Videos: Add "codec" search filter and auto-enable nvidia encoder #4848

Signed-off-by: Michael Mayer <michael@photoprism.app>

Makefile

@@ -386,6 +386,9 @@ docker-pull:
 docker-build:
 	$(DOCKER_COMPOSE) --profile=all pull --ignore-pull-failures
 	$(DOCKER_COMPOSE) build --pull
+docker-nvidia: docker-nvidia-up
+docker-nvidia-up:
+	docker compose -f compose.nvidia.yaml up
 docker-local-up:
 	$(DOCKER_COMPOSE) -f compose.local.yaml up --force-recreate
 docker-local-down:

compose.nvidia.yaml

@@ -0,0 +1,181 @@
+services:
+  ## PhotoPrism (Development Environment for Nvidia)
+  photoprism:
+    build: .
+    image: photoprism/photoprism:develop
+    depends_on:
+      - mariadb
+      - dummy-webdav
+      - dummy-oidc
+    stop_grace_period: 10s
+    security_opt:
+      - seccomp:unconfined
+      - apparmor:unconfined
+    ## Expose HTTP and debug ports
+    ports:
+      - "2342:2342"   # Default HTTP port (host:container)
+      - "2443:2443"   # Default TLS port (host:container)
+      - "2343:2343"   # Acceptance Test HTTP port (host:container)
+      - "40000:40000" # Go Debugger (host:container)
+    shm_size: "2gb"
+    ## Set links and labels for use with Traefik reverse proxy
+    links:
+      - "traefik:localssl.dev"
+      - "traefik:app.localssl.dev"
+      - "traefik:keycloak.localssl.dev"
+      - "traefik:dummy-oidc.localssl.dev"
+      - "traefik:dummy-webdav.localssl.dev"
+    labels:
+      - "traefik.enable=true"
+      - "traefik.http.services.photoprism.loadbalancer.server.port=2342"
+      - "traefik.http.services.photoprism.loadbalancer.server.scheme=http"
+      - "traefik.http.routers.photoprism.entrypoints=websecure"
+      - "traefik.http.routers.photoprism.rule=Host(`localssl.dev`) || HostRegexp(`^.+\\.localssl\\.dev`)"
+      - "traefik.http.routers.photoprism.priority=2"
+      - "traefik.http.routers.photoprism.tls.domains[0].main=localssl.dev"
+      - "traefik.http.routers.photoprism.tls.domains[0].sans=*.localssl.dev"
+      - "traefik.http.routers.photoprism.tls=true"
+    ## Configure development environment
+    environment:
+      ## Run as a non-root user after initialization (supported: 0, 33, 50-99, 500-600, and 900-1200):
+      PHOTOPRISM_UID: ${UID:-1000}                         # user id, should match your host user id
+      PHOTOPRISM_GID: ${GID:-1000}                         # group id
+      ## Access Management:
+      PHOTOPRISM_ADMIN_USER: "admin"                       # admin login username
+      PHOTOPRISM_ADMIN_PASSWORD: "photoprism"              # initial admin password (8-72 characters)
+      PHOTOPRISM_AUTH_MODE: "password"                     # authentication mode (public, password)
+      PHOTOPRISM_REGISTER_URI: "https://keycloak.localssl.dev/admin/"
+      PHOTOPRISM_PASSWORD_RESET_URI: "https://keycloak.localssl.dev/realms/master/login-actions/reset-credentials"
+      ## OpenID Connect (pre-configured for local tests):
+      ## see https://keycloak.localssl.dev/realms/master/.well-known/openid-configuration
+      PHOTOPRISM_OIDC_URI: "https://keycloak.localssl.dev/realms/master"
+      PHOTOPRISM_OIDC_CLIENT: "photoprism-develop"
+      PHOTOPRISM_OIDC_SECRET: "9d8351a0-ca01-4556-9c37-85eb634869b9"
+      PHOTOPRISM_OIDC_PROVIDER: "Keycloak"
+      PHOTOPRISM_OIDC_REGISTER: "true"
+      PHOTOPRISM_OIDC_WEBDAV: "true"
+      PHOTOPRISM_DISABLE_OIDC: "false"
+      ## LDAP Authentication (pre-configured for local tests):
+      PHOTOPRISM_LDAP_URI: "ldap://dummy-ldap:389"
+      PHOTOPRISM_LDAP_INSECURE: "true"
+      PHOTOPRISM_LDAP_SYNC: "true"
+      PHOTOPRISM_LDAP_BIND: "simple"
+      PHOTOPRISM_LDAP_BIND_DN: "cn"
+      PHOTOPRISM_LDAP_BASE_DN: "dc=localssl,dc=dev"
+      PHOTOPRISM_LDAP_ROLE: ""
+      PHOTOPRISM_LDAP_ROLE_DN: "ou=photoprism-*,ou=groups,dc=localssl,dc=dev"
+      PHOTOPRISM_LDAP_WEBDAV_DN: "ou=photoprism-webdav,ou=groups,dc=localssl,dc=dev"
+      ## HTTPS/TLS Options:
+      ## see https://docs.photoprism.app/getting-started/using-https/
+      PHOTOPRISM_DISABLE_TLS: "true"
+      PHOTOPRISM_DEFAULT_TLS: "true"
+      ## Site Information:
+      PHOTOPRISM_SITE_URL: "https://app.localssl.dev/"    # server URL in the format "http(s)://domain.name(:port)/(path)"
+      PHOTOPRISM_SITE_CAPTION: "AI-Powered Photos App"
+      PHOTOPRISM_SITE_DESCRIPTION: "Tags and finds pictures without getting in your way!"
+      PHOTOPRISM_SITE_AUTHOR: "@photoprism_app"
+      PHOTOPRISM_DEBUG: "true"
+      PHOTOPRISM_READONLY: "false"
+      PHOTOPRISM_EXPERIMENTAL: "true"
+      PHOTOPRISM_HTTP_MODE: "debug"
+      PHOTOPRISM_HTTP_HOST: "0.0.0.0"
+      PHOTOPRISM_HTTP_PORT: 2342
+      PHOTOPRISM_HTTP_COMPRESSION: "gzip"                  # improves transfer speed and bandwidth utilization (none or gzip)
+      PHOTOPRISM_DATABASE_DRIVER: "mysql"
+      PHOTOPRISM_DATABASE_SERVER: "mariadb:4001"
+      PHOTOPRISM_DATABASE_NAME: "photoprism"
+      PHOTOPRISM_DATABASE_USER: "root"
+      PHOTOPRISM_DATABASE_PASSWORD: "photoprism"
+      PHOTOPRISM_TEST_DRIVER: "sqlite"
+      # PHOTOPRISM_TEST_DSN_MYSQL8: "root:photoprism@tcp(mysql:4001)/photoprism?charset=utf8mb4,utf8&collation=utf8mb4_unicode_ci&parseTime=true&timeout=15s"
+      PHOTOPRISM_ASSETS_PATH: "/go/src/github.com/photoprism/photoprism/assets"
+      PHOTOPRISM_STORAGE_PATH: "/go/src/github.com/photoprism/photoprism/storage"
+      PHOTOPRISM_ORIGINALS_PATH: "/go/src/github.com/photoprism/photoprism/storage/originals"
+      PHOTOPRISM_ORIGINALS_LIMIT: 128000             # sets originals file size limit to 128 GB
+      PHOTOPRISM_IMPORT_PATH: "/go/src/github.com/photoprism/photoprism/storage/import"
+      PHOTOPRISM_DISABLE_CHOWN: "false"              # disables updating storage permissions via chmod and chown on startup
+      PHOTOPRISM_DISABLE_BACKUPS: "false"            # disables backing up albums and photo metadata to YAML files
+      PHOTOPRISM_DISABLE_WEBDAV: "false"             # disables built-in WebDAV server
+      PHOTOPRISM_DISABLE_SETTINGS: "false"           # disables settings UI and API
+      PHOTOPRISM_DISABLE_PLACES: "false"             # disables reverse geocoding and maps
+      PHOTOPRISM_DISABLE_EXIFTOOL: "false"           # disables creating JSON metadata sidecar files with ExifTool
+      PHOTOPRISM_DISABLE_TENSORFLOW: "false"         # disables all features depending on TensorFlow
+      PHOTOPRISM_DISABLE_RAW: "false"                # disables indexing and conversion of RAW images
+      PHOTOPRISM_RAW_PRESETS: "false"                # enables applying user presets when converting RAW images (reduces performance)
+      PHOTOPRISM_DETECT_NSFW: "false"                # automatically flags photos as private that MAY be offensive (requires TensorFlow)
+      PHOTOPRISM_UPLOAD_NSFW: "false"                # allows uploads that MAY be offensive (no effect without TensorFlow)
+      PHOTOPRISM_THUMB_LIBRARY: "auto"               # image processing library to be used for generating thumbnails (auto, imaging, vips)
+      PHOTOPRISM_THUMB_FILTER: "auto"                # downscaling filter (imaging best to worst: blackman, lanczos, cubic, linear, nearest)
+      PHOTOPRISM_THUMB_UNCACHED: "true"              # enables on-demand thumbnail rendering (high memory and cpu usage)
+      TF_CPP_MIN_LOG_LEVEL: 0                        # show TensorFlow log messages for development
+      ## Nvidia Video Transcoding (https://docs.photoprism.app/getting-started/advanced/transcoding/#nvidia-container-toolkit):
+      NVIDIA_VISIBLE_DEVICES: "all"
+      NVIDIA_DRIVER_CAPABILITIES: "compute,video,utility"
+      PHOTOPRISM_FFMPEG_ENCODER: "nvidia"            # H.264/AVC encoder (software, intel, nvidia, apple, raspberry, or vaapi)
+      PHOTOPRISM_FFMPEG_SIZE: "1920"                 # video size limit in pixels (720-7680) (default: 3840)
+      PHOTOPRISM_FFMPEG_BITRATE: "50"                # video bitrate limit in Mbit/s (default: 50)
+      ## Run/install on first startup (options: update https gpu ffmpeg tensorflow davfs clitools clean):
+      PHOTOPRISM_INIT: "https tensorflow"
+    ## Share hardware devices with FFmpeg and TensorFlow (optional):
+    # devices:
+    #  - "/dev/dri:/dev/dri"                         # Intel QSV (Broadwell and later) or VAAPI (Haswell and earlier)
+    #  - "/dev/nvidia0:/dev/nvidia0"                 # Nvidia CUDA
+    #  - "/dev/nvidiactl:/dev/nvidiactl"
+    #  - "/dev/nvidia-modeset:/dev/nvidia-modeset"
+    #  - "/dev/nvidia-nvswitchctl:/dev/nvidia-nvswitchctl"
+    #  - "/dev/nvidia-uvm:/dev/nvidia-uvm"
+    #  - "/dev/nvidia-uvm-tools:/dev/nvidia-uvm-tools"
+    #  - "/dev/video11:/dev/video11"                 # Video4Linux Video Encode Device (h264_v4l2m2m)
+    working_dir: "/go/src/github.com/photoprism/photoprism"
+    volumes:
+      - ".:/go/src/github.com/photoprism/photoprism"
+      - "./storage:/photoprism"
+      - "go-mod:/go/pkg/mod"
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: "nvidia"
+              count: 1
+              capabilities: [ gpu ]
+  mariadb:
+    extends:
+      file: ./compose.yaml
+      service: mariadb
+  traefik:
+    extends:
+      file: ./compose.yaml
+      service: traefik
+  dummy-webdav:
+    extends:
+      file: ./compose.yaml
+      service: dummy-webdav
+  dummy-oidc:
+    extends:
+      file: ./compose.yaml
+      service: dummy-oidc
+  dummy-ldap:
+    extends:
+      file: ./compose.yaml
+      service: dummy-ldap
+  keycloak:
+    extends:
+      file: ./compose.yaml
+      service: keycloak
+  prometheus:
+    extends:
+      file: ./compose.yaml
+      service: prometheus
+
+## Create named volume for Go module cache
+volumes:
+  go-mod:
+    driver: local
+  mariadb:
+    driver: local
+
+## Create shared "photoprism-develop" network for connecting with services in other compose.yaml files
+networks:
+  default:
+    name: photoprism
+    driver: bridge

compose.yaml

@@ -14,12 +14,14 @@ services:
     security_opt:
       - seccomp:unconfined
       - apparmor:unconfined
+    ## Expose HTTP and debug ports
     ports:
       - "2342:2342"   # Default HTTP port (host:container)
       - "2443:2443"   # Default TLS port (host:container)
       - "2343:2343"   # Acceptance Test HTTP port (host:container)
       - "40000:40000" # Go Debugger (host:container)
     shm_size: "2gb"
+    ## Set links and labels for use with Traefik reverse proxy
     links:
       - "traefik:localssl.dev"
       - "traefik:app.localssl.dev"
@@ -36,6 +38,11 @@ services:
       - "traefik.http.routers.photoprism.tls.domains[0].main=localssl.dev"
       - "traefik.http.routers.photoprism.tls.domains[0].sans=*.localssl.dev"
       - "traefik.http.routers.photoprism.tls=true"
+    ## Override variables with optional env file, see https://docs.docker.com/reference/compose-file/services/#required
+    env_file:
+      - path: ".env"
+        required: false
+    ## Configure development environment
     environment:
       ## Run as a non-root user after initialization (supported: 0, 33, 50-99, 500-600, and 900-1200):
       PHOTOPRISM_UID: ${UID:-1000}                         # user id, should match your host user id

internal/config/config_ffmpeg.go

@@ -20,8 +20,10 @@ func (c *Config) FFmpegEnabled() bool {
 
 // FFmpegEncoder returns the FFmpeg AVC encoder name.
 func (c *Config) FFmpegEncoder() encode.Encoder {
-	if c.options.FFmpegEncoder == "" || c.options.FFmpegEncoder == encode.SoftwareAvc.String() {
+	if c.options.FFmpegEncoder == encode.SoftwareAvc.String() {
 		return encode.SoftwareAvc
+	} else if c.options.FFmpegEncoder == "" {
+		return encode.DefaultAvcEncoder()
 	}
 
 	return encode.FindEncoder(c.options.FFmpegEncoder)

internal/config/config_ffmpeg_test.go

@@ -12,7 +12,7 @@ import (
 
 func TestConfig_FFmpegEncoder(t *testing.T) {
 	c := NewConfig(CliTestContext())
-	assert.Equal(t, encode.SoftwareAvc, c.FFmpegEncoder())
+	assert.Equal(t, encode.DefaultAvcEncoder(), c.FFmpegEncoder())
 	c.options.FFmpegEncoder = "nvidia"
 	assert.Equal(t, encode.NvidiaAvc, c.FFmpegEncoder())
 	c.options.FFmpegEncoder = "intel"
@@ -20,7 +20,7 @@ func TestConfig_FFmpegEncoder(t *testing.T) {
 	c.options.FFmpegEncoder = "xxx"
 	assert.Equal(t, encode.SoftwareAvc, c.FFmpegEncoder())
 	c.options.FFmpegEncoder = ""
-	assert.Equal(t, encode.SoftwareAvc, c.FFmpegEncoder())
+	assert.Equal(t, encode.DefaultAvcEncoder(), c.FFmpegEncoder())
 }
 
 func TestConfig_FFmpegEnabled(t *testing.T) {

internal/entity/search/photos.go

@@ -555,11 +555,16 @@ func searchPhotos(frm form.SearchPhotos, sess *entity.Session, resultCols string
 		s = s.Where("files.file_aspect_ratio = 1")
 	}
 
-	// Filter by main color.
-	if frm.Color != "" {
+	// Filter by file main color.
+	if txt.NotEmpty(frm.Color) {
 		s = s.Where("files.file_main_color IN (?)", SplitOr(strings.ToLower(frm.Color)))
 	}
 
+	// Filter by file codec.
+	if txt.NotEmpty(frm.Codec) {
+		s = s.Where("files.file_codec IN (?)", SplitOr(strings.ToLower(frm.Codec)))
+	}
+
 	// Filter by chroma.
 	if frm.Mono {
 		s = s.Where("files.file_chroma = 0")
@@ -684,7 +689,7 @@ func searchPhotos(frm form.SearchPhotos, sess *entity.Session, resultCols string
 		s = s.Where(where, values...)
 	}
 
-	// Filter by hash.
+	// Filter by file hash.
 	if txt.NotEmpty(frm.Hash) {
 		s = s.Where("files.file_hash IN (?)", SplitOr(strings.ToLower(frm.Hash)))
 	}

internal/entity/search/photos_geo.go

@@ -461,11 +461,16 @@ func UserPhotosGeo(frm form.SearchPhotosGeo, sess *entity.Session) (results GeoR
 		s = s.Where("files.file_aspect_ratio = 1")
 	}
 
-	// Filter by main color.
-	if frm.Color != "" {
+	// Filter by file main color.
+	if txt.NotEmpty(frm.Color) {
 		s = s.Where("files.file_main_color IN (?)", SplitOr(strings.ToLower(frm.Color)))
 	}
 
+	// Filter by file codec.
+	if txt.NotEmpty(frm.Codec) {
+		s = s.Where("files.file_codec IN (?)", SplitOr(strings.ToLower(frm.Codec)))
+	}
+
 	// Filter by chroma.
 	if frm.Mono {
 		s = s.Where("files.file_chroma = 0")

internal/ffmpeg/encode/default.go

@@ -0,0 +1,31 @@
+package encode
+
+import (
+	"os"
+	"strings"
+
+	"github.com/photoprism/photoprism/pkg/fs"
+)
+
+// defaultAvcEncoder is the default FFmpeg AVC encoder if it has already been determined.
+var defaultAvcEncoder = Encoder("")
+
+// DefaultAvcEncoder determines and returns the default FFmpeg AVC encoder type:
+func DefaultAvcEncoder() Encoder {
+	if defaultAvcEncoder != "" {
+		return defaultAvcEncoder
+	}
+
+	switch {
+	// Default to Nvidia AVC encoder if the NVIDIA_DRIVER_CAPABILITIES variable is set and contains "video":
+	case fs.DeviceExists("/dev/nvidia0") &&
+		strings.Contains(os.Getenv("NVIDIA_DRIVER_CAPABILITIES"), "video") &&
+		!strings.Contains(os.Getenv("PHOTOPRISM_INIT"), "ffmpeg"):
+		defaultAvcEncoder = NvidiaAvc
+	// Otherwise, use the standard software AVC encoder:
+	default:
+		defaultAvcEncoder = SoftwareAvc
+	}
+
+	return defaultAvcEncoder
+}

internal/ffmpeg/encode/encoder.go

@@ -1,6 +1,8 @@
 package encode
 
-import "github.com/photoprism/photoprism/pkg/clean"
+import (
+	"github.com/photoprism/photoprism/pkg/clean"
+)
 
 // Encoder represents a supported FFmpeg AVC encoder name.
 type Encoder string

internal/form/search_photos.go

@@ -61,6 +61,7 @@ type SearchPhotos struct {
 	Mm        string    `form:"mm" example:"mm:28-35" notes:"Focal Length (35mm equivalent)"`
 	F         string    `form:"f" example:"f:2.8-4.5" notes:"Aperture (f-number)"`
 	Color     string    `form:"color" example:"color:\"red|blue\"" notes:"Color Name (purple, magenta, pink, red, orange, gold, yellow, lime, green, teal, cyan, blue, brown, white, grey, black) (separate with |)"` // Main color
+	Codec     string    `form:"codec" example:"codec:avc1" notes:"Media Codec (e.g. jpeg, avc1, hvc1); separate with |"`
 	Chroma    int16     `form:"chroma" example:"chroma:70" notes:"Chroma (0-100)"`
 	Mono      bool      `form:"mono" notes:"Finds pictures with few or no colors"`
 	Diff      uint32    `form:"diff" notes:"Differential Perceptual Hash (000000-FFFFFF)"`

internal/form/search_photos_geo.go

@@ -13,7 +13,7 @@ type SearchPhotosGeo struct {
 	Filter    string    `form:"filter" serialize:"-" notes:"-"`
 	ID        string    `form:"id" example:"id:123e4567-e89b-..." notes:"Finds pictures by Exif UID, XMP Document ID or Instance ID"`
 	UID       string    `form:"uid" example:"uid:pqbcf5j446s0futy" notes:"Limits results to the specified internal unique IDs"`
-	Type      string    `form:"type"`
+	Type      string    `form:"type" example:"type:raw" notes:"Media Type (image, video, raw, live, animated); separate with |"`
 	Path      string    `form:"path"`
 	Folder    string    `form:"folder"` // Alias for Path
 	Name      string    `form:"name"`
@@ -63,6 +63,7 @@ type SearchPhotosGeo struct {
 	Mm        string    `form:"mm" example:"mm:28-35" notes:"Focal Length (35mm equivalent)"`
 	F         string    `form:"f" example:"f:2.8-4.5" notes:"Aperture (f-number)"`
 	Color     string    `form:"color"`
+	Codec     string    `form:"codec" example:"codec:avc1" notes:"Media Codec (e.g. jpeg, avc1, hvc1); separate with |"`
 	Chroma    int16     `form:"chroma" example:"chroma:70" notes:"Chroma (0-100)"`
 	Mono      bool      `form:"mono" notes:"Finds pictures with few or no colors"`
 	Person    string    `form:"person"`   // Alias for Subject

pkg/fs/fs.go

@@ -106,6 +106,23 @@ func PathExists(path string) bool {
 	return m&os.ModeDir != 0 || m&os.ModeSymlink != 0
 }
 
+// DeviceExists tests if a path exists, and is a device.
+func DeviceExists(path string) bool {
+	if path == "" {
+		return false
+	}
+
+	info, err := os.Stat(path)
+
+	if err != nil {
+		return false
+	}
+
+	m := info.Mode()
+
+	return m&os.ModeDevice != 0 || m&os.ModeCharDevice != 0
+}
+
 // Writable checks if the path is accessible for reading and writing.
 func Writable(path string) bool {
 	if path == "" {

pkg/fs/fs_test.go

@@ -44,6 +44,11 @@ func TestPathExists(t *testing.T) {
 	assert.False(t, PathExists(""))
 }
 
+func TestDeviceExists(t *testing.T) {
+	assert.True(t, DeviceExists("/dev/null"))
+	DeviceExists("/dev/nvidia0")
+}
+
 func TestPathWritable(t *testing.T) {
 	assert.True(t, PathWritable("./testdata"))
 	assert.False(t, PathWritable("./testdata/test.jpg"))