Photo comparison is the problem of deciding whether visible differences between two photographs come from the subject or from camera position, focal length, lighting, exposure, white balance, pose, focus, crop, scale, processing and output. Search for “before and after photography,” “compare two photos,” “same angle before after,” “progress photos,” “scientific image comparison,” or “photo documentation change,” and the key requirement is not that the pictures merely look similar. The relevant variables must be controlled well enough for the difference to mean what we think it means.
A face photographed close with a wide lens can look different from the same face photographed farther away with a longer lens. Side light can reveal texture that frontal light suppresses. A room can look larger from a corner. Auto exposure can brighten a darker specimen. White balance can shift colour. Sharpening can create stronger edges. If those variables change between frames, “before versus after” becomes a mixture of subject change and camera change.
This is why before-and-after photography, photo comparison, camera position, focal length, lighting, exposure, white balance, pose, calibration, scientific photography and image analysis belong to one mechanism. Define the quantity being compared, identify every imaging variable capable of changing its appearance, lock or calibrate those variables, preserve originals, and report residual uncertainty.
Central proposition: a visible difference between photographs is evidence of subject change only after relevant imaging differences have been controlled or modelled.
1. Define the Quantity
Size, colour, texture, crowd density, damage and posture have different confounders.
2. Lock Camera Position
Moving the camera changes perspective and apparent proportions.
3. Lock Camera Height
Height changes visible surfaces, horizon relationships and body proportions.
4. Lock Camera Orientation
Tilt and yaw alter projective geometry.
5. Match Focal Length
Different fields of view can change framing and encourage different working distances.
6. Match Subject Distance
Perspective depends strongly on camera-to-subject position.
7. Match Subject Pose
Rotation, posture and expression can create apparent structural change.
8. Match Lighting Direction
Side light reveals texture; front light suppresses it.
9. Match Light Quality
Hard and soft light create different shadow boundaries and perceived surface detail.
10. Match Light Intensity or Exposure Strategy
Automatic exposure can hide genuine brightness change by compensating for it.
11. Match White Balance
Colour comparison is weak when illuminant interpretation changes.
12. Control the Illuminant
Two lamps with similar apparent white can render materials differently.
13. Use Colour References
A calibrated target helps separate subject colour change from pipeline change.
14. Match Focus
Defocus can remove fine texture and edge contrast.
15. Match Aperture
Depth of field and diffraction alter visible detail.
16. Match Shutter When Motion Matters
Blur can imitate shape or texture change.
17. Match ISO and Processing
Noise reduction and sharpening can change apparent detail.
18. Preserve RAW
Upstream files retain more evidence about exposure and colour.
19. Do Not Compare Screenshots
Screenshots can rescale, recompress and strip metadata.
20. Control Cropping
Different crops alter scale references and contextual relationships.
21. Control Output Size
A detail visible in one enlargement may disappear in another.
22. Use Registration
Overlaying aligned images reveals change more reliably than memory.
23. Use Fixed Landmarks
Stable scene points help recover camera alignment.
24. Use Tripod Marks
Repeatable physical setup is cheaper than correcting geometry later.
25. Use Tethered Overlays
Ghosted previous frames can help reproduce viewpoint in long-term projects.
26. Before-and-After Portraits Need Discipline
Pose, expression, makeup, lens distance and light can dominate perceived change.
27. Property Comparisons Need Geometry
Wide-angle viewpoint changes can exaggerate renovation effects.
28. Environmental Monitoring Needs Season Control
Weather and season can overwhelm the change being studied.
29. Medical or Scientific Images Need Protocols
High-stakes comparisons require validated acquisition and analysis, not persuasive presentation.
30. Product Tests Need Identical Rendering
Lighting and polarisation can alter gloss, scratches and colour.
31. Historical Rephotography Needs Viewpoint Reconstruction
Old lens metadata may be missing, so architecture and horizon clues become alignment evidence.
32. Difference Images Are Powerful
Subtracting registered frames can reveal where pixels changed, but illumination changes can dominate the difference.
33. Calibration Converts Appearance Into Measurement
Scale targets and known geometry allow numerical comparison.
34. Report Uncertainty
A 2% apparent change is meaningless if setup variability is 5%.
35. Replicate
Repeated matched captures reveal how much the imaging system itself varies.
36. Blind Evaluation When Possible
If the assessor knows which image is “after,” expectation can influence judgement.
37. Comparison Can Be Artistic
Deliberately mismatched conditions can express time or memory; they simply should not masquerade as controlled measurement.
38. Reader Diagnosis
Ask what changed in the world, what changed in the camera, and what changed in processing.
39. Repair
Recreate geometry, light, exposure and output; then compare.
40. World Return
Two photographs are not a measurement merely because they are placed side by side. Comparability is engineered.
The Clementi Compression
Define the reader job first. Separate subject change from camera change, name the mechanism, expose the confounders, preserve the original evidence, and make the repair operational. Search language belongs naturally at the entrance; the body earns authority through mechanism, diagnosis, worked reasoning, failure modes, transfer and a return to the wider Photography system.
The Final Idea
The photograph becomes more useful when we stop asking it to be self-explanatory. Control what can be controlled, document what changed, and let uncertainty remain visible where the image cannot discriminate.