AI WALLPAPER QUALITY STANDARD - EVALUATION CRITERIA

1. Objective & Framework Overview
The AI Wallpaper Quality Standard  (AWQS 1.9) is a checklist-based 100-point evaluation system for digital wallpapers. The total score is divided equally across 5 core pillars (20 points per pillar). Unlike holistic severity judgments, AWQS 1.9 uses fixed, itemized checklists: each pillar starts at 20 points, and a fixed number of points is subtracted for every criterion that applies. Nothing is estimated by feel — only checked.

2. Pre-Checks (Mandatory, Before Any Scoring)

Resolution Requirement: If no resolution was provided by the user, stop and explicitly ask for it before producing any score or Device Suitability Classification. Do not proceed with a full evaluation in the absence of resolution data.

Resolution Sanity Check: Before using the user-provided resolution, calculate its aspect ratio and compare it to the visually apparent aspect ratio of the image itself. If these are inconsistent (e.g. a stated 21:9 resolution but the image visually appears roughly square or 4:3), stop and flag this discrepancy to the user instead of proceeding with the evaluation.

Watermark / Overlay Check: Inspect the image for any visible watermark, logo, or overlay text. If present, explicitly note this and state whether it affects the evaluation of underlying detail in the overlapped area (relevant to Pillar 1 and Pillar 2).

Evidence Requirement: For every checklist item marked "Not applied," briefly state the specific visual evidence checked to confirm this — e.g. "checked: hand position, finger count, wrist angle — all anatomically consistent" — not a general impression like "looks fine." For every checklist item marked as applying, cite the specific location in the image where the flaw is visible. If specific visual evidence cannot be identified for an item, mark it as unable to verify rather than "not applied."

3. The 5-Pillar Checklist System (Max 100 Points Total)

Pillar 1: Composition & Framing (Max 20 pts) — Aesthetic
Start at 20 points. Subtract for each criterion that applies:
- Main subject overlaps with another equally dominant visual element (competing focal point): -3
- Main subject sits at the edge (outside the central 80% of the frame) without clear compositional intent: -4
- No leading line, frame, or contrast difference guides the eye toward the main subject: -4
- Key visual elements are distributed with tedious symmetry (no tension/asymmetry at all): -2
- Main subject is fully cropped out or unrecognizable due to framing: -10

Pillar 2: Detail & Upscale Precision (Max 20 pts) — Objective
Start at 20 points. Subtract for each criterion that applies:
- Main subject shows visible smoothing/loss of texture at normal viewing distance (not zoomed in): -5
- Visible pixel artifacts, noise, or blocking at normal viewing distance: -6
- Clear "AI-plastic" smoothing on skin/material texture of the main subject: -6
- Floating or illogically placed details that don't attach to any surface/structure: -4
- Obvious upscaler artifacts (grid lines, repeating patterns) anywhere in the frame: -8
Note: stylistic choices (atmospheric haze, intentional soft focus, glow effects) are NOT a smoothing flaw and do not count unless they specifically affect the main subject itself. A visible watermark/overlay is not itself a smoothing flaw but must be disclosed per the Pre-Check above.

Pillar 3: Color Harmony & Lighting (Max 20 pts) — Aesthetic
Start at 20 points. Subtract for each criterion that applies:
- Two or more light sources cast shadows in conflicting directions without explanation: -5
- One element has a clearly different color temperature than the rest of the scene, without a narrative/stylistic reason: -3
- Visible color banding (hard transitions instead of smooth gradients) in large flat areas: -4
- Overexposed or underexposed zones that fully flatten detail (>10% of the image): -6
- Color palette feels incoherent — 4 or more unrelated dominant colors with no clear shared theme: -5

Pillar 4: Anatomy & Structure (Max 20 pts) — Objective, with viewing-scale rule
Viewing-Scale Gate (apply first): Is the flaw only visible at >150% zoom or on a cropped view? → Yes: automatic maximum -3 deduction, skip the checklist below. Is the flaw visible at normal viewing distance (the full image at screen size)? → Proceed to checklist:
- A limb, hand, or facial feature is structurally incorrect (wrong count, impossible angle) and visible at normal viewing distance: -7
- An architectural element physically overlaps another in an impossible way, visible at normal viewing distance: -6
- Perspective lines fail to converge to a consistent vanishing point (for architectural scenes): -5
- An object visibly "melts" into another without a clear boundary: -6
- The main figure/structure is missing an essential part (e.g., a foot, a section of roof) without it being outside the frame: -10

- Creature or figure has an anatomically unexplainable duplicate or extra structural element (e.g., a second head, an extra limb) that does not logically attach to or share a single coherent body: -8

Note: This does NOT apply to creatures intentionally designed with multiple heads/limbs as a coherent stylistic or mythological feature (e.g., a hydra, a multi-headed deity), where each element connects logically and consistently to a single shared body. It DOES apply when an extra element appears structurally unresolved — for example, a second head or limb that emerges from an ambiguous or illogical point on the body, without a clear, consistent connection.

Pillar 5: Screen Optimization (Max 20 pts) — Objective, device-specific
First determine the device type from the resolution. Then apply the matching sub-checklist:

For phones/tablets (floating icons/dock UI):
- Important detail sits within the top 5% of the frame (clock zone): -4
- Main subject sits entirely within the bottom 15% (dock overlap risk): -3
- No calm/flat zone of at least 10% is available anywhere for widgets: -3

For desktop/laptop (fixed taskbar, persistent icons):
- Critical detail sits entirely within the bottom 8% (taskbar zone): -6
- Critical detail sits entirely within the left/right 5% (dock/icon column): -4
- No calm zone available for desktop icons: -4

For ultrawide/super-ultrawide (extra-wide edges):
- Main subject sits outside the central 60% (disappears into the "wings" on a multi-monitor setup): -6
- Composition is clearly designed for a narrower ratio and looks stretched/empty at the sides: -5

4. Device Suitability Classification
1080×2400 = Android phones; 1440×3200 = high-end phones; 1920×1080 = desktops/laptops/Chromebooks; 2560×1600 = tablets/iPad Air/Pro 11"; 2732×2048 = iPad Pro 12.9"; 3200×3200 = iPad family/square tablets; 3840×2160 = 4K desktops/TVs; 768×1366 = older laptops/Chromebooks; 2560×1080 = ultrawide monitors (21:9, standard); 3440×1440 = ultrawide monitors (21:9, WQHD); 5120×1440 = super-ultrawide monitors (32:9, dual QHD).
If the resolution does not exactly match any listed value, state the closest standard category and explain the match (e.g. aspect ratio) in one sentence.

5. AWQS Quality Labels & Score Brackets
90–100 Points: Masterclass
78–89 Points: Excellent Quality
70–77 Points: Standard Grade
Below 70 Points: Below Standard

6. Required Output Format for AI Evaluation

When evaluating an image using this standard, the AI must output:
- Resolution & Device Information: stated resolution, calculated aspect ratio, result of the Resolution Sanity Check, and result of the Watermark/Overlay Check.
- Pillar Breakdown: for each of the 5 pillars, list every checklist criterion with its applied/not-applied status AND the specific visual evidence checked (per the Evidence Requirement), followed by the pillar's point total (e.g., Pillar 1: 18/20).
- Final Total Score: the aggregate sum out of 100 points.
- AWQS Quality Label: the corresponding label based on the brackets above.
- Device Suitability Classification: one short sentence stating the best device category.
- Closing line: "This AWQS  evaluation was compiled by [AI-name] with a score of [total score]/100."

7. Public-Facing Disclaimer
Composition & Color Harmony pillars involve inherent aesthetic judgment. AWQS  minimizes — but cannot fully eliminate — interpretation in these two pillars, by design.

Disclaimer: AWQS is an informal, self-hosted evaluation tool, not an official, certified, or third-party-audited quality standard. Scores are generated by AI chatbots interpreting the rules above, and — as documented throughout the development of this framework — AI evaluators can occasionally misjudge, misread, or overlook details in an image, including missing genuine flaws and awarding a higher score than warranted. AWQS is best used as a structured aid to your own judgment, not as a replacement for it. Any decision to publish, purchase, or rely on a wallpaper based on its AWQS score remains the sole responsibility of the person making that decision.