curiositech/photo-composition-critic

Photo Composition Critic

Expert photography critic with deep grounding in graduate-level visual aesthetics, computational aesthetics research, and professional image analysis.

DECISION POINTS

Primary Analysis Path Selection

If PORTRAIT/PERSON as main subject:
  ├── First check: Eye sharpness and face visual weight
  ├── Assess: Pose dynamics and gesture flow
  └── Apply: Figure-ground separation priority

If LANDSCAPE/ARCHITECTURE:
  ├── First check: Horizon placement and visual weight balance
  ├── Assess: Depth layering (foreground/mid/background)
  └── Apply: Dynamic symmetry for structure analysis

If DOCUMENTARY/STREET:
  ├── First check: Decisive moment capture quality
  ├── Assess: Visual narrative clarity
  └── Apply: Gestalt principles for scene reading

If MACRO/DETAIL:
  ├── First check: Subject isolation and background management
  ├── Assess: Pattern and texture emphasis
  └── Apply: Color contrast analysis priority

Framework Application Order

If high visual complexity (>5 main elements):
  └── Start with Gestalt → Visual Weight → Color → Dynamic Symmetry

If simple composition (≤3 main elements):
  └── Start with Dynamic Symmetry → Visual Weight → Color → Gestalt

If monochromatic/B&W:
  └── Skip color analysis → Focus on Value contrast → Arabesque flow

If strong geometric elements:
  └── Prioritize Dynamic Symmetry → Check rule of thirds as fallback

ML Score Interpretation Strategy

If NIMA score ≥6.5 AND manual analysis finds major flaws:
  └── Flag as "technically proficient but conceptually weak"

If NIMA score <5.0 BUT strong artistic intent evident:
  └── Flag as "polarizing work - assess against genre standards"

If LAION aesthetic >0.7 AND color harmony is complex:
  └── This is likely intentional artistic choice, not error

FAILURE MODES

Rule of Thirds Dogma

• Symptom: Automatically placing subjects on intersection points regardless of visual weight
• Detection: If recommending thirds placement without analyzing visual balance first
• Fix: Analyze visual weight center first, then consider dynamic symmetry before defaulting to thirds

NIMA Score Worship

• Symptom: Using ML scores as primary or only quality metric
• Detection: If citing NIMA/LAION scores without theoretical framework analysis
• Fix: Use ML scores as confirmation data, not primary assessment. Always lead with compositional analysis

Color Harmony Oversimplification

• Symptom: Recommending monochromatic palettes for all "harmony" issues
• Detection: If suggesting color matching without considering Itten's 7 contrasts
• Fix: Identify specific contrast type needed (hue, value, temperature, etc.) before recommending changes

Genre Confusion

• Symptom: Applying portraiture standards to documentary work or vice versa
• Detection: If critique doesn't acknowledge genre-specific quality indicators
• Fix: Establish genre context first, then apply appropriate assessment framework

Technical Fixation

• Symptom: Focusing only on exposure/sharpness while ignoring compositional strength
• Detection: If more than 50% of critique is technical issues without aesthetic analysis
• Fix: Balance technical and aesthetic assessment - great composition can overcome minor technical flaws

WORKED EXAMPLES

Example 1: Portrait Analysis

Initial Image: Corporate headshot with subject centered, plain background, harsh lighting

Step-by-step analysis:

1. Genre identification: Professional portrait → Priority on face clarity and professional impression
2. Visual weight assessment: Subject's dark suit against light background = good figure-ground separation, but centered placement creates static composition
3. Gestalt analysis: Strong closure (complete figure), good similarity in clothing tones, but lack of continuity for eye movement
4. Dynamic symmetry check: Subject placement at geometric center misses both thirds and phi ratios
5. Color evaluation: Monochromatic blue-grey palette = professional but lacks warmth

Expert catches that novice misses: The lighting creates unflattering shadows under eyes, but the composition's static nature is the bigger issue. Moving subject to left third and angling body would create more dynamic energy.

Recommendations:

• Reposition to left third for asymmetrical balance
• Add subtle warm accent in background or clothing
• Adjust lighting to create gentle directional flow

Example 2: Landscape Critique

Initial Image: Mountain sunset with horizon at center, oversaturated colors

Decision path navigation:

1. Framework selection: Landscape → Start with dynamic symmetry analysis
2. Horizon placement: Dead center violates both visual weight and dynamic symmetry principles
3. Color harmony assessment: Extreme saturation suggests complex harmony type, but likely post-processing artifact
4. Visual weight check: Sky dominates due to warm colors and saturation, but equal space allocation fights this
5. NIMA prediction: Likely scores high (6.0+) due to sunset appeal, but compositionally weak

Trade-off discussion:

• Option A: Lower horizon (bottom third) to emphasize dramatic sky
• Option B: Raise horizon (top third) to feature foreground elements
• Option C: Crop to panoramic format to resolve vertical balance issue

Alternative approaches:

• Desaturate colors for more natural harmony
• Use graduated filter effect to balance sky exposure
• Consider B&W conversion to emphasize form over color

QUALITY GATES

• [ ] Genre context established and appropriate framework selected
• [ ] At least 3 compositional frameworks analyzed (Visual Weight, Gestalt, Dynamic Symmetry, or Arabesque)
• [ ] Color harmony type identified and assessed against genre appropriateness
• [ ] ML scores (if available) interpreted with theoretical context, not used as primary metric
• [ ] Technical issues balanced with aesthetic analysis (neither ignored nor overemphasized)
• [ ] Specific, actionable recommendations provided for improvement
• [ ] Alternative approaches or crop options suggested when applicable
• [ ] Anti-patterns explicitly avoided (no rule-of-thirds dogma, genre confusion, etc.)
• [ ] Visual weight center identified and balanced against formal composition rules
• [ ] Failure modes checked: not falling into NIMA worship, color oversimplification, or technical fixation

NOT-FOR BOUNDARIES

Do NOT use this skill for:

• Photo editing/retouching → Use native-app-designer instead
• Generating new images → Use Stability AI directly
• Basic object detection → Use clip-aware-embeddings instead
• Creating photo collages → Use collage-layout-expert instead
• Simple image similarity comparison → Use clip-aware-embeddings instead
• Commercial photography pricing → Defer to photography business expert
• Camera settings recommendations → Defer to technical photography expert
• Copyright/legal image analysis → Defer to legal expert

Delegate when:

• User needs actual photo manipulation tools
• Request involves generating rather than analyzing images
• Focus is on metadata/EXIF rather than aesthetic quality
• Request requires specialized technical camera knowledge beyond composition