QI

AI Image Quality Diagnostic Workbench V2.4 Local Vision CV

Canvas CV Pipeline: Active
Benchmark Presets:
Model: Flux 1.0 Dev
Filter Mode: Edge Gradient Coherence (Sobel)
Mean Gradient Dispersion: 14.8 px
Probe: X: 412 Y: 328 | Normal Vector: [0.72, -0.48, 0.50]
Diagnostic Filter:
1.2x
Diagnostic Scores FLUX-DEV
Composite Quality Index
Multi-axis Coherence Score
89.7
Anatomical Continuity
91.2
High-Freq Texture Fidelity
88.0
Specular Consistency
84.5
Typography Sharpness
95.0
Detected Artifact Taxonomy
Prompt Remediation Engine
Original Stress Prompt:
Macro photo of an artisan jeweler holding a faceted ruby ring in soft directional studio light
Recommended Prompt Adjustments:
Inject specular angle grounding to fix ray alignment disparity between ring facet and cuticle.
--lighting "single key rim light at 45 deg, subsurface keratin scatter, specular reflection matching"
State-of-the-Art Generative Model Quality Benchmark
Model Architecture Anatomy & Hands Micro-Texture / Skin Lighting Vector Physics Typography Coherence Prompt Adherence Failure Mode Archetype
Flux 1.0 (Schnell / Dev) 91.2% High 88.0% Crisp 84.5% Coherent 95.0% Sharp 94.8% Exact Minor specular dispersion jitter on metallic edges
Midjourney v6.0 85.4% Solid 93.2% Ultra 90.1% Natural 76.4% Fair 88.2% High Hallucinated glyphs in complex embedded lettering
Stable Diffusion XL 1.0 72.3% Variable 82.1% Good 74.0% Inconsistent 61.5% Smudged 81.0% Good Interdigital webbing and joint over-multiplication
DALL-E 3 82.0% Good 71.5% Plastic 83.2% Decent 92.0% Clean 96.5% Superior Over-smoothed synthetic skin; loss of pore detail

Sobel Edge Coherence and Specular Vector Diagnostics

Read the explanation

The diagnostic workbench evaluates synthetic images by applying real-time convolution filters directly across pixels in canvas memory. Selecting Sobel Edge Coherence computes horizontal and vertical luminance gradients Gx and Gy, calculating edge magnitude to reveal contour disruptions. Moving the probe over the image samples normal vectors dx and dy, identifying lighting mismatches between facets and fingers to generate automated prompt remedies.

Super generates helpful tools and automates fact-checking across the internet proactively. If you enjoyed this tool, build your own with Super and share it with a friend.