diff --git a/_sql/actions_and_analytics/backup/_inventory_media_files.sql b/_sql/actions_and_analytics/backup/_inventory_media_files.sql new file mode 100644 index 0000000000..009c8f7db2 --- /dev/null +++ b/_sql/actions_and_analytics/backup/_inventory_media_files.sql @@ -0,0 +1,196 @@ +-- TODO: TOP maybe_creator_user_token + +-- TODO: maybe_style_transfer_source_media_file_token (original videos) + +-- TODO: Later: media_type = (mp4, video) + +-- origin_product_category +select distinct origin_category, +count(*) +from media_files +group by origin_category; + +-- +-----------------+----------+ +-- | origin_category | count(*) | +-- +-----------------+----------+ +-- | inference | 40933502 | +-- | processed | 365 | +-- | story_engine | 688 | +-- | studio | 148 | +-- | upload | 796516 | +-- +-----------------+----------+ +-- 5 rows in set (54.69 sec) -- TODO [QUERY DONE]: Get origin_category = `story_engine`, `studio`, and `processed` (???) + + +-- origin_product_category +select distinct origin_product_category, +count(*) +from media_files +group by origin_product_category; + +-- +-------------------------+----------+ +-- | origin_product_category | count(*) | +-- +-------------------------+----------+ +-- | face_animator | 145770 | +-- | face_fusion | 1524 | +-- | face_mirror | 12743 | +-- | image_gen | 450196 | +-- | mocap | 436 | +-- | studio | 148 | +-- | tts | 33757597 | +-- | unknown | 94182 | +-- | voice_conversion | 6664633 | +-- | workflow | 46011 | +-- | zs_voice | 558212 | +-- +-------------------------+----------+ +-- 11 rows in set (45.88 sec) +-- TODO [QUERIES DONE(2)]: Get origin_product_category = `mocap`, `studio`, `workflow`, `face_mirror`, `face_fusion` + + +-- maybe_origin_model_type (+token) +select distinct maybe_origin_model_type, +count(*) +from media_files +group by maybe_origin_model_type; + +-- +-------------------------+----------+ +-- | maybe_origin_model_type | count(*) | +-- +-------------------------+----------+ +-- | NULL | 347531 | +-- | comfy_ui | 46013 | +-- | face_fusion | 1524 | +-- | gpt_sovits | 304046 | +-- | live_portrait | 12743 | +-- | mocap_net | 71 | +-- | rvc_v2 | 6107295 | +-- | sad_talker | 156308 | +-- | so_vits_svc | 557377 | +-- | stable_diffusion_1_5 | 450196 | +-- | styletts2 | 263575 | +-- | tacotron2 | 33453882 | +-- | vall_e_x | 31188 | +-- +-------------------------+----------+ +-- 13 rows in set (51.98 sec) +-- TODO [DONEx2]: maybe_origin_model_type = comfy_ui, mocap_net, face_fusion, live_portrait + + +-- media_type +select distinct media_type, +count(*) +from media_files +group by media_type; + +-- +------------+----------+ +-- | media_type | count(*) | +-- +------------+----------+ +-- | audio | 40676557 | +-- | bvh | 107 | +-- | fbx | 325 | +-- | gif | 7 | +-- | glb | 3100 | +-- | gltf | 45 | +-- | image | 481274 | +-- | jpg | 91 | +-- | mp4 | 13280 | +-- | pmd | 11 | +-- | pmx | 239 | +-- | png | 208 | +-- | scene_json | 4562 | +-- | scene_ron | 688 | +-- | video | 247391 | +-- | vmd | 168 | +-- | wav | 303996 | +-- +------------+----------+ +-- 17 rows in set (43.68 sec) +-- TODO [DONE WITH QUERY]: media_type = bvh, fbx, gif, glb, gltf, jpg, png, pmx, pmd, scene_json, scene_ron, vmd +-- TODO: Later: media_type = (mp4, video) + + +-- media_class +select distinct media_class, +count(*) +from media_files +group by media_class; + +-- +-------------+----------+ +-- | media_class | count(*) | +-- +-------------+----------+ +-- | audio | 40786861 | +-- | dimensional | 7846 | +-- | image | 30446 | +-- | unknown | 653032 | +-- | video | 254270 | +-- +-------------+----------+ +-- 5 rows in set (46.78 sec) +-- TODO [QUERY DONE]: media_class = dimensional + + +-- maybe_media_subtype +select distinct maybe_media_subtype, +count(*) +from media_files +group by maybe_media_subtype; + +-- +---------------------+----------+ +-- | maybe_media_subtype | count(*) | +-- +---------------------+----------+ +-- | NULL | 41730497 | +-- | mixamo | 950 | +-- | storyteller_scene | 766 | +-- | scene_import | 385 | +-- | mocap_net | 2 | +-- | animation_only | 15 | +-- | object | 1 | +-- +---------------------+----------+ +-- 7 rows in set (25.48 sec) +-- TODO [QUERY DONE]: maybe_media_subtype IS NOT NULL + + +-- maybe_animation_type +select distinct maybe_animation_type, +count(*) +from media_files +group by maybe_animation_type; + +-- +----------------------+----------+ +-- | maybe_animation_type | count(*) | +-- +----------------------+----------+ +-- | NULL | 41732552 | +-- | mixamo | 360 | +-- | mocap_net_ar_kit | 1 | +-- | ar_kit | 41 | +-- | move_ai | 46 | +-- | move_ai_ar_kit | 7 | +-- | rigify | 7 | +-- | mixamo_ar_kit | 114 | +-- | miku_miku_dance | 290 | +-- | mocap_net | 2 | +-- +----------------------+----------+ +-- 10 rows in set (27.46 sec) +-- TODO [QUERY DONE]: maybe_animation_type IS NOT NULL + + +-- maybe_engine_category +select distinct maybe_engine_category, +count(*) +from media_files +group by maybe_engine_category; + +-- +-----------------------+----------+ +-- | maybe_engine_category | count(*) | +-- +-----------------------+----------+ +-- | NULL | 41729660 | +-- | animation | 344 | +-- | character | 707 | +-- | creature | 71 | +-- | expression | 45 | +-- | image_plane | 306 | +-- | location | 105 | +-- | object | 694 | +-- | scene | 4493 | +-- | set_dressing | 44 | +-- | skybox | 1 | +-- | video_plane | 77 | +-- +-----------------------+----------+ +-- 12 rows in set (36.41 sec) +-- TODO [QUERY DONE]: maybe_engine_category IS NOT NULL diff --git a/_sql/actions_and_analytics/backup/backup_cover_images.sql b/_sql/actions_and_analytics/backup/backup_cover_images.sql new file mode 100644 index 0000000000..8313b65463 --- /dev/null +++ b/_sql/actions_and_analytics/backup/backup_cover_images.sql @@ -0,0 +1,12 @@ + +-- Media file cover images +select distinct maybe_cover_image_media_file_token +from media_files +where maybe_cover_image_media_file_token is not null +group by maybe_cover_image_media_file_token + +-- Model weights cover images +select distinct maybe_cover_image_media_file_token +from model_weights +where maybe_cover_image_media_file_token is not null +group by maybe_cover_image_media_file_token diff --git a/_sql/actions_and_analytics/backup/backup_studio_files.sql b/_sql/actions_and_analytics/backup/backup_studio_files.sql new file mode 100644 index 0000000000..c5c70c7f7a --- /dev/null +++ b/_sql/actions_and_analytics/backup/backup_studio_files.sql @@ -0,0 +1,59 @@ + +-- STUDIO_maybe_media_subtype_not_null.csv +select token +from media_files +where maybe_media_subtype IS NOT NULL; + +-- STUDIO_dimensional_media_class.csv +select token +from media_files +where media_class = 'dimensional'; + +-- STUDIO_maybe_animation_type_not_null.csv +select token +from media_files +where maybe_animation_type IS NOT NULL; + +-- STUDIO_IMPORTANT_maybe_engine_category_not_null.csv +select token +from media_files +where maybe_engine_category IS NOT NULL; + +-- STUDIO_IMPORTANT_scene_source_token_not_null.csv +select distinct maybe_scene_source_media_file_token +from media_files +where maybe_scene_source_media_file_token IS NOT NULL +group by maybe_scene_source_media_file_token; + +-- STUDIO_media_file_origin_categories.csv +select token +from media_files +where origin_category IN ('processed', 'studio', 'story_engine'); + +-- STUDIO_media_file_studio_asset_product_category.csv +select token +from media_files +where origin_product_category IN ('mocap', 'studio', 'workflow'); + +-- VIDEO_media_file_video_product_category.csv +select token +from media_files +where origin_product_category IN ('face_mirror', 'face_fusion'); + +-- VIDEO_media_file_video_origin_model_type.csv +select token +from media_files +where maybe_origin_model_type IN ('comfy_ui', 'face_fusion', 'live_portrait'); + +-- STUDIO_media_file_mocap_model_type.csv +select token +from media_files +where maybe_origin_model_type IN ('mocap_net'); + +select token +from media_files +where media_type IN ('bvh', 'fbx', 'gif', 'glb', 'gltf', 'jpg', 'pmd', 'pmx', 'png', 'scene_json', 'scene_ron', 'vmd'); + +select token +from media_files +where media_type IN ('mp4', 'video'); diff --git a/_sql/actions_and_analytics/backup/backup_studio_intermediate_videos.sql b/_sql/actions_and_analytics/backup/backup_studio_intermediate_videos.sql new file mode 100644 index 0000000000..fabfe2c996 --- /dev/null +++ b/_sql/actions_and_analytics/backup/backup_studio_intermediate_videos.sql @@ -0,0 +1,4 @@ + +select token +from media_files +where is_intermediate_system_file IS TRUE