Account, Billing & Admin
Studio Admin: Model Calibration & Parameters
Configuring Azure AI Foundry model deployments, token pricing, and tuning SystemCullingConfig boundaries.
Studio Administrator Panel
System administrators and studio owners have access to configuration surfaces under /admin to tune evaluation parameters and manage AI infrastructure.
AI Model Deployment Management (/admin/settings/ai-model)
- Deployment Routing: Configure active model endpoints hosted on Microsoft Azure AI Foundry (e.g. GPT-5-mini).
- Cost Controls: Define input/output token pricing per 1M tokens to monitor studio operational expenses in real time.
- Reasoning Effort Configuration: Tune default reasoning effort levels (
low,medium,high) across different project types.
System Culling Calibration (/admin/settings/culling)
The culling calibration interface visualizes the distribution curves calculated by SystemCullingConfig:
| Parameter | Default Value | Operational Function |
|---|---|---|
fiveStarBoundary | 85 | Percentile cutoff for 5-star ratings. |
fourStarBoundary | 55 | Percentile cutoff for 4-star ratings. |
threeStarBoundary | 25 | Percentile cutoff for 3-star ratings. |
twoStarBoundary | 10 | Percentile cutoff for 2-star ratings. |
excellentOverride | 90 | Score threshold unconditionally forcing 5 stars. |
poorOverride | 35 | Score threshold unconditionally forcing 1 star / Discard. |
spreadThreshold | 20 | Minimum spread required before activating percentile compression. |
Adjusting these boundaries recalculates real-time sample shoot histograms, allowing studio leads to verify calibration before rolling changes out to active photographer teams.