Application Environment Variables Explainer

This document details the environment variables used in the application, as defined in config/default.py. These variables control infrastructure settings, model versions, storage locations, and feature configurations.

Quick Start: A dotenv.template file is provided in the root directory. To set up your local environment, copy this file to .env and populate the values:
Terminal window
cp dotenv.template .env

🌍 Core Infrastructure & Environment

These variables define the fundamental operating context of the application.

Variable	Default	Description
`PROJECT_ID`	None (Required)	The Google Cloud Project ID where resources (Google Cloud AI, Firestore, Storage) are located.
`LOCATION`	`us-central1`	The default GCP region for most services (Google Cloud AI, etc.).
`APP_ENV`	`""` (Empty)	Defines the environment name (e.g., `dev`, `godemos`). This is used as a metadata tag on the Config page.
`GMCS_OVERRIDE_PATH`	None	(Development Only) An absolute path to a directory containing configuration overrides. If a file exists in this path (e.g., `config/about_content.json`), the app will prioritize it over the local version.
`API_BASE_URL`	`http://localhost:{PORT}`	The base URL for the application’s backend APIs.
`PORT`	`8080`	The port the application server listens on.
`SERVICE_ACCOUNT_EMAIL`	None	The email of the service account used for authentication, if applicable.
`GA_MEASUREMENT_ID`	None	Google Analytics Measurement ID for tracking user interactions.
`REQUIRE_AUTHENTICATED_USER`	`true` outside local/dev/test, otherwise `false`	Reject non-health requests that do not include a trusted upstream identity header. Set this to `true` behind Netskope NPA/oauth2-proxy or IAP.
`AUTH_EMAIL_HEADERS`	IAP + oauth2-proxy/Netskope defaults	Comma-separated trusted request headers to inspect for the authenticated user email. Defaults: `X-Goog-Authenticated-User-Email`, `X-Auth-Request-Email`, `X-Forwarded-Email`, `X-Email`, `X-Authenticated-User`.

🧠 Gemini Models (Text & Multimodal)

Controls which versions of the Gemini models are used for various tasks.

Variable	Default	Description
`MODEL_ID`	`gemini-3.5-flash`	The primary Gemini model used for general text and reasoning tasks throughout the app.
`GEMINI_IMAGE_GEN_MODEL`	`gemini-2.5-flash-image`	The default model used for image generation features (supports `gemini-3.1-flash-lite-image`, `gemini-3.1-flash-image`, `gemini-3-pro-image`, or `gemini-2.5-flash-image`).
`GEMINI_IMAGE_GEN_LOCATION`	`global`	The region for the Gemini Image Generation API.
`GEMINI_AUDIO_ANALYSIS_MODEL_ID`	`gemini-3.1-flash-lite`	The model used specifically for analyzing audio content.
`GEMINI_WRITERS_WORKSHOP_MODEL_ID`	`MODEL_ID`	The model used for the Gemini Writers Workshop page. Defaults to `MODEL_ID`.
`GEMINI_CRITIQUE_MODEL_ID`	`gemini-3-flash-preview`	The specific model used for the Imagen critique functionality.
`GEMINI_CRITIQUE_LOCATION`	`global`	The region for the Gemini image critique model.
`CHARACTER_CONSISTENCY_GEMINI_MODEL`	`MODEL_ID`	The model used for Character Consistency tasks.
`CHARACTER_CONSISTENCY_GEMINI_LOCATION`	`global`	The region for the Character Consistency Gemini model.

🎞️ Gemini Omni (Video Generation & Editing)

Configuration for the Gemini Omni Flash multimodal video interaction model.

Variable	Default	Description
`DEFAULT_OMNI_MODEL_NAME`	`gemini-omni-flash-preview`	The fallback model identifier for Gemini Omni when starting a session.
`OMNI_LOCATION`	`global`	Region for the Gemini Omni API endpoint.
`OMNI_MODEL_ID`	`gemini-omni-flash-preview`	The standard Gemini Omni model ID.
`OMNI_PROJECT_ID`	`PROJECT_ID`	Allows using a different GCP project for Gemini Omni quota if needed.
`OMNI_TIMEOUT_MS`	`600000`	Client-side HTTP timeout in milliseconds (default 10 minutes) for long-running video generation and editing requests.

🎥 Veo (Video Generation)

Configuration for the Veo video generation models.

Variable	Default	Description
`DEFAULT_VEO_MODEL_NAME`	`veo-3.1-fast-generate-001`	The fallback model name for the UI when a user clears their session or deep links.
`VEO_LOCATION`	`us-central1`	Region for GA Veo API calls.
`PREVIEW_LOCATION`	`global`	Region for Preview Veo API calls (models with “preview” in their name).
`VEO_MODEL_ID`	`veo-3.1-fast-generate-001`	The standard Veo model version.
`VEO_PROJECT_ID`	`PROJECT_ID`	Allows using a different project for Veo quota if needed.
`VEO_EXP_MODEL_ID`	`veo-3.1-generate-001`	The experimental/newer Veo model version.
`VEO_EXP_FAST_MODEL_ID`	`veo-3.1-fast-generate-001`	The faster, lower-latency experimental Veo model.
`VEO_EXP_PROJECT_ID`	`PROJECT_ID`	Project ID for experimental Veo models.

🎨 Imagen (Image Generation & Editing)

Settings for Imagen models, including specialized versions for editing and product shots.

Variable	Default	Description
`IMAGEN_GENERATED_SUBFOLDER`	`generated_images`	Subfolder in the GCS bucket where generated images are saved.
`IMAGEN_EDITED_SUBFOLDER`	`edited_images`	Subfolder for images resulting from editing operations.

🛍️ Virtual Try-On (VTO)

Specific configuration for the Virtual Try-On feature.

Variable	Default	Description
`VTO_LOCATION`	`us-central1`	Region for the VTO API.
`VTO_MODEL_ID`	`virtual-try-on-001`	The specific VTO model version.
`GENMEDIA_VTO_MODEL_COLLECTION_NAME`	`genmedia-vto-model`	Firestore collection for VTO model data.
`GENMEDIA_VTO_CATALOG_COLLECTION_NAME`	`genmedia-vto-catalog`	Firestore collection for VTO product catalog data.

🎵 Lyria (Music Generation)

Configuration for the Lyria music generation model.

Variable	Default	Description
`LYRIA_LOCATION`	`us-central1`	Region for Lyria API calls.
`LYRIA_MODEL_VERSION`	`lyria-002`	The version of the Lyria model to use.
`LYRIA_PROJECT_ID`	`PROJECT_ID`	Project ID for Lyria quota.

💾 Storage & Database (Firebase/GCS)

Defines where data and media assets are stored.

Variable	Default	Description
`GENMEDIA_FIREBASE_DB`	`(default)`	The Firestore database ID.
`GENMEDIA_COLLECTION_NAME`	`genmedia`	The main Firestore collection for storing generation metadata.
`SESSIONS_COLLECTION_NAME`	`sessions`	Firestore collection for user session data.
`GENMEDIA_BUCKET`	`{PROJECT_ID}-assets`	The primary GCS bucket for storing generated media.
`VIDEO_BUCKET`	`{PROJECT_ID}-assets/videos`	Specific bucket/path for video files.
`IMAGE_BUCKET`	`{PROJECT_ID}-assets/images`	Specific bucket/path for image files.
`MEDIA_BUCKET`	`{PROJECT_ID}-assets`	Used by Lyria and potentially other legacy components.
`GCS_ASSETS_BUCKET`	None	Bucket for static assets used in the “About” page.

⚙️ Application Logic

Variable	Default	Description
`LIBRARY_MEDIA_PER_PAGE`	`15`	Controls how many items appear per page in the media library.
`USE_MEDIA_PROXY`	`true`	If `true`, media URLs are proxied to avoid CORS/hotlinking issues.

📦 Build Metadata

The application can display detailed build information on the Config page. This is populated from an optional JSON file:

File Path: config/build.json (or via GMCS_OVERRIDE_PATH)

Format:

{
  "commit": "a1b2c3d",
  "date": "2026-01-13 14:30:00"
}

If present, these values populate the BUILD_COMMIT and BUILD_DATE fields in the application state.

🏗️ Terraform Configuration & Deployment

When deploying this application using Terraform (via main.tf), not all environment variables are exposed for configuration. The Terraform setup manages a specific subset of variables, primarily those related to infrastructure and core model IDs.

1. Variables Controllable via `variables.tf`

These variables are exposed in variables.tf and directly map to environment variables in the Cloud Run service. You can customize these by setting the corresponding Terraform variable during deployment.

Terraform Variable	Maps to App Env Var	Default in Terraform
`project_id`	`PROJECT_ID`	(Required)
`region`	`LOCATION`	`us-central1`
`model_id`	`MODEL_ID`	`gemini-3.5-flash`
`gemini_audio_analysis_model_id`	`GEMINI_AUDIO_ANALYSIS_MODEL_ID`	`gemini-3.1-flash-lite`
`gemini_critique_model_id`	`GEMINI_CRITIQUE_MODEL_ID`	`gemini-3-flash-preview`
`gemini_critique_location`	`GEMINI_CRITIQUE_LOCATION`	`global`
`character_consistency_gemini_location`	`CHARACTER_CONSISTENCY_GEMINI_LOCATION`	`global`
`veo_model_id`	`VEO_MODEL_ID`	`veo-3.1-fast-generate-001`
`veo_location`	`VEO_LOCATION`	`region`
`veo_exp_model_id`	`VEO_EXP_MODEL_ID`	`veo-3.1-generate-001`
`lyria_model_id`	`LYRIA_MODEL_VERSION`	`lyria-002`
`edit_images_enabled`	`EDIT_IMAGES_ENABLED`	`true`

2. Variables Automatically Managed by Terraform

These variables are computed within main.tf based on the resources Terraform creates (e.g., bucket names, service account emails). You generally cannot change these via variables.tf as they ensure the application correctly connects to the provisioned infrastructure.

App Env Var	Source in `main.tf`	Value Logic
`GENMEDIA_BUCKET`	`local.asset_bucket_name`	`creative-studio-{project_id}-assets`
`VIDEO_BUCKET`	`local.asset_bucket_name`	Same as above
`MEDIA_BUCKET`	`local.asset_bucket_name`	Same as above
`IMAGE_BUCKET`	`local.asset_bucket_name`	Same as above
`GCS_ASSETS_BUCKET`	`local.asset_bucket_name`	Same as above
`GENMEDIA_FIREBASE_DB`	Resource Attribute	Name of the created Firestore DB
`SERVICE_ACCOUNT_EMAIL`	Resource Attribute	Email of the created Service Account
`LYRIA_PROJECT_ID`	`var.project_id`	Forces Lyria to use the main project ID

3. Variables NOT Set by Terraform (Using Python Defaults)

The following variables are not explicitly set in the main.tf configuration. This means the application will use the default values defined in config/default.py when deployed via Terraform.

Gemini Models: GEMINI_IMAGE_GEN_MODEL (supports Nano Banana 2 Lite gemini-3.1-flash-lite-image), GEMINI_IMAGE_GEN_LOCATION, GEMINI_AUDIO_ANALYSIS_MODEL_ID
Gemini Omni: DEFAULT_OMNI_MODEL_NAME, OMNI_LOCATION, OMNI_MODEL_ID, OMNI_PROJECT_ID, OMNI_TIMEOUT_MS
Veo: DEFAULT_VEO_MODEL_NAME, PREVIEW_LOCATION, VEO_PROJECT_ID, VEO_EXP_FAST_MODEL_ID, VEO_EXP_PROJECT_ID
VTO (Virtual Try-On): VTO_LOCATION, VTO_MODEL_ID, GENMEDIA_VTO_* collection names.
Imagen: MODEL_IMAGEN_PRODUCT_RECONTEXT, IMAGEN_GENERATED_SUBFOLDER, IMAGEN_EDITED_SUBFOLDER
App Logic: APP_ENV, API_BASE_URL, GA_MEASUREMENT_ID, LIBRARY_MEDIA_PER_PAGE, USE_MEDIA_PROXY
Auth: REQUIRE_AUTHENTICATED_USER, AUTH_EMAIL_HEADERS
Collections: GENMEDIA_COLLECTION_NAME, SESSIONS_COLLECTION_NAME

🛠️ How to Deploy with Custom Values

To change a variable from Group 3 (e.g., GEMINI_IMAGE_GEN_MODEL) when deploying with Terraform:

Modify variables.tf: Add a new variable definition.

variable "gemini_image_model" {
  description = "Model ID for Gemini Image Generation"
  type        = string
  default     = "gemini-3-pro-image"
}

Modify main.tf: Update the locals block to include the new environment variable mapping.

locals {
  creative_studio_env_vars = {
    # ... existing vars ...
    GEMINI_IMAGE_GEN_MODEL = var.gemini_image_model
  }
}