the box, the safelight and the developer are all exonerated and the fault is in the camera or the exposure. If it is fogged, nothing else can be interpreted until that is fixed.

From the plate. A scan at known resolution and a laser diffraction pattern, both from the pinhole lab’s method. A clean round hole gives concentric rings; a burr gives a fan or a lopsided pattern. And a micrometer or feeler gauge on the plate thickness, which is the number people never record and then need.

From the camera. A torch inside it in a dark room, looking for anything that shines. A straight-edge across the loaded back, looking for a bow. And the two commissioning tests re-run: shutter closed in full sun for thirty minutes, then shutter open with the pinhole capped.

From a re-shoot. One frame of a static, high-contrast subject, on a tripod, in still air, at a computed exposure, with everything logged. Half the cases below are settled by whether the fault survives that.

Hints

In increasing order. Stop as soon as one of them is enough.

Hint 1. Look at the log before the negative, and specifically at the plate letter and the focal distance together. Most softness on a first camera is a hole that is wrong for the distance it is being used at, and that is arithmetic rather than diagnosis.

Hint 2. Take one high-contrast edge and look at it under a loupe twice: once along the edge and once across it. If the two look different, you are finished — it is movement, and no optical cause can produce a direction-dependent blur, because a hole has no preferred direction.

Hint 3. Ask whether the whole frame is affected or only part of it. Whole-frame faults live in the aperture; part-frame faults live in the geometry — the near limit, the plate thickness, the flatness of the sheet.

Hint 4. For anything to do with contrast rather than sharpness, find a place on the sheet where the picture should be at its darkest, and ask whether it got there. Non-image light shows itself in the shadows first, always, because that is where the picture delivered least and a constant addition is therefore proportionally largest.

Hint 5. For a thin or dense negative, compute what the exposure should have been from the log’s own numbers, in the order speed, aperture, extension, reciprocity, and compare with what was given. The answer is nearly always a factor of two, four or five and a half, and each of those has a name.

Hint 6. If the arithmetic looks right, check the source of the reciprocity correction rather than the correction. A number from the wrong maker’s sheet is a real number correctly applied to the wrong material, and it will survive every check except that one.

The diagnosis

Seven cases. Each gives the physical cause, the confirming test, the fix and the prevention, in a form that can be lifted straight into the diagnostic atlas.

Case 1 — The burred hole

What you see. The whole frame is soft, and the softness has a preferred direction that is the same everywhere in the frame — every point is smeared the same way, in the same amount, whether it is at the centre or in a corner.

The physical cause. Piercing thin metal pushes a lip of material up on the far side. That lip stands into the light path and makes the effective aperture something other than a circle: on one side the opening is a clean edge, on the other it is a ragged flap standing proud. The blur figure of an object point is the shape of the aperture, so a lopsided aperture gives a lopsided blur, and every point in the frame gets the same lopsided blur because every point sees the same hole. Wall’s 1912 Dictionary gives the instruction without the explanation — “the edges of the pinhole should be quite clean and free from burr” — and it has been true for as long as people have pierced holes in metal.

Why it is not movement. Motion smear is the same direction everywhere too, which is why the two get confused. The difference is that motion smears only what moved, or, for camera shake, smears the whole frame including things at different distances by the same angular amount — and it leaves an edge running parallel to the motion completely sharp. A burr blurs perpendicular edges as well, because the aperture still has width in that direction.

The confirming test. The laser diffraction pattern from the pinhole lab. A clean circular hole gives concentric rings; a burred hole gives a fan, a streak, or rings that are brighter on one side. The loupe on the plate itself, from the burr side, is the second check, and a scan at 4800 dpi is the third.

The fix. Re-make the plate. A burr can be abraded off with fine wet-and-dry on a hard flat backing, working from the burr side, and the hole re-measured afterwards, because abrading changes the diameter. If the hole was already at the small end, abrading will take it out of the band entirely.

The prevention. Pierce from one side only, part way, then reverse the plate and abrade; measure after every operation, not just at the end; and record the result in the register so that a plate that has been reworked is identifiable later.

Case 2 — The plate too thick

What you see. Not softness at all, on inspection: the edges and corners are dark, and the boundary is comparatively hard rather than the gentle gradient of cos-fourth falloff. It is read as softness because a corner with no light in it has no detail in it.

The physical cause. The tunnel effect. A hole in a plate of thickness t is a short cylinder, and looking into a cylinder from the side you see less of the far opening. The overlap closes completely at

θ_cut = arctan(d / t)

Tunnel cut-off angle

and beyond that angle no light reaches the film at all. This is true vignetting — a physical obstruction — and it multiplies the geometric cos-fourth falloff rather than replacing it.

The numbers, which are startling. A 0.30 mm hole in 0.5 mm steel — a perfectly ordinary shim thickness, and a plate somebody will make because steel is what was in the drawer — cuts off at arctan(0.30/0.5) = 31.0°. On 4 × 5 inch paper the corner sits at 34.1° at f = 120 mm and 58.4° at f = 50 mm. The corners are gone at every setting the camera has, and at the wide setting most of the frame is gone with them. The geometry lesson’s rule — keep t ≤ d/5 — puts the cut-off beyond 78.7°, which is past the corner of anything you are likely to build.

The confirming test. Photograph nothing but a large, evenly lit blank wall, filling the frame. The negative is then a map of the camera’s illumination with no subject structure in it. Cos-fourth falloff is a smooth gradient all the way to the corner; a tunnel cut-off has a visible boundary at a computable radius, f tan θ_cut. At f = 120 mm and θ_cut = 31° that radius is 72 mm, which is inside the 81 mm half-diagonal — so the corners are dark and the edge centres are not.

The fix. A thinner plate. Drinks-can aluminium is about 0.1 mm and brass shim is available at 0.05 mm; either is a better home for a 0.3 mm hole than 0.5 mm steel. If the thick plate must be used, counter-sink or chamfer the hole from both sides so that the cylinder becomes a short hourglass — which is what a laser-drilled pinhole in a thicker carrier does — and then re-measure.

The prevention. Put the plate thickness in the pinhole register beside the diameter, and compute θ_cut for every plate when it is entered. It is one line of arithmetic and it retires a plate before it wastes a session.

Case 3 — The subject inside the near limit

What you see. One object is soft and everything behind it is as sharp as the camera ever gets. The soft object is the nearest thing in the frame.

The physical cause. The blur circle is b = d(1 + f/u), and the bracket grows as u falls. At infinity a point becomes a disc exactly the width of the hole; at u = f it becomes two hole widths. There is no focus to lose, so this is not “out of focus” — it is the geometric blur getting larger, and it is the one softness on this page that is a function of subject distance.

Why a near subject blurs more: the same hole, two subject distances

The rays are computed from the element positions, so the inversion and the magnification here are the ones the numbers give. Nothing about this is a focusing error: a pinhole has no focus, and the near limit is a property of the geometry alone.

A worked near limit. At f = 120 mm with a 0.30 mm hole, the distant blur is 0.30 mm and the diffraction blur is 0.44 mm, so the total is 0.74 mm and geometry contributes 41 per cent of it. Bring the subject to 300 mm and the geometric part becomes 0.30 × (1 + 120/300) = 0.42 mm, so the total goes to 0.86 mm — 16 per cent worse. At 150 mm it is 0.54 + 0.44 = 0.98 mm, 32 per cent worse, and that is visible.

The confirming test. The one that settles it in ten seconds: is the background sharp? Optical blur from the aperture affects the whole frame equally; only the near-limit blur is distance-dependent. If the near object is soft and a wall thirty metres away is at the camera’s usual standard, this is the answer and nothing else on this page is.

The fix. Move back, or accept it. A smaller hole reduces the geometric term in proportion, but the diffraction term rises in inverse proportion, so below the optimum you lose more than you gain.

The prevention. Compute the near limit for each configuration once and write it on the frame in white pencil: the distance at which the total blur has grown by whatever fraction you decide is your limit. For the camera above, a 20 per cent limit puts the near limit at about 250 mm.

Case 4 — The sheet that was not flat

What you see. A denser centre, straight lines that are subtly bowed, and — here is the trap — a centre that is very slightly sharper rather than softer.

The physical cause, worked through rather than assumed. Everything about a pinhole scales with f. If the sheet bows so that the centre is 2 mm closer to the hole at f = 50 mm, that is a 4 per cent change, and three things follow. Illuminance goes as 1/f², so the centre receives about 8 per cent more light — a denser middle. Magnification is f/u, so the middle of the picture is drawn at a slightly smaller scale than the edges, which bows straight lines. And the blur circle is proportional to f, so the centre’s blur is 4 per cent smaller: the middle is fractionally sharper.

So a buckled sheet is a density fault masquerading as a sharpness fault, and diagnosing it as softness sends you looking at the plate, which is fine and will be found innocent.

The confirming test. A straight-edge across the loaded back before the dark slide goes in; and, on the negative, whether the density gradient is radial, centred on the middle of the sheet rather than falling towards the corners as cos-fourth does. A cos-fourth gradient is symmetric about the optical axis; a bow gradient is symmetric about the middle of the paper, and if your carrier is offset those are not the same point.

The fix. Thicker foam on the pressure plate, or a second leaf spring at the middle of the long side.

The prevention. Load one sheet, close the back, and check with a straight-edge through the dark-slide slot before every session that starts a new box — resin-coated paper from a freshly opened packet curls more than paper that has been in a camera all week.

And the related movement fault. A sheet that is not properly held can change shape during the exposure, popping from one curvature to another. That gives a faint complete second image, displaced by a small amount, over an otherwise normal negative — and it is a movement fault, not an optical one, so it belongs on the left branch of the tree.

Case 5 — The shiny interior

What you see. Correct overall density, flat grey shadows, clean highlights, and sometimes a broad bright patch near one edge. The veil stops at the strips that were under the paper stops.

The physical cause. Light that comes through the hole at an angle greater than the frame’s own half angle misses the film and lands on the interior walls. A matt black wall absorbs most of it; a shiny one returns it, and what returns is spread more or less evenly over the whole frame. A constant addition of exposure is negligible where the picture already delivered a lot of light and dominant where it delivered almost none, so the shadows lift and the highlights barely move: the toe compresses and the contrast of the whole picture falls. Part IV owns the latent-image side of this and Part XIII owns the curve; what belongs here is the consequence and where to look for it.

The hot spot, and the ghost. Two related marks come from specular rather than diffuse reflection. A bright, flat interior wall reflects a recognisable patch of light back onto the film, roughly the shape of the wall it came from and on the opposite side of the frame from the source. And an unblackened pinhole plate, whose inside face is a small mirror parallel to the film at distance f, returns light that has already reached the film and scattered, giving a faint, broad, displaced ghost of the frame’s own bright areas rather than a sharp second picture. This course has not photographed either, and describes both from the geometry; treat them as predictions to be tested rather than as observations.

The confirming test. The flare pair from the commissioning session: the same scene with a bright source just outside the frame, once with and once without a card flag on the source side. If the flagged sheet has cleaner shadows, the camera is adding flare. The second test is a torch inside the camera in a dark room: anything that shines is a candidate.

The fix. Matt black paint, then flocking or black felt on the two walls that face the film most directly, and a blackened pinhole plate — both faces, not just the outside.

The prevention. Test the flocking rather than assume it: two identical inserts for one wall, one painted and one felted, and one scene photographed twice.

Case 6 — The shutter leak

What you see. A band of raised density along the edge of the sheet nearest the shutter, which crosses the strip that was under the paper stop, on a sheet whose picture is otherwise normal.

The physical cause. A sliding blade that lies flat against the outside of the front panel rather than running in a rebated channel leaves a wedge-shaped gap at shallow angles. Light entering there does not go through the hole, so it forms no image; it illuminates whatever part of the film it can reach, which is the part nearest the gap.

The confirming test is already built: sheet 2 and sheet 3 of the commissioning session. Shutter closed, thirty minutes in full sun tests every path except the shutter; shutter open with the pinhole capped, thirty minutes in full sun tests the shutter and nothing else. A band on sheet 3 and a clean sheet 2 is a shutter leak and cannot be anything else.

The fix. Cut the rebated channel. Tape on the outside of a blade that is lying flat will not fix a gap that points at the hole, because the gap is on the inside.

The prevention. Re-run the two-sheet test after any work on the front of the camera, and after adding a frame — a stack that passes at 50 mm can leak at 120, because each frame adds two joints.

Case 7 — The wrong reciprocity data, and its three relatives

This is the case with real arithmetic in it, and the numbers below come from two makers’ published sheets so that the wrong answer is a real wrong answer rather than an invented one.

🧮One metered time, four ways of getting it wrong

The situation. A metered exposure of 10 seconds, on sheet film, at the camera’s f-number.

The right answer for FOMAPAN 100. FOMA publish a multiplier table rather than an exponent: at a metered 10 seconds the lengthening factor is ×8, so the correct exposure is 80 seconds.

The right answer for FP4 PLUS. ILFORD publish the relation T_c = T_m^P with P = 1.26 for FP4 PLUS. At 10 seconds, T_c = 10^1.26 = 18.2 seconds.

Now the four errors.

Error 1 — the wrong maker’s data. FOMAPAN in the holder, ILFORD’s exponent in the notebook. You give 18.2 s where 80 s was needed: a factor of 4.4, which is 2.14 stops underexposed. On a film with FOMAPAN’s latitude the shadows are gone. Reverse the mistake — FOMA’s factor applied to FP4 PLUS — and you give 80 s where 18.2 was needed, 2.14 stops over.

Error 2 — the correction applied twice. FP4 PLUS, corrected once to 18.2 s, then corrected again by someone who did not notice: 18.2^1.26 = 38.7 s, which is 1.09 stops over. This is the error that hides best, because every individual step was done correctly.

Error 3 — the extension factor forgotten. Metered and computed for the body alone at f = 50 mm, then both frames stacked to f = 120 mm to reframe, and the time not recomputed: (120/50)² = 5.76, which is 2.53 stops underexposed. The pinhole has no bellows factor, as the exposure lesson proves — but it does have an N that changed when f changed.

Error 4 — a daylight paper speed used indoors. ILFORD’s ISO 3 to 6 for MULTIGRADE RC is determined under a daylight illuminant, and the paper is blue-sensitive. Carry that number into tungsten light and the arithmetic returns thirteen hours for a still life, which the exposure lesson works through in full. Thirteen hours is not a long exposure; it is the arithmetic telling you the speed does not transfer.

How to tell them apart from the negative alone. You mostly cannot: 2.14 stops under and 2.53 stops under look much the same. The log is what separates them, and specifically the field that names the sheet the correction came from. That is why the exposure log has that field.

Four log rows, each with one wrong number, and how the error shows

1. 1Wrong maker — FOMAPAN 100, ILFORD P = 1.26 applied → 18.2 s where 80 s was needed: 2.14 stops under
2. 2Applied twice — FP4 PLUS, 10 s → 18.2 s → 38.7 s: 1.09 stops over. Every single step was done correctly
3. 3Extension forgotten — metered at f = 50, exposed at f = 120: (120/50)² = 5.76, 2.53 stops under
4. 4Daylight speed indoors — ISO 3 is a daylight figure for a blue-sensitive paper; the 13-hour answer is the warning, not the exposure

The ringed field in each row is the one the negative cannot tell you about. Drawn from published data in two makers' sheets; the arithmetic is worked in full in the callout above.

Four edge profiles, drawn: what each kind of unsharpness does to one hard edge

1. 1Hole too large: geometric — a straight ramp of definite width, the same in every direction; the hole is a flat-topped disc
2. 2Hole too small: diffraction — a narrow core with far-reaching tails — glow, not blur; same in every direction
3. 3Motion smear — ramp width depends on the direction of the edge: widest across the motion, zero along it
4. 4Burr smear — asymmetric, spreading further on one side, and the same way everywhere in the frame

Drawn to show the shapes rather than measured from negatives; the course has not yet photographed the fault set. The decisive comparison is panel 3 against panels 1, 2 and 4: only movement cares which way the edge runs.

The contrast cases, side by side

Four things lift the shadows and flatten a negative, and they are separated by two questions.

Cause	Does it cross the strip under the paper stop?	Is it on the unexposed control sheet?	Other signature
Flare from a shiny interior or a bright plate	No — it came through the hole, so it stops at the image circle	No	Worst with a bright source near the frame edge; a flag reduces it
A leak	Yes	No	Usually a gradient from one edge or corner, not uniform
Safelight fog	Yes, and it is uniform right to the cut edge of the sheet, because the sheet was out of the camera when it happened	Yes — this is the one that shows on the control	Grows with handling time; ILFORD warn it can show as a general loss of contrast before it shows as visible fog
Exhausted or oxidised developer	Yes	Yes	Low maximum black as well as lifted shadows; a fresh-developer reference sheet settles it

The control sheet does most of this work by itself. A sheet from the same box that was handled identically and never went in the camera carries the safelight and the developer and nothing else. If the control is clean and the picture is veiled, the fault is inside the camera. That is worth one sheet of paper a session for the rest of your life.

On the developer, this page says only what it can see: reduced contrast, a maximum black that is not black, and sometimes a faint overall stain. The mechanism — what oxidation does to a developing agent and why the effect shows first in the deepest tones — belongs to Part VIII and Part XXVIII, and is not anticipated here.

The fix

Summarised, one line per fault, so the table can be used at the bench.

Fault	Confirming test	Fix	Prevention
Burred hole	Laser pattern fans or is lopsided	Re-make or abrade and re-measure	Pierce part way, reverse, abrade, measure after each step
Plate too thick	Blank-wall negative shows a hard-edged dark rim at radius f tan θcut	Thinner plate, or chamfer both faces	Record t in the register; compute θcut = arctan(d/t) on entry
Subject inside the near limit	Is the distant background sharp?	Move back	Compute and label the near limit per configuration
Sheet not flat	Straight-edge on the loaded back; radial density about the paper’s centre	Thicker foam, second spring	Check the back at the start of every new box
Shiny interior	Flare pair with and without a flag; torch inside in the dark	Matt black, flocking, blackened plate both faces	Two-insert controlled test of the flocking
Shutter leak	Capped-pinhole sheet fogged, closed-shutter sheet clean	Cut the rebated channel	Re-run both sheets after any front-end work or a new frame stack
Wrong reciprocity data	Recompute from the maker’s sheet named in the log	Use the sheet for the material in the camera	Log the sheet and its edition, not just the number
Correction applied twice	Recompute from the metered time	Apply once, to the whole metered time	Record the metered time and the corrected time as separate fields
Extension forgotten	(factual/fassumed)²	Recompute N whenever the stack changes	Write the configuration in the log before the frames go on
Daylight speed used indoors	An absurd computed time	Change the material, not the time	Log the light’s description, not just its level

What to record

Two things: a corrected log, and an atlas entry.

The corrected log row. Do not erase the wrong number. Strike it through, write the right one beside it, and add one sentence saying which field was wrong and how you know. A log that shows its own corrections is the only kind that teaches anything, and a season of them is a map of your own characteristic mistakes.

The atlas entry. Each of the seven cases above is written to be lifted into the diagnostic atlas in the form the atlas uses: one defect, one set of causes, one confirming test, one corrective action, one prevention. The atlas entry has structured data at the top and six fixed sections in the body, so an entry drafted from this page looks like this:

---
title: Tunnel-effect vignetting from an over-thick pinhole plate
pageType: reference
refKind: defect
defect:
  id: tunnel-effect-vignetting
  name: Tunnel-effect vignetting
  aka: [thick-plate cut-off, pinhole tunnel vignetting]
  appearsOn: [negative]
  stage: [equipment, exposure]
  likelyCauses:
    - A pinhole plate thicker than about one fifth of the hole diameter
    - A hole drilled rather than pierced, leaving a parallel-sided bore
---

## What you see
Dark edges and corners with a comparatively hard boundary, at a radius that can be
computed; read as softness because a corner with no light in it has no detail.

## Likely causes
A plate whose thickness t is large relative to the hole diameter d.

## The chemistry and physics
The hole is a short cylinder. Beyond arctan(d/t) the near and far openings no longer
overlap and no light passes. True vignetting, multiplying the cos-fourth falloff.

## Diagnostic questions
Is the boundary hard or gradual? What is t, and what is arctan(d/t)? Does a blank-wall
negative show a rim at f tan(theta_cut)?

## Corrective action
A thinner plate, or chamfer both faces of the existing one, then re-measure.

## Prevention
Record t beside d in the pinhole register and compute the cut-off angle on entry.

Write one of those for every fault you produce, whether or not it is on this page’s list, and keep the negative with it. Seven entries and seven negatives is the beginning of a personal atlas, and Part XXVIII builds the general method on exactly this foundation.

📓Three faults this page does not cover, and where they go

So that the boundaries are clear. Uneven development, flow marks and streaking are processing faults and belong to Part IX and Part XXVIII, not to the camera. Fixing and washing faults — stains, residual thiosulfate, fading — belong to Part XI and Part XII. Emulsion faults, including coating defects on a plate you made yourself, belong to Part V.

If your negative is soft and streaked, you have two faults, and this page can only help with one of them. Diagnose them separately. The commonest way to get a diagnosis wrong is to look for a single cause of two independent symptoms.

Check your understanding

Question 1. A negative is soft. Under a loupe, an edge running vertically is crisp and an edge running horizontally is smeared. What does that establish, and what does it rule out?

• The hole is too large; it rules out diffraction
• The hole has a burr; it rules out motion
• Something moved during the exposure; it rules out every optical cause, because a hole has no preferred direction and blurs both edges equally
• The plate is too thick; it rules out the near limit

Show the answer and why

Answer: Something moved during the exposure; it rules out every optical cause, because a hole has no preferred direction and blurs both edges equally

This is the single most useful observation on the page, and it takes ten seconds. Every optical blur on a pinhole camera comes from the shape of the aperture, and the aperture is a hole with width in every direction, so it spreads a point into a two-dimensional figure and blurs an edge whichever way that edge runs. Motion does not: a smear is a line, and it lengthens an edge perpendicular to the motion while leaving an edge parallel to it untouched. So a direction-dependent difference is decisive for movement. Note the one trap: a burred hole also has a preferred direction, which is why option 2 is tempting. The difference is that a burr still has width across the smear, so the perpendicular edge is soft as well, just less so — and the burr smears in the same direction at every point in the frame, whereas subject motion smears only what moved. Once movement is established, three sub-cases remain, and they separate on whether everything in the frame is smeared alike (camera), only some objects are (subject), or there is a faint complete second image (a shutter jar, or a sheet that changed shape mid-exposure).

Question 2. A student uses a 0.30 mm hole drilled in 0.5 mm steel on 4 × 5 inch paper at f = 120 mm. Predict what changes in the image and why.

• Nothing changes; only the hole diameter matters
• The corners go dark with a comparatively hard boundary, because the tunnel cut-off is arctan(0.30/0.5) = 31.0° while the corners of the sheet sit at 34.1° — so no light reaches them at all
• The whole image gets softer, because a thicker plate diffracts more
• The exposure needs to be doubled

Show the answer and why

Answer: The corners go dark with a comparatively hard boundary, because the tunnel cut-off is arctan(0.30/0.5) = 31.0° while the corners of the sheet sit at 34.1° — so no light reaches them at all

A hole in a plate of thickness t is a short cylinder, and past the angle at which the near and far openings stop overlapping, no light gets through at all. That angle is arctan(d/t) = arctan(0.30/0.5) = 31.0°. The corner of a 4 × 5 inch sheet at f = 120 mm sits at arctan(81.3/120) = 34.1°, which is beyond it, so the corners receive nothing — and at f = 50 mm, where the corner is at 58.4°, most of the frame is gone. The cut-off radius on the film is f tan(31.0°) = 72 mm, inside the 81 mm half-diagonal, which is why the edge centres survive and the corners do not. Option 3 has the physics wrong twice: diffraction is set by the aperture width and the wavelength, not by the plate thickness, and this fault is a loss of light rather than a loss of definition. The geometry lesson's rule keeps you out of it: t no greater than d/5 puts the cut-off past 78°, which is beyond the corner of anything you would build.

Question 3. A log shows a metered 10 seconds on FOMAPAN 100 sheet film, corrected with ILFORD's published exponent of 1.26, and 18 seconds given. Recompute the correct time from FOMA's own data and state the error in stops.

• 80 seconds; the negative is about 2.1 stops underexposed
• 20 seconds; the error is negligible
• 12.6 seconds; the negative is about half a stop over
• 160 seconds; the negative is 3 stops under

Show the answer and why

Answer: 80 seconds; the negative is about 2.1 stops underexposed

Two makers publish the same physics in two different formats, and neither transfers. ILFORD give an exponent relation, Tc = Tm^P, with P = 1.26 for FP4 PLUS, which at a metered 10 seconds gives 10^1.26 = 18.2 s — a correct number, for the wrong film. FOMA publish a multiplier table for FOMAPAN 100 Classic, and their entry at a metered 10 seconds is a lengthening factor of eight, so the right answer is 80 seconds. Giving 18 where 80 was needed is a factor of 4.4, and log₂(4.4) = 2.14 stops underexposed. Note two things about this mistake. First, it is invisible on the negative: 2.1 stops under looks like 2.5 stops under from a forgotten extension factor, and only the log field naming the source sheet separates them. Second, it is not a small effect at the margins — the two materials differ by more than a factor of four at the same metered time, which is why the exposure lesson insists that the correction is taken from the sheet for the material actually in the camera and that the sheet is named in the log.

Question 4. A negative has correct overall density, clean highlights and flat grey shadows. The strip that was under the paper stop is clean, and the session's unexposed control sheet is clean. What is it?

• Safelight fog
• A light leak at one edge
• Flare inside the camera, because the veil came through the pinhole — so it stops at the image circle and leaves the covered strip clean — and the clean control exonerates the box, the safelight and the developer
• Exhausted developer

Show the answer and why

Answer: Flare inside the camera, because the veil came through the pinhole — so it stops at the image circle and leaves the covered strip clean — and the clean control exonerates the box, the safelight and the developer

Two pieces of evidence, each ruling out a different family. The covered strip is clean, which means the light that caused the veil arrived through the pinhole: image-forming light and flare both stop at the edge of the image circle, whereas a leak or handling fog carries straight on under the 3 mm paper stop. That rules out a leak. The control sheet is clean, and the control went through the same box, the same safelight and the same developer while never entering the camera — so safelight fog and an exhausted developer are both excluded, since either would have marked it. What is left is non-image light that entered the hole and was returned by the camera's own interior, which is flare, and the symptom pattern confirms it: a constant addition of exposure is negligible where the picture already delivered a lot of light and dominant where it delivered almost none, so the shadows lift, the highlights barely move, the toe compresses and the contrast falls. The confirming test is the flare pair with and without a flag, and the fix is mechanical — matt black, flocking, and a blackened pinhole plate on both faces.

Question 5. Why does this page insist on reading the exposure log before the negative?

• Because the negative is fragile
• Because a negative read with a hypothesis already in mind will supply evidence for it, and because most of the distinguishing information — which plate, which focal distance, which reciprocity sheet — is not visible on the negative at all
• Because the log is quicker to read
• Because the negative can only be read when dry

Show the answer and why

Answer: Because a negative read with a hypothesis already in mind will supply evidence for it, and because most of the distinguishing information — which plate, which focal distance, which reciprocity sheet — is not visible on the negative at all

Two reasons, and the second is the stronger. The first is perceptual: unsharpness of every kind looks alike at low magnification, and someone who has decided the hole is at fault will find evidence that the hole is at fault. Reading the log first means the hypothesis arrives with a number attached rather than from an impression. The second is that a great deal of the evidence is simply not in the image. A thin negative cannot tell you whether the exposure was short, the extension factor forgotten, or a reciprocity correction taken from the wrong maker's sheet — all three produce a uniformly thin picture with a normal rebate, and 2.1 stops under looks exactly like 2.5 stops under. Only the log distinguishes them, and only if it was written before the shutter opened. That is why the exposure log carries a field for the source of the reciprocity correction and a field for the configuration, and why a log completed after processing is, as the exposure lesson puts it, a memoir.

Check my answersClear