Add 2D-to-3D market and product technical reports

.gitattributes

@@ -34,3 +34,141 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 基于多阶段深度变换与修补的2D视频转3D视频项目文档.docx filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/accenture_ai_spatial_computing_workforce_2024.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/accenture_metaverse_evolution_before_revolution_2023.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/accenture_technology_vision_2025.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/autodesk_state_design_make_2025.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/citi_metaverse_and_money_2022.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/deloitte_future_of_media_and_entertainment_2024.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/deloitte_state_of_gen_ai_enterprise_2024.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/deloitte_virtual_production_future_content_creation_2024.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/01_market_size/accenture_2023_p10_metaverse_value.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/01_market_size/citi_2022_p12_metaverse_tam.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/01_market_size/mckinsey_2022_p37_metaverse_impact.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/01_market_size/pwc_2025_p05_global_em_revenue.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/01_market_size/pwc_2025_p07_revenue_structure.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/01_market_size/pwc_2025_p16_ott_market.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/01_market_size/pwc_2026_p01_global_em_2030.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/01_market_size/pwc_2026_p05_sphere_immersive.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/01_market_size/pwc_2026_p06_cinema_live_experiences.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/01_market_size/pwc_2026_p08_global_ott.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/02_cost_value/autodesk_2025_p07_digital_transformation_roi.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/02_cost_value/deloitte_2024_p07_virtual_production_cost.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/02_cost_value/morgan_stanley_2025_p01_cost_estimate.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/02_cost_value/morgan_stanley_2025_p03_tv_film_savings.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/02_cost_value/movielabs_2026_p146_jiostar_case.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/02_cost_value/movielabs_2026_p148_jiostar_outcomes.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/02_cost_value/netflix_2026_p12_ai_cost_case.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/02_cost_value/wef_2025_p10_ai_creative_value.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/02_cost_value/wef_2025_p14_ai_productivity_cost.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/02_cost_value/wef_2025_p15_ai_content_creation.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/03_competitor_product/primefocus_2026_p04_business_and_products.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/03_competitor_product/primefocus_2026_p09_stereo_conversion_capability.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/03_competitor_product/primefocus_2026_p14_immersive_business.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/key_charts/03_competitor_product/primefocus_2026_p17_content_tam.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/market_chapter_assets/figure3_global_em_growth.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/market_chapter_assets/figure4_application_map.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/market_chapter_assets/figure5_dneg_capabilities_cn.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/market_chapter_assets/figure6_cost_evidence.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/mckinsey_value_creation_in_the_metaverse_2022.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/morgan_stanley_ai_entertainment_costs_2025_official_web_print.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/movielabs_2030_vision_evolution_media_creation.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/movielabs_industry_forum_spring_summit_2026.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/movielabs_interoperability_media_creation.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/movielabs_software_defined_workflows.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/netflix_q2_2026_earnings_call_transcript_ai_cost.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/netflix_q2_2026_shareholder_letter_ai_production.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/oecd_immersive_technologies_policy_primer_2025.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/primefocus_investor_presentation_jan_2026.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/pwc_global_entertainment_media_outlook_2025_2029.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/pwc_global_entertainment_media_outlook_2026_2030_official_web_print.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/salesforce_ai_media_entertainment_redefined.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_market_tech_reports/wef_ai_in_media_entertainment_sport_2025.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/01_direct_conversion_products/acer_spatiallabs_official_web_report_print_2026.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/01_direct_conversion_products/immersity_ai_technical_backgrounder_2024.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/01_direct_conversion_products/lenovo_thinkvision_27_3d_product_datasheet_2025.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/01_direct_conversion_products/samsung_odyssey_3d_g90xf_product_datasheet_2025.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/02_display_and_delivery_products/apple_hevc_stereo_video_interoperability_profile_2023.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/02_display_and_delivery_products/apple_stereo_video_isobmff_extensions_2025.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/02_display_and_delivery_products/continental_post_ces_2023_3d_display_technology_one_pagers.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/02_display_and_delivery_products/samsung_display_solutions_spatial_signage_catalog_2026.pdf filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/acer_spatiallabs/page-04.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/acer_spatiallabs/page-05.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/apple_stereo_delivery/page-09.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/apple_stereo_delivery/page-10.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/apple_stereo_delivery/page-11.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/apple_stereo_delivery/page-13.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/continental_3d_display/page-04.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/immersity_backgrounder/page-1.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/immersity_backgrounder/page-2.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/immersity_backgrounder/page-3.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/lenovo_thinkvision_27_3d/page-1.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/lenovo_thinkvision_27_3d/page-2.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/official_web_diagrams/acer_spatiallabs_go_product_pipeline.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/official_web_diagrams/asus_spatial_vision_eye_tracking_section.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/official_web_diagrams/asus_spatial_vision_optical_stack_section.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/samsung_odyssey_3d/page-1.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/samsung_odyssey_3d/page-2.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/samsung_odyssey_3d/page-3.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/samsung_spatial_signage/page-04.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/key_figures/samsung_spatial_signage/page-05.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/asus_spatial_vision_technology_2023/assets/01_og_image_97ff017213.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/asus_spatial_vision_technology_2023/assets/13_source_409909bf19.webp filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/asus_spatial_vision_technology_2023/assets/14_source_fa6dc1aa6a.webp filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/asus_spatial_vision_technology_2023/assets/15_image_7a13b8517d.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/asus_spatial_vision_technology_2023/assets/16_source_4579c8b9f7.webp filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/dneg_godzilla_x_kong_stereo_conversion_case_2024/assets/01_og_image_08c8694bce.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/dneg_godzilla_x_kong_stereo_conversion_case_2024/assets/02_image_92dd997ace.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/foundry_ocula_product_technology/assets/11_Get_online_help_and_user_guides_for_Ocula_the_st_294d14a3ea.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/foundry_ocula_product_technology/assets/12_Get_going_with_free_online_tutorials_from_Foundr_3e903d36e8.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/foundry_ocula_product_technology/assets/13_Find_out_what_you_need_to_know_from_experts_in_t_48f6d33db4.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_4_product_release_2025/assets/01_og_image_3b3317e999.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_4_product_release_2025/assets/03_image_d606b7a41a.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_4_product_release_2025/assets/04_image_4a307c821d.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_4_product_release_2025/assets/05_image_accbbd4c1e.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_4_product_release_2025/assets/06_image_29148429fb.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_4_product_release_2025/assets/07_image_bd7836f5c1.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_product_launch_technical_overview_2024/assets/01_og_image_9c8add9f60.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_product_launch_technical_overview_2024/assets/03_image_f483e057ee.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_product_launch_technical_overview_2024/assets/04_image_4a307c821d.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_product_launch_technical_overview_2024/assets/05_image_accbbd4c1e.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_product_launch_technical_overview_2024/assets/06_image_29148429fb.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/immersity_product_launch_technical_overview_2024/assets/07_image_bd7836f5c1.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/lenovo_thinkvision_27_3d_product_technology/assets/02_image_558cc71b02.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/lenovo_thinkvision_27_3d_product_technology/assets/05_image_1111f3b94b.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/lenovo_thinkvision_27_3d_product_technology/assets/06_image_0dc18ebce3.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/lenovo_thinkvision_27_3d_product_technology/assets/07_image_34d2bd2ef0.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/01_og_image_13fe37066b.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/12_image_21ff287ce5.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/16_image_766e1d4ab7.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/18_image_487959a9b5.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/19_image_06396b6f0f.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/20_image_5136f01d32.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/21_image_2da98b76e9.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/22_image_f774a57bf4.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/23_image_eca807d79e.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/24_image_cca2957a8c.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/25_image_58e6dd0211.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/looking_glass_product_software_stack/assets/26_image_baa7e94a3e.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/01_og_image_aa3a37ffb2.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/02_image_1b229b7ae9.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/04_image_8be35802fd.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/05_image_5fe2b3dd9c.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/06_image_dec945cb71.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/07_image_c7175b3768.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/08_image_141cd54a4e.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/09_image_75343a9bb2.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/10_image_047f18a644.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/11_image_1858d69cab.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/12_image_23451d9569.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_odyssey_3d_engineering_interview_2025/assets/13_image_67689ec55f.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_spatial_signage_technology_2026/assets/09_4_ways_Samsung_The_Wall_MMF-A_All-in-One_LED_dis_c21d4adb86.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_spatial_signage_technology_2026/assets/10_A_woman_stands_in_a_sunlit_rustic_office_with_ex_d3a0cfe338.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_spatial_signage_technology_2026/assets/11_A_woman_stands_in_a_sunlit_rustic_office_with_ex_c1400d16fe.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_spatial_signage_technology_2026/assets/13_How_interactive_displays_help_enterprises_reduce_428c370c20.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_spatial_signage_technology_2026/assets/15_Why_government_meeting_rooms_need_bright_vivid_d_830d0f8cb7.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_spatial_signage_technology_2026/assets/17_4_ways_to_design_a_modern_office_for_collaborati_df3394ab78.jpg filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/samsung_spatial_signage_technology_2026/assets/22_digital_signage_category_488969a584.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/sony_spatial_reality_display_whitepaper_2025/assets/01_og_image_f4560179b3.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/xreal_real_3d_engineering_report_2026/assets/07_XREAL_Beam_Pro_5603446ef6.png filter=lfs diff=lfs merge=lfs -text
+2d_to_3d_verified_product_technical_reports/official_web_archives/xreal_real_3d_engineering_report_2026/assets/08_image_d657ab031a.png filter=lfs diff=lfs merge=lfs -text

2d_to_3d_market_tech_reports/README.md

@@ -0,0 +1,210 @@
+# 2D 视频转 3D / 空间视频市场调研与技术报告索引
+
+整理日期：2026-07-22
+
+## 项目文档匹配点
+
+你的项目文档定位是“基于多阶段深度估计、视图变换与视频修补的双目视频生成系统”，核心链路是：
+
+`Depth -> Warp -> Inpaint -> 左右眼视频 / SBS / 空间视频`
+
+因此，这批资料按以下问题筛选：
+
+1. 3D 电影、XR、空间计算、沉浸式内容、个人消费市场是否有足够市场空间。
+2. 大厂和行业组织是否在推动 AI/云端/自动化媒体生产技术。
+3. 自动化生成、虚拟制作、AI 后期是否有明确节省时间和成本的证据。
+4. 是否存在与 2D-to-3D、stereo conversion、immersive content creation 接近的竞品或产业落地。
+
+## 本次新增的可用成果
+
+- `market_evidence_matrix.csv`：逐条记录数字、PDF 页码、图表文件、项目用途和引用限制。
+- `report_to_project_mapping.md`：把市场总盘、相邻市场、直接竞品和降本证据映射到项目文档章节。
+- `key_charts/`：从原始报告导出的 26 张关键页，分为市场规模、降本价值、竞品/产品三类。
+- `autodesk_state_design_make_2025.pdf`：包含数字化转型、成本控制和 M&E 的 AI 影响调查图表。
+- `morgan_stanley_ai_entertainment_costs_2025_official_web_print.pdf`：Morgan Stanley 官方文章本地打印件，含媒体约 10%、电视/电影最高约 30% 的降本估计。
+- `pwc_global_entertainment_media_outlook_2026_2030_official_web_print.pdf`：PwC 2026-2030 官方网页本地打印件，含 2030 年 4.2 万亿美元 E&M 和 3040 亿美元 OTT 预测。
+
+其中两个 `official_web_print` 文件用于保存官方网页内容，不是厂商发布的原生 PDF，引用时应链接原始官方网页。
+
+## 最建议优先引用的近年资料
+
+### 1. Netflix Q2 2026 Earnings Call Transcript
+
+- 文件：`netflix_q2_2026_earnings_call_transcript_ai_cost.pdf`
+- 年份：2026
+- 类型：大厂财报电话会文字稿
+- 适合引用点：Netflix 管理层提到 `American Experiment` 中 17 分钟 AI-enhanced footage，以“两倍速度、半成本”完成。还提到 AI 已用于 set references、pre-vis、VFX、sequence prep、shot planning 等制作环节。
+- 项目对应价值：可支撑“AI/生成式后期技术能缩短内容制作周期、降低单位内容成本”。虽然不是 2D 转 3D 专项报告，但和本项目的 Inpaint/VFX/后期自动化价值最贴近。
+
+### 2. Deloitte - The Future of Content Creation: Virtual Production
+
+- 文件：`deloitte_virtual_production_future_content_creation_2024.pdf`
+- 年份：2024
+- 类型：影视制作技术/商业价值报告
+- 适合引用点：报告指出 Hollywood 传统制作方式接近效率上限，虚拟制作可带来显著时间和成本节省；高预算电影重拍常见，可占最终制作成本 5%-20% 甚至更高；高预算科幻/奇幻片 VFX 成本可达总预算 20%。
+- 项目对应价值：可用于论证“减少原生 3D 拍摄、重拍和复杂后期的人力/周期成本”，与 2D 存量视频自动立体化的成本价值直接相关。
+
+### 3. WEF - Artificial Intelligence in Media, Entertainment and Sport
+
+- 文件：`wef_ai_in_media_entertainment_sport_2025.pdf`
+- 年份：2025
+- 类型：国际组织产业报告
+- 适合引用点：报告系统整理 AI 在内容创作、资产管理、商业化、受众体验中的机会，并明确讨论 GenAI 提升生产力、降低成本。
+- 项目对应价值：可作为“媒体娱乐行业正在把生成式 AI 纳入生产流程”的权威背景材料。
+
+### 4. Prime Focus / DNEG Investor Presentation
+
+- 文件：`primefocus_investor_presentation_jan_2026.pdf`
+- 年份：2026
+- 类型：竞品/产业公司投资者材料
+- 适合引用点：Prime Focus/DNEG 声明其服务能力覆盖 Stereo Conversion、VFX、Immersive & Experiential、Production、Post-production 等；材料还提到 2030 年超过 1300 亿美元的可寻址市场机会。
+- 项目对应价值：这是最像“竞品/产业落地”的材料，可证明 stereo conversion 仍是大型 VFX 公司商业能力的一部分。
+
+### 5. MovieLabs Industry Forum Spring Summit
+
+- 文件：`movielabs_industry_forum_spring_summit_2026.pdf`
+- 年份：2026
+- 类型：好莱坞行业组织技术论坛材料
+- 适合引用点：包含 2030 Vision 在 AI 时代的落地、专业媒体创作模型训练、真实部署案例；其中云端制作案例写到 30%+ cost effective、85% lesser IT staff。
+- 项目对应价值：可作为“好莱坞技术路线正在转向云端、AI、自动化工作流”的产业依据。
+
+### 6. PwC Global Entertainment & Media Outlook
+
+- 文件：`pwc_global_entertainment_media_outlook_2025_2029.pdf`
+- 补充文件：`pwc_global_entertainment_media_outlook_2026_2030_official_web_print.pdf`
+- 年份：2025
+- 类型：娱乐传媒市场报告摘要
+- 适合引用点：全球娱乐传媒收入预计 2029 年达到约 3.5 万亿美元。PwC 2026 官方网页还进一步给出 2030 年 4.2 万亿美元预测。
+- 项目对应价值：支撑“电影、长视频、OTT、广告、沉浸式体验属于万亿级内容经济的一部分”。
+- 直接下游数字：全球电影票房预计 2030 年达到 395 亿美元；全球 OTT 收入预计达到 3040 亿美元；Sphere 2025 年收入 7.81 亿美元并包含 4D 电影放映。
+- 2026 网页来源：https://www.pwc.com/gx/en/issues/business-model-reinvention/outlook/insights-and-perspectives.html
+
+### 7. Morgan Stanley - How AI Benefits—and Threatens—the Entertainment Industry
+
+- 文件：`morgan_stanley_ai_entertainment_costs_2025_official_web_print.pdf`
+- 年份：2025
+- 类型：投行行业分析官方网页打印件
+- 适合引用点：估计 GenAI 可使媒体行业成本下降约 10%，电视和电影最多约 30%；还讨论动画、VFX、数字/虚拟制作和减少重拍。
+- 使用限制：这是广义 GenAI 行业估计，不是 2D 转 3D 产品实测。
+
+### 8. Autodesk - 2025 State of Design & Make
+
+- 文件：`autodesk_state_design_make_2025.pdf`
+- 年份：2025
+- 类型：大厂全球行业调查报告
+- 适合引用点：5594 名行业领导者调查，包含数字化转型改善幅度、成本控制优先级及 M&E 的 AI 影响图表。
+- 使用限制：调查覆盖多个 Design & Make 行业；“50%+”是受访者报告的改善分档，不应写成已审计财务收益率。
+
+## 市场规模背景资料
+
+### McKinsey - Value Creation in the Metaverse
+
+- 文件：`mckinsey_value_creation_in_the_metaverse_2022.pdf`
+- 年份：2022
+- 类型：市场调研/咨询报告
+- 适合引用点：McKinsey 估计 metaverse 到 2030 年可产生最高 5 万亿美元影响。
+- 使用建议：不是最新技术报告，但适合用于“沉浸式/空间内容长期市场上限”的背景论证。
+
+### Citi GPS - Metaverse and Money
+
+- 文件：`citi_metaverse_and_money_2022.pdf`
+- 年份：2022
+- 类型：市场调研/金融机构报告
+- 适合引用点：Citi 估计 2030 年 metaverse TAM 可达 8-13 万亿美元，潜在用户可达约 50 亿。
+- 使用建议：适合与 McKinsey 交叉引用，说明市场空间预测区间很大，但多家机构都认为上限巨大。
+
+### Accenture - Metaverse: Evolution, Then Revolution
+
+- 文件：`accenture_metaverse_evolution_before_revolution_2023.pdf`
+- 年份：2023
+- 类型：咨询报告
+- 适合引用点：Accenture 调研称企业预计 4.2% 收入来自 metaverse 相关产品、服务或商业模式，代表约 1 万亿美元价值；89% 的 3200 名高管认为 metaverse 对未来增长重要。
+- 使用建议：适合论证企业端和消费端都存在沉浸式内容/空间体验需求。
+
+## 好莱坞媒体生产技术路线
+
+### MovieLabs - The Evolution of Media Creation: 2030 Vision
+
+- 文件：`movielabs_2030_vision_evolution_media_creation.pdf`
+- 年份：2019，仍是 2030 Vision 系列基础白皮书
+- 类型：好莱坞技术白皮书
+- 适合引用点：提出未来媒体创作会走向云端、软件定义、资产复用、自动化、XR/沉浸式格式。报告提到资产复用可覆盖 feature、marketing、XR experiences、games、spin-offs 等。
+- 使用建议：虽然不是近两年，但它是 MovieLabs 2030 系列基础文件，后续 2026 论坛材料仍在引用。
+
+### MovieLabs - Software-Defined Workflows
+
+- 文件：`movielabs_software_defined_workflows.pdf`
+- 类型：好莱坞生产工作流技术白皮书
+- 适合引用点：说明媒体生产工作流如何软件定义、自动化、可组合。
+- 项目对应价值：本项目的 Depth/Warp/Inpaint 模块化接口，可以和“软件定义工作流、模块可替换、自动化处理链”呼应。
+
+### MovieLabs - Interoperability in Media Creation
+
+- 文件：`movielabs_interoperability_media_creation.pdf`
+- 类型：好莱坞互操作白皮书
+- 适合引用点：强调工具、数据、元数据、工作流互操作，降低切换成本。
+- 项目对应价值：可支撑你们文档中“统一输入输出、坐标、掩码、评价协议”的工程化必要性。
+
+## 其他支撑资料
+
+### Salesforce - Media and Entertainment Redefined
+
+- 文件：`salesforce_ai_media_entertainment_redefined.pdf`
+- 年份：约 2023/2024
+- 类型：行业趋势报告
+- 适合引用点：生成式 AI 正在影响内容创作、营销、广告、订阅增长和媒体电商；报告引用 Market.us，称媒体娱乐 GenAI 收入预计 2032 年达到 116 亿美元。
+
+### Deloitte - Future of Media and Entertainment
+
+- 文件：`deloitte_future_of_media_and_entertainment_2024.pdf`
+- 年份：2024
+- 类型：媒体娱乐行业未来趋势
+- 适合引用点：内容爆炸、设备与格式演进、生成式 AI 对媒体娱乐的影响。
+
+### Deloitte - State of Generative AI in the Enterprise
+
+- 文件：`deloitte_state_of_gen_ai_enterprise_2024.pdf`
+- 年份：2024
+- 类型：企业 GenAI 调研
+- 适合引用点：企业采用 GenAI 的收益包括效率、生产力、成本下降。
+
+### Accenture - Technology Vision 2025
+
+- 文件：`accenture_technology_vision_2025.pdf`
+- 年份：2025
+- 类型：大厂技术趋势报告
+- 适合引用点：AI 自主性、模型工作流、企业级 AI 应用趋势。
+
+### Accenture - Getting Your Workforce Ready for AI and Spatial Computing
+
+- 文件：`accenture_ai_spatial_computing_workforce_2024.pdf`
+- 年份：2024
+- 类型：AI + Spatial Computing 报告
+- 适合引用点：空间计算与 AI 可提升沉浸式协作、降低企业运营成本、扩展内容创建能力。
+
+### OECD - An Immersive Technologies Policy Primer
+
+- 文件：`oecd_immersive_technologies_policy_primer_2025.pdf`
+- 年份：2025
+- 类型：政策/产业技术报告
+- 适合引用点：沉浸式技术应用方向、政策与产业机会、生产率和成本影响。
+
+## 建议写进项目文档的表述方向
+
+1. 市场空间：全球娱乐传媒市场已是数万亿美元规模，PwC 预计 2030 年达到 4.2 万亿美元；沉浸式/metaverse 相关市场上限被 McKinsey、Citi、Accenture 估到万亿美元级。
+2. 供给瓶颈：XR、空间视频、裸眼 3D 和 3D 影院的消费体验需要大量空间内容，但原生双目/多目拍摄对设备、同步、标定、现场调度和后期校正要求高，难以覆盖海量存量 2D 视频。
+3. 降本逻辑：自动 2D 转 3D不替代所有原生拍摄，但可把大量存量影视、广告、短视频、纪录片、动画和 UGC 转成可消费的双目/空间视频，降低内容供给成本。
+4. 技术价值：本项目的 `Depth -> Warp -> Inpaint` 多阶段结构对应影视后期中的几何理解、视图生成和缺失区域补全；模块化设计符合 MovieLabs 软件定义工作流和互操作趋势。
+5. 竞品与落地：DNEG/Prime Focus 仍将 Stereo Conversion 列为端到端媒体内容服务能力；Netflix 已在 2026 年财报会中公开 AI 生产案例并给出半成本、两倍速度的效率口径。
+
+## 重要口径说明
+
+- PwC 的 4.2 万亿美元是整个娱乐传媒市场，McKinsey 的 5 万亿美元是潜在经济影响，Citi 的 8-13 万亿美元是广义元宇宙情景 TAM；三者定义不同，不能直接比较或相加。
+- 本项目的直接市场应通过目标客户数、可转换片库分钟数、单分钟报价和采用率测算；外部万亿美元数字只用于行业背景。
+- Netflix、Deloitte、Morgan Stanley、MovieLabs 的降本数字分别对应单一 AI 案例、虚拟制作、广义 GenAI 和云编辑平台，不能替代项目自身试点数据。
+
+## 年份说明
+
+- 核心近年资料：2024-2026。
+- 市场规模背景：2022-2023，但仍常被引用，主要用于“万亿级市场潜力”。
+- MovieLabs 2030 Vision 基础白皮书年份较早，但与 2026 论坛材料形成连续体系，建议作为好莱坞技术路线背景使用。

2d_to_3d_market_tech_reports/accenture_ai_spatial_computing_workforce_2024.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00975bbf15d886a4637918323fc4f0132ecb4cd6cfbda07886df75458243b3dc
+size 3121834

2d_to_3d_market_tech_reports/accenture_metaverse_evolution_before_revolution_2023.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62a4be05220b5b0196e84bed8d6604baabbc7c6a68a60938fcf510d8dabbdbae
+size 2117146

2d_to_3d_market_tech_reports/accenture_technology_vision_2025.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d03519073f8182aa0b90264a569f2bfd7fe834cedab2a22ef544f876b1e160aa
+size 5498156

2d_to_3d_market_tech_reports/autodesk_state_design_make_2025.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b78dcc282c49d8246c3c6a451fcf773e9558eb72fc70a03ca46c7e426f4a8c48
+size 17695087

2d_to_3d_market_tech_reports/build_market_chapter.py

@@ -0,0 +1,710 @@
+from __future__ import annotations
+
+import os
+import shutil
+import sys
+import zipfile
+from pathlib import Path
+
+import matplotlib.pyplot as plt
+from matplotlib import font_manager
+from matplotlib.patches import FancyBboxPatch
+from docx import Document
+from docx.enum.table import WD_CELL_VERTICAL_ALIGNMENT
+from docx.enum.text import WD_ALIGN_PARAGRAPH
+from docx.oxml import OxmlElement
+from docx.oxml.ns import qn
+from docx.shared import Cm, Pt, RGBColor
+
+
+ROOT_BYTES = b"/mnt/hdd/wjy/czy/diaoyan"
+REPORT_ROOT = Path("/mnt/hdd/wjy/czy/diaoyan/2d_to_3d_market_tech_reports")
+ASSET_DIR = REPORT_ROOT / "market_chapter_assets"
+BLUE = "1F4E79"
+MID_BLUE = "5B9BD5"
+LIGHT_BLUE = "DCE6F1"
+PALE_BLUE = "EDF3F8"
+ORANGE = "ED7D31"
+GRAY = "66727C"
+
+
+def mpl_color(value: str) -> str:
+    return value if value.startswith("#") else f"#{value}"
+
+
+def encoded_path(display_name: str) -> bytes:
+    return ROOT_BYTES + b"/" + display_name.encode("gbk")
+
+
+TARGET_BYTES = encoded_path("修改文档.docx")
+BACKUP_BYTES = encoded_path("修改文档_补充市场调研前备份.docx")
+TEMP_BYTES = ROOT_BYTES + b"/.market_chapter_edit.tmp.docx"
+
+
+def configure_chinese_fonts() -> None:
+    regular = "/usr/share/fonts/opentype/noto/NotoSansCJK-Regular.ttc"
+    font_manager.fontManager.addfont(regular)
+    family = font_manager.FontProperties(fname=regular).get_name()
+    plt.rcParams["font.family"] = family
+    plt.rcParams["axes.unicode_minus"] = False
+
+
+def create_growth_chart(path: Path) -> None:
+    years = list(range(2020, 2030))
+    revenue = [2.31, 2.53, 2.66, 2.78, 2.93, 3.06, 3.20, 3.30, 3.42, 3.51]
+    growth = [None, 9.4, 5.0, 4.7, 5.5, 4.5, 4.3, 3.4, 3.4, 2.7]
+    colors = [mpl_color(MID_BLUE) if y <= 2024 else "#9DC3E6" for y in years]
+
+    fig, ax = plt.subplots(figsize=(8.4, 4.35), dpi=220)
+    bars = ax.bar(years, revenue, color=colors, width=0.66, zorder=2)
+    ax.set_ylim(0, 4.05)
+    ax.set_ylabel("全球娱乐与媒体收入（万亿美元）", fontsize=9.5, color="#333333")
+    ax.set_xticks(years)
+    ax.set_xticklabels([str(y) for y in years], fontsize=8.5)
+    ax.grid(axis="y", color="#D9E1E8", linewidth=0.7, zorder=1)
+    ax.spines[["top", "right"]].set_visible(False)
+    ax.spines[["left", "bottom"]].set_color("#AAB4BE")
+    for bar, value in zip(bars, revenue):
+        ax.text(bar.get_x() + bar.get_width() / 2, value + 0.06, f"{value:.2f}",
+                ha="center", va="bottom", fontsize=7.8, color="#34495E")
+
+    ax2 = ax.twinx()
+    xs = years[1:]
+    ys = growth[1:]
+    ax2.plot(xs, ys, color=mpl_color(ORANGE), marker="o", linewidth=2.0, markersize=4.2, zorder=3)
+    ax2.set_ylim(0, 10.6)
+    ax2.set_ylabel("年度增速（%）", fontsize=9.5, color=mpl_color(ORANGE))
+    ax2.tick_params(axis="y", labelsize=8.2, colors=mpl_color(ORANGE))
+    ax2.spines["top"].set_visible(False)
+    ax2.spines["left"].set_visible(False)
+    ax2.spines["right"].set_color("#D7A17B")
+    for x, value in zip(xs, ys):
+        ax2.text(x, value + 0.35, f"{value:.1f}%", ha="center", va="bottom",
+                 fontsize=7.3, color="#B45F23")
+
+    ax.set_title("全球娱乐与媒体行业收入保持增长", loc="left", fontsize=15,
+                 fontweight="bold", color="#1F4E79", pad=12)
+    ax.text(2024.5, 3.88, "2025-2029为预测值", ha="center", va="center",
+            fontsize=8.2, color=mpl_color(GRAY), bbox=dict(boxstyle="round,pad=0.25", fc="#F2F5F7", ec="none"))
+    fig.text(0.01, 0.01, "数据来源：PwC《Global Entertainment & Media Outlook 2025-2029》[13]",
+             fontsize=7.4, color=mpl_color(GRAY))
+    fig.tight_layout(rect=(0.015, 0.045, 0.985, 0.965))
+    fig.savefig(path, bbox_inches="tight", facecolor="white")
+    plt.close(fig)
+
+
+def add_box(ax, x, y, w, h, title, body, fc, ec=BLUE, title_color=BLUE):
+    box = FancyBboxPatch((x, y), w, h, boxstyle="round,pad=0.018,rounding_size=0.018",
+                         linewidth=1.1, edgecolor=mpl_color(ec), facecolor=fc)
+    ax.add_patch(box)
+    ax.text(x + 0.03, y + h - 0.055, title, ha="left", va="top", fontsize=11,
+            fontweight="bold", color=mpl_color(title_color))
+    ax.text(x + 0.03, y + h - 0.125, body, ha="left", va="top", fontsize=8.4,
+            color="#334455", linespacing=1.5)
+
+
+def create_application_map(path: Path) -> None:
+    fig, ax = plt.subplots(figsize=(8.4, 4.55), dpi=220)
+    ax.set_xlim(0, 1)
+    ax.set_ylim(0, 1)
+    ax.axis("off")
+    ax.text(0.02, 0.96, "技术能力与重点应用市场的对应关系", fontsize=15,
+            fontweight="bold", color="#1F4E79", va="top")
+
+    add_box(ax, 0.02, 0.55, 0.22, 0.30, "电影与长视频",
+            "存量片库立体化\n影院3D/大画幅版本\n纪录片、动画、剧集", "#EDF3F8")
+    add_box(ax, 0.26, 0.55, 0.22, 0.30, "OTT与个人消费",
+            "流媒体空间版本\n头显与家庭终端\n短视频与UGC", "#F5F8FB")
+    add_box(ax, 0.50, 0.55, 0.22, 0.30, "XR与空间计算",
+            "双目/空间视频\n可调视差与零视差\n��终端格式适配", "#EDF3F8")
+    add_box(ax, 0.74, 0.55, 0.24, 0.30, "商业与沉浸体验",
+            "裸眼3D与数字标牌\n主题乐园、数字演出\n4D放映与沉浸式场馆", "#F5F8FB")
+
+    pipeline = FancyBboxPatch((0.08, 0.18), 0.84, 0.20,
+                              boxstyle="round,pad=0.02,rounding_size=0.025",
+                              linewidth=1.3, edgecolor=mpl_color(BLUE), facecolor="#DCE6F1")
+    ax.add_patch(pipeline)
+    ax.text(0.5, 0.335, "统一内容生产底座", ha="center", va="center",
+            fontsize=10.2, color=mpl_color(BLUE), fontweight="bold")
+    ax.text(0.5, 0.255, "单目视频  →  Depth  →  Warp  →  Inpaint/Fusion  →  左右眼 / SBS / 空间视频",
+            ha="center", va="center", fontsize=10, color="#203A52", fontweight="bold")
+    ax.text(0.5, 0.105, "共性价值：复用海量2D内容 · 减少重新拍摄与逐帧处理 · 支持批量化和参数化交付",
+            ha="center", va="center", fontsize=8.6, color=mpl_color(GRAY))
+    fig.text(0.02, 0.012, "资料来源：PwC[12-13]、Prime Focus/DNEG[19]、WEF[17]；项目组整理",
+             fontsize=7.4, color=mpl_color(GRAY))
+    fig.tight_layout(pad=0.2)
+    fig.savefig(path, bbox_inches="tight", facecolor="white")
+    plt.close(fig)
+
+
+def create_cost_evidence(path: Path) -> None:
+    fig, ax = plt.subplots(figsize=(8.4, 4.45), dpi=220)
+    ax.set_xlim(0, 1)
+    ax.set_ylim(0, 1)
+    ax.axis("off")
+    ax.text(0.02, 0.965, "影视内容生产与后期自动化的公开价值证据", fontsize=15,
+            fontweight="bold", color="#1F4E79", va="top")
+
+    cards = [
+        (0.02, "传统成本基线", "重拍占最终制作成本\n5%-20%（有时更高）\n高预算科幻/奇幻片VFX\n可达总预算20%", "Deloitte 虚拟制作报告[15]"),
+        (0.265, "真实制作案例", "17分钟AI增强画面\n制作速度达到2倍\n成本降至此前方案的\n约50%", "Netflix 2026财报会[14]"),
+        (0.51, "行业降本估计", "媒体行业成本约降10%\n电视和电影制作\n最高约降30%", "Morgan Stanley 研究[16]"),
+        (0.755, "云端后期案例", "成本改善30%以上\n所需IT人员减少85%\n零CAPEX与100%利用率", "MovieLabs/JioStar案例[20]"),
+    ]
+    for idx, (x, title, body, source) in enumerate(cards):
+        fc = "#EDF3F8" if idx % 2 == 0 else "#F7F9FB"
+        box = FancyBboxPatch((x, 0.20), 0.225, 0.62,
+                             boxstyle="round,pad=0.018,rounding_size=0.02",
+                             linewidth=1.0, edgecolor="#6E91AE", facecolor=fc)
+        ax.add_patch(box)
+        ax.text(x + 0.018, 0.765, title, ha="left", va="top", fontsize=10.4,
+                fontweight="bold", color=mpl_color(BLUE))
+        ax.text(x + 0.018, 0.665, body, ha="left", va="top", fontsize=9.0,
+                color="#263C4D", linespacing=1.55)
+        ax.text(x + 0.018, 0.245, source, ha="left", va="bottom", fontsize=7.2,
+                color=mpl_color(GRAY), wrap=True)
+
+    ax.text(0.5, 0.095, "注意：四组数字分别对应传统成本结构、单一项目、行业预测和云编辑平台，口径不同，不可相加或直接视为本项目实测值。",
+            ha="center", va="center", fontsize=8.1, color="#9C4A1A",
+            bbox=dict(boxstyle="round,pad=0.35", fc="#FFF3E8", ec="#F2C29B"))
+    fig.tight_layout(pad=0.15)
+    fig.savefig(path, bbox_inches="tight", facecolor="white")
+    plt.close(fig)
+
+
+def create_dneg_capability_chart(path: Path) -> None:
+    fig, ax = plt.subplots(figsize=(8.4, 4.725), dpi=220)
+    ax.set_xlim(0, 1)
+    ax.set_ylim(0, 1)
+    ax.axis("off")
+    ax.text(0.02, 0.965, "DNEG端到端媒体内容处理能力", fontsize=15,
+            fontweight="bold", color="#1F4E79", va="top")
+
+    center = FancyBboxPatch((0.30, 0.33), 0.25, 0.30,
+                            boxstyle="round,pad=0.025,rounding_size=0.035",
+                            linewidth=1.5, edgecolor="#1F4E79", facecolor="#F5A14B")
+    ax.add_patch(center)
+    ax.text(0.425, 0.515, "DNEG", ha="center", va="center", fontsize=18,
+            fontweight="bold", color="#FFFFFF")
+    ax.text(0.425, 0.42, "端到端媒体内容\n处理生态", ha="center", va="center",
+            fontsize=9.3, color="#253B50", fontweight="bold", linespacing=1.45)
+
+    items = [
+        (0.03, 0.66, "电影与剧集\n视觉特效"),
+        (0.18, 0.73, "长片与剧集\n动画"),
+        (0.03, 0.37, "立体转换"),
+        (0.18, 0.08, "前期创意\n服务"),
+        (0.43, 0.72, "制作服务、摄影棚\n与设备租赁"),
+        (0.56, 0.46, "沉浸式体验\n与动作捕捉"),
+        (0.43, 0.08, "全球内容处理\n与发行"),
+        (0.03, 0.12, "虚拟制作与\n实时服务"),
+    ]
+    for idx, (x, y, label) in enumerate(items):
+        width = 0.20 if idx in (4, 5, 6) else 0.15
+        box = FancyBboxPatch((x, y), width, 0.14,
+                             boxstyle="round,pad=0.012,rounding_size=0.018",
+                             linewidth=1.0, edgecolor="#6E91AE",
+                             facecolor="#EDF3F8" if idx % 2 == 0 else "#F7F9FB")
+        ax.add_patch(box)
+        ax.text(x + width / 2, y + 0.07, label, ha="center", va="center",
+                fontsize=8.4, color="#1F4E79", fontweight="bold", linespacing=1.35)
+
+    side = FancyBboxPatch((0.72, 0.16), 0.25, 0.66,
+                          boxstyle="round,pad=0.018,rounding_size=0.025",
+                          linewidth=1.1, edgecolor="#1F4E79", facecolor="#F4F7FA")
+    ax.add_patch(side)
+    ax.text(0.755, 0.755, "覆盖制作全流程", ha="left", va="top",
+            fontsize=11, color="#1F4E79", fontweight="bold")
+    processes = "前期制作\n拍摄制作\n后期制作\n视觉特效\n动画与游戏\n母版制作与发行"
+    ax.text(0.755, 0.655, processes, ha="left", va="top", fontsize=9.2,
+            color="#2D4355", linespacing=1.65)
+    ax.text(0.02, 0.025,
+            "来源：Prime Focus/DNEG《Investor Presentation: January 2026》第9页[19]；项目组据官方能力图重绘",
+            fontsize=7.2, color="#66727C")
+    fig.tight_layout(pad=0.15)
+    fig.savefig(path, bbox_inches="tight", facecolor="white")
+    plt.close(fig)
+
+
+def set_cell_shading(cell, fill: str) -> None:
+    tc_pr = cell._tc.get_or_add_tcPr()
+    shd = tc_pr.find(qn("w:shd"))
+    if shd is None:
+        shd = OxmlElement("w:shd")
+        tc_pr.append(shd)
+    shd.set(qn("w:fill"), fill)
+
+
+def set_cell_margins(cell, top=70, start=90, bottom=70, end=90) -> None:
+    tc = cell._tc
+    tc_pr = tc.get_or_add_tcPr()
+    tc_mar = tc_pr.first_child_found_in("w:tcMar")
+    if tc_mar is None:
+        tc_mar = OxmlElement("w:tcMar")
+        tc_pr.append(tc_mar)
+    for margin, value in (("top", top), ("start", start), ("bottom", bottom), ("end", end)):
+        node = tc_mar.find(qn(f"w:{margin}"))
+        if node is None:
+            node = OxmlElement(f"w:{margin}")
+            tc_mar.append(node)
+        node.set(qn("w:w"), str(value))
+        node.set(qn("w:type"), "dxa")
+
+
+def format_run(run, size=9.0, bold=False, color=None) -> None:
+    run.font.name = "Times New Roman"
+    run._element.get_or_add_rPr().get_or_add_rFonts().set(qn("w:eastAsia"), "Noto Sans SC")
+    run.font.size = Pt(size)
+    run.bold = bold
+    if color:
+        run.font.color.rgb = RGBColor.from_string(color)
+
+
+def format_table(table, widths_cm: list[float]) -> None:
+    table.style = "Table Grid"
+    table.autofit = False
+    tbl_pr = table._tbl.tblPr
+    layout = tbl_pr.find(qn("w:tblLayout"))
+    if layout is None:
+        layout = OxmlElement("w:tblLayout")
+        tbl_pr.append(layout)
+    layout.set(qn("w:type"), "fixed")
+
+    for row_index, row in enumerate(table.rows):
+        tr_pr = row._tr.get_or_add_trPr()
+        cant_split = OxmlElement("w:cantSplit")
+        tr_pr.append(cant_split)
+        if row_index == 0:
+            header = OxmlElement("w:tblHeader")
+            header.set(qn("w:val"), "true")
+            tr_pr.append(header)
+        for col_index, cell in enumerate(row.cells):
+            cell.width = Cm(widths_cm[col_index])
+            cell.vertical_alignment = WD_CELL_VERTICAL_ALIGNMENT.CENTER
+            set_cell_margins(cell)
+            set_cell_shading(cell, BLUE if row_index == 0 else ("F4F7FA" if row_index % 2 == 0 else "FFFFFF"))
+            for paragraph in cell.paragraphs:
+                paragraph.alignment = WD_ALIGN_PARAGRAPH.CENTER if row_index == 0 else WD_ALIGN_PARAGRAPH.LEFT
+                paragraph.paragraph_format.space_after = Pt(0)
+                paragraph.paragraph_format.line_spacing = 1.15
+                for run in paragraph.runs:
+                    format_run(run, size=8.5, bold=(row_index == 0), color="FFFFFF" if row_index == 0 else None)
+
+
+def add_table(doc, headers: list[str], rows: list[list[str]], widths: list[float]):
+    table = doc.add_table(rows=1, cols=len(headers))
+    for index, text in enumerate(headers):
+        table.rows[0].cells[index].text = text
+    for row_data in rows:
+        cells = table.add_row().cells
+        for index, text in enumerate(row_data):
+            cells[index].text = text
+    format_table(table, widths)
+    return table
+
+
+def add_paragraph(doc, text: str, style="Normal", size=None, bold_prefix: str | None = None):
+    paragraph = doc.add_paragraph(style=style)
+    if bold_prefix and text.startswith(bold_prefix):
+        first = paragraph.add_run(bold_prefix)
+        format_run(first, size=size or 10.5, bold=True, color=BLUE)
+        rest = paragraph.add_run(text[len(bold_prefix):])
+        format_run(rest, size=size or 10.5)
+    else:
+        run = paragraph.add_run(text)
+        if size:
+            format_run(run, size=size)
+    return paragraph
+
+
+def add_caption(doc, text: str):
+    paragraph = doc.add_paragraph(style="Caption")
+    paragraph.add_run(text)
+    paragraph.paragraph_format.keep_with_next = True
+    return paragraph
+
+
+def add_figure(doc, image_path: Path, width_cm: float, caption: str):
+    paragraph = doc.add_paragraph(style="Normal")
+    paragraph.alignment = WD_ALIGN_PARAGRAPH.CENTER
+    paragraph.paragraph_format.space_after = Pt(0)
+    paragraph.add_run().add_picture(str(image_path), width=Cm(width_cm))
+    cap = doc.add_paragraph(caption, style="Caption")
+    return paragraph, cap
+
+
+def move_before(anchor, block) -> None:
+    element = block._p if hasattr(block, "_p") else block._tbl
+    anchor._p.addprevious(element)
+
+
+def remove_paragraph(paragraph) -> None:
+    element = paragraph._element
+    element.getparent().remove(element)
+
+
+def build_chapter(doc: Document, growth: Path, applications: Path, cost: Path, dneg: Path) -> None:
+    if any(p.text.strip().startswith("7.1 调研口径") for p in doc.paragraphs):
+        raise RuntimeError("第7章已经包含本次市场调研内容，停止重复写入。")
+
+    heading_8 = next(p for p in doc.paragraphs if p.text.strip() == "8 参考资料")
+    heading_7_index = next(i for i, p in enumerate(doc.paragraphs) if p.text.strip() == "7 市场调研")
+    blank = doc.paragraphs[heading_7_index + 1]
+    if not blank.text.strip():
+        remove_paragraph(blank)
+
+    blocks = []
+
+    blocks.append(add_paragraph(doc, "7.1 调研口径与市场边界", style="Heading 2"))
+    blocks.append(add_paragraph(
+        doc,
+        "本节围绕本项目“单目视频输入—时序深度估计—视图变换—遮挡修补—双目/空间视频输出”的技术链路，选取2022—2026年由PwC、Netflix、Deloitte、Morgan Stanley、世界经济论坛、Autodesk、MovieLabs及Prime Focus/DNEG等机构公开的市场报告、投资者材料和制作案例。调研重点不是泛化罗列3D产业，而是回答三个问题：目标内容市场是否持续增长，立体化与空间化内容是否存在明确应用入口，自动化后期是否具有可验证的时间和成本价值。",
+    ))
+    blocks.append(add_paragraph(
+        doc,
+        "为避免夸大市场规模，本节将行业总盘、相邻可服务市场和项目近期目标市场分开表述。PwC的数万亿美元娱乐传媒收入、McKinsey和Citi的万亿美元级沉浸式情景只用于说明内容经济和空间体验的长期上限，不等同于2D转3D技术自身的可寻址市场。项目直接市场仍应依据可转换片库分钟数、单分钟报价、采用率和目标客户数量测算。",
+    ))
+    blocks.append(add_caption(doc, "表7-1  市场口径与本项目的对应关系"))
+    blocks.append(add_table(
+        doc,
+        ["市场层级", "口径说明", "本项目对应关系", "引用原则"],
+        [
+            ["行业总盘", "全球娱乐传媒、广告、连接及消费支出", "电影、OTT、广告和沉浸式内容的上游环境", "仅说明行业背景，不作为项目TAM"],
+            ["相邻市场", "OTT、影院、XR、沉浸式场馆和数字标牌", "需要双目、空间视频或高规格视听版本的分发场景", "说明需求入口，不等同于转换收入"],
+            ["直接服务市场", "Stereo Conversion、VFX、AI后期和内容版本制作", "Depth—Warp—Inpaint自动转换与专业校正", "用于竞品和商业能力验证"],
+            ["近期目标市场", "可触达客户及可处理内容形成的订单", "片库修复、广告短片、演示内容和专业后期工具", "通过项目报价和试点数据测算SAM/SOM"],
+        ],
+        [2.5, 4.3, 5.0, 4.0],
+    ))
+
+    blocks.append(add_paragraph(doc, "7.2 市场规模与发展趋势", style="Heading 2"))
+    blocks.append(add_paragraph(
+        doc,
+        "PwC《Global Entertainment & Media Outlook 2026—2030》显示，全球娱乐与媒体行业收入由2025年的3.5万亿美元预计增长至2030年的4.2万亿美元，预测期复合年增长率为3.4%，相较2025年形成约6000亿美元新增收入[12]。其2025版报告给出的连续数据也表明，全球娱乐传媒收入由2020年的2.31万亿美元增长至2024年的2.93万亿美元，并预计在2029年达到3.51万亿美元[13]。该趋势意味着内容生产、版本适配和跨终端分发仍处于扩张期，为立体化和空间化内容生产提供了稳定的上游需求。",
+    ))
+    fig, cap = add_figure(doc, growth, 15.4, "图3  全球娱乐与媒体收入规模及增速（数据来源：PwC[13]）")
+    blocks.extend([fig, cap])
+    blocks.append(add_paragraph(
+        doc,
+        "与视频内容直接相关的下游市场同样保持增长。PwC预计全球OTT收入将由2025年的2266亿美元增至2030年的3040亿美元，复合年增长率为6.1%；全球电影票房预计在2030年达到395亿美元，影院将继续投资大画幅、高规格视听和家庭环境难以复制的体验[12]。在沉浸式场馆方面，Las Vegas Sphere于2025年报告7.81亿美元收入，其节目已经包含4D电影放映，且运营方正在推进新的地区扩张[12]。这些数据不能直接转换为2D转3D收入，但说明影院、流媒体、头显和沉浸式大屏具有持续的高质量内容版本需求。",
+    ))
+    blocks.append(add_caption(doc, "表7-2  与本项��相关的市场指标"))
+    blocks.append(add_table(
+        doc,
+        ["指标", "基期/现状", "预测或公开结果", "与项目的关系"],
+        [
+            ["全球娱乐与媒体收入", "2025年3.5万亿美元", "2030年4.2万亿美元，CAGR 3.4%", "内容生产与分发的行业总盘"],
+            ["全球OTT收入", "2025年2266亿美元", "2030年3040亿美元，CAGR 6.1%", "片库立体化和空间视频版本的分发入口"],
+            ["全球电影票房", "持续恢复", "2030年395亿美元，CAGR 3.2%", "3D电影、大画幅和高规格影院体验"],
+            ["Sphere沉浸式场馆", "2025年收入7.81亿美元", "包含4D电影内容并规划扩张", "沉浸式大屏和专业空间内容应用"],
+            ["沉浸式长期情景", "预测口径差异较大", "McKinsey：2030年4万亿-5万亿美元潜在影响", "仅作为空间内容长期需求上限"],
+        ],
+        [3.2, 3.4, 5.0, 4.2],
+    ))
+    blocks.append(add_paragraph(
+        doc,
+        "对于更宽口径的空间计算与沉浸式市场，McKinsey估计元宇宙到2030年可能产生4万亿—5万亿美元经济影响[21]，Citi在设备无关的广义定义下给出8万亿—13万亿美元情景TAM[22]，Accenture调查则显示，具有相关战略的企业预计约4.2%的收入来自元宇宙相关产品、服务或商业模式，对应约1万亿美元价值[23]。上述预测包含电商、游戏、硬件和基础设施，技术与监管不确定性较高，因此本项目仅将其作为空间内容长期需求的辅助背景。",
+    ))
+
+    blocks.append(add_paragraph(doc, "7.3 应用方向与需求结构", style="Heading 2"))
+    blocks.append(add_paragraph(
+        doc,
+        "市场需求的核心矛盾并非缺少2D内容，而是空间终端和沉浸式场景增长后，可直接消费的双目、多视点内容供给不足。原生双目或多目拍摄需要额外机位、同步、标定、视差设计和后期校正，适合重点项目，却难以覆盖海量存量电影、剧集、纪录片、广告、动画和用户视频。本项目通过可解释的几何估计、确定性视图变换和受掩码约束的视频修补，将存量2D资产转换为可控的左右眼、SBS或空间视频版本，定位于“内容再利用与版本生产”环节。",
+    ))
+    fig, cap = add_figure(doc, applications, 15.4, "图4  技术能力与重点应用市场的对应关系（项目组根据公开报告整理[12-13,17,19]）")
+    blocks.extend([fig, cap])
+    blocks.append(add_caption(doc, "表7-3  重点应用场景、需求痛点与项目价值"))
+    blocks.append(add_table(
+        doc,
+        ["应用场景", "主要需求", "现有痛点", "本项目可提供的价值", "优先级"],
+        [
+            ["电影/剧集/纪录片", "存量片库和新增内容的立体版本", "原生立体拍摄与人工转换周期长、成本高", "离线高质量转换、镜头级视差控制、长视频稳定", "高"],
+            ["OTT与个人消费", "头显、家庭终端的空间视频内容", "内容规模不足，终端格式分散", "批量生成左右眼/SBS/空间视频并适配终端", "高"],
+            ["XR与空间计算", "舒适、稳定、可调节的双目内容", "视差抖动和左右眼冲突影响体验", "零视差平面、视差范围和设备参数可配置", "高"],
+            ["短视频/广告/UGC", "低成本、多版本、快速交付", "单条预算低，人工处理难以规模化", "自动化处理、局部复核、低边际成本", "中高"],
+            ["裸眼3D/商业展示", "多视点和大屏沉浸内容", "内容定制多、显示规格不统一", "以双目为基础扩展多视点和显示封装", "中"],
+            ["专业后期工具", "深度、视差、掩码和修补可编辑", "端到端黑盒难以满足导演和后期控制", "提供中间结果、参数控制和人工校正接口", "高"],
+        ],
+        [2.7, 3.6, 3.9, 4.7, 1.3],
+    ))
+
+    blocks.append(add_paragraph(doc, "7.4 产业主体与竞品验证", style="Heading 2"))
+    blocks.append(add_paragraph(
+        doc,
+        "Prime Focus/DNEG的2026年投资者材料将立体转换（Stereo Conversion）与电影/剧集视觉特效、动画、虚拟制作、沉浸式体验、后期处理和发行并列，纳入其端到端媒体内容处理生态[19]。这是一项直接的产业验证：立体转换并非只停留在研究论文中，而是国际头部视觉特效服务商持续提供的商业能力。材料同时显示，DNEG正由单一视觉特效服务扩展至Brahma AI内容生成与管理平台，反映行业竞争正在由“单个镜头制作”向“AI工具、资产管理、自动化流程和多渠道交付”延伸。",
+    ))
+    fig, cap = add_figure(doc, dneg, 15.4, "图5  DNEG端到端媒体内容处理能力（根据Prime Focus/DNEG官方材料重绘[19]）")
+    blocks.extend([fig, cap])
+    blocks.append(add_caption(doc, "表7-4  代表性产业主体及其公开动向"))
+    blocks.append(add_table(
+        doc,
+        ["���体", "公开能力或案例", "产业角色", "对本项目的启示"],
+        [
+            ["Prime Focus/DNEG", "立体转换、视觉特效、虚拟制作、沉浸式体验和Brahma AI", "直接竞品与产业服务商", "产品需同时具备高质量转换、可控后期和端到端交付能力"],
+            ["Netflix", "生成式AI用于预演、VFX、镜头规划和后期，已覆盖约300部作品", "内容平台与需求方", "自动化工具只有进入真实制作流程并提升单位投入产出才具有商业价值"],
+            ["MovieLabs/JioStar", "云端虚拟编辑、集成工作流及AI本地化", "好莱坞技术组织与规模化案例", "长视频处理应支持云端、批量任务、权限和资产复用"],
+            ["WEF/Autodesk", "AI内容生产价值链和M&E行业数字化调查", "行业趋势与采用度证据", "成本控制、生产力和可治理的AI工作流是采购决策重点"],
+        ],
+        [3.2, 5.0, 3.3, 4.6],
+    ))
+    blocks.append(add_paragraph(
+        doc,
+        "从竞争关系看，本项目不宜仅以“能生成右眼画面”作为差异点。专业市场更关注几何和时间稳定性、可靠区域保真、视差可控、长视频效率、失败帧定位、人工干预以及交付格式。Depth—Warp—Inpaint分阶段架构能够显式暴露深度、视差、遮挡掩码和修补结果，更容易形成专业后期工具、质量审计与模块替换能力，这是相对于纯端到端黑盒生成路线的重要产品化优势。",
+    ))
+
+    blocks.append(add_paragraph(doc, "7.5 降本提效与商业价值", style="Heading 2"))
+    blocks.append(add_paragraph(
+        doc,
+        "Deloitte对虚拟制作的调研指出，高预算电影的重拍较为常见，可占最终制作成本的5%—20%甚至更多；高预算科幻/奇幻电影的VFX成本可达到总预算的20%。更早介入的数字资产、可视化和自动化流程能够减少重拍、合成、抠像、差旅和外景成本[15]。虽然该报告讨论的是虚拟制作而非2D转3D，但它揭示了本项目所处的后期与版本生产环节具有明确成本压力。",
+    ))
+    fig, cap = add_figure(doc, cost, 15.4, "图6  影视内容生产与后期自动化的公开价值证据（来源：[14-16,20]）")
+    blocks.extend([fig, cap])
+    blocks.append(add_paragraph(
+        doc,
+        "Netflix在2026年第二季度财报会披露，纪录片系列American Experiment包含17分钟AI增强画面，该部分相较此前方案以两倍速度、约一半成本完成；Netflix同时表示生成式AI工作流已在约300部作品中使用，当前最集中于后期[14]。这一案例表明，AI画面处理不仅能够降低既有方案成本，还可以使原本受预算和周期限制的镜头进入成片。由于案例范围仅为17分钟AI增强画面，不能外推为所有影片均可节省50%。",
+    ))
+    blocks.append(add_paragraph(
+        doc,
+        "在行业层面，Morgan Stanley估计生成式AI可使媒体行业成本降低约10%，电视和电影制作成本最多降低约30%，主要机会包括动画、脚本、编辑、声音、VFX以及以数字或虚拟组件减少重拍[16]。MovieLabs公布的JioStar/Postudio云端后期案例则报告，相较本地部署成本改善30%以上、所需IT人员减少85%，并实现零资本性投入和资源充分利用[20]。Autodesk 2025年对5594名行业领导者的调查也将成本控制列为首要业务挑战，并认为媒体娱乐行业同时面临显著AI扰动和收益[18]。这些材料共同说明，客户采购新工具的核心依据将是单位内容成本、交付周期、人员投入和可复用资产，而不只是单帧视觉效果。",
+    ))
+    blocks.append(add_caption(doc, "表7-5  外部价值证据及其在本项目中的使用边界"))
+    blocks.append(add_table(
+        doc,
+        ["证据来源", "公开结果", "可支持的项目论点", "使用限制"],
+        [
+            ["Deloitte[15]", "重拍占5%-20%；高预算类型片VFX可达预算20%", "传统制作与后期存在较高成本基线", "虚拟制作口径，不是本项目节省比例"],
+            ["Netflix[14]", "17分钟AI增强画面：2倍速度、约50%成本", "AI后期已产生真实制作价值", "单一项目局部案例，不可普遍外推"],
+            ["Morgan Stanley[16]", "媒体约降10%，电视/电影最高约降30%", "影视GenAI具有行业级降本空间", "研究估计，覆盖广义GenAI"],
+            ["MovieLabs[20]", "成本改善30%+，IT人员减少85%", "云化和批量化工作流可降低工程成本", "云编辑平台案例，不是算法指标"],
+            ["WEF[17]", "创意生产力约提升25%，编辑工作时间潜在减少72%", "自动化可释放重复编辑工作量", "不同研究任务，数字不可相加"],
+        ],
+        [3.0, 4.7, 4.6, 3.8],
+    ))
+
+    blocks.append(add_paragraph(doc, "7.6 项目市场定位与实施建议", style="Heading 2"))
+    blocks.append(add_paragraph(
+        doc,
+        "综合市场规模、��用需求和产业案例，本项目适合定位为“面向存量和新增2D视频的可控双目/空间视频生成底座”，而不是宣称替代所有原生立体拍摄。近期应优先进入对离线质量、可控参数和版本交付要求明确的专业后期、片库样片、广告展示和沉浸式演示场景，通过小规模试点积累单分钟处理成本、人工复核时间、返修率和观看舒适度数据；中期再面向OTT、XR终端和批量内容平台形成标准化服务接口；远期在双目系统稳定后扩展多视点、裸眼3D和实时处理。",
+    ))
+    blocks.append(add_caption(doc, "表7-6  项目能力、客户价值与验证指标"))
+    blocks.append(add_table(
+        doc,
+        ["项目能力", "解决的客户问题", "商业价值", "建议验证指标"],
+        [
+            ["时序深度与视差控制", "镜头间层次漂移、观看不舒适", "减少人工逐镜头调节", "视差稳定性、舒适区超限率、人工调参时长"],
+            ["可解释Warp与遮挡分析", "边界重影、空洞和错误区域难定位", "缩短质量检查和返修定位时间", "空洞率、边界错误率、问题帧定位时间"],
+            ["受掩码约束的视频修补", "逐帧补洞成本高、纹理闪烁", "降低重复修补工时并保护原内容", "修补区域工时、时序闪烁、可靠区域变化率"],
+            ["长视频批处理与统一输出", "格式多、编解码和任务管理复杂", "提高吞吐率并降低单位分钟成本", "分钟/GPU小时、峰值显存、任务成功率、交付周期"],
+            ["中间结果和人工校正接口", "黑盒结果难审计、专业客户不可控", "提升可用率、客户信任和复购可能", "人工接管率、返修率、客户验收通过率"],
+        ],
+        [3.6, 4.6, 4.3, 3.8],
+    ))
+
+    spacer = doc.add_paragraph(style="Normal")
+    spacer.paragraph_format.space_after = Pt(4)
+    blocks.append(spacer)
+    note = add_table(
+        doc,
+        ["内部商业测算建议"],
+        [["单分钟净节省 = 传统人工立体转换单价 - 系统推理成本 - 人工复核成本；年度净价值 = 年处理分钟数 × 单分钟净节省 - 固定研发与运维成本。外部报告用于证明行业需求和降本方向，项目试点数据用于证明本系统实际ROI。"]],
+        [16.3],
+    )
+    set_cell_shading(note.rows[1].cells[0], "EDF3F8")
+    blocks.append(note)
+    blocks.append(add_paragraph(
+        doc,
+        "因此，市场化评价应与前文实验体系同步建设：在PSNR、SSIM、LPIPS、视差误差和时序稳定性之外，增加每分钟GPU成本、每分钟人工复核工时、一次验收通过率、返修率和交付周期等指标。只有将技术质量与单位经济性同时量化，才能把多阶段框架的工程优势转化为可验证的市场竞争力。",
+    ))
+
+    for block in blocks:
+        move_before(heading_8, block)
+
+
+def update_references(doc: Document) -> None:
+    heading_index = next(i for i, p in enumerate(doc.paragraphs) if p.text.strip() == "8 参考资料")
+    existing = [p for p in doc.paragraphs[heading_index + 1:] if p.text.strip()]
+    for number, paragraph in enumerate(existing, start=1):
+        if not paragraph.text.lstrip().startswith("["):
+            paragraph.text = f"[{number}] {paragraph.text}"
+        p_pr = paragraph._p.get_or_add_pPr()
+        num_pr = p_pr.find(qn("w:numPr"))
+        if num_pr is not None:
+            p_pr.remove(num_pr)
+        paragraph.paragraph_format.left_indent = Cm(0.72)
+        paragraph.paragraph_format.first_line_indent = Cm(-0.72)
+        paragraph.paragraph_format.space_after = Pt(4)
+        paragraph.paragraph_format.line_spacing = 1.2
+        for run in paragraph.runs:
+            format_run(run, size=9.5)
+    new_references = [
+        "[12] PwC. Global Entertainment & Media Outlook 2026-30: Digital Innovation Drives Demand for Human Experiences [EB/OL]. 2026-06-22 [2026-07-22]. https://www.pwc.com/gx/en/issues/business-model-reinvention/outlook/insights-and-perspectives.html.",
+        "[13] PwC. Global Entertainment & Media Outlook 2025-2029: Hong Kong Summary [R]. Hong Kong: PwC, 2025.",
+        "[14] Netflix, Inc. FQ2 2026 Earnings Call Transcript [EB/OL]. 2026-07-16 [2026-07-22]. https://s22.q4cdn.com/959853165/files/doc_financials/2026/q2/Netflix-Inc-_Earnings-Call_2026-07-16T00_00_00_English-1.pdf.",
+        "[15] Deloitte. The Future of Content Creation: Virtual Production [R]. Deloitte Insights, 2024.",
+        "[16] Morgan Stanley. How AI Benefits—and Threatens—the Entertainment Industry [EB/OL]. 2025-07-17 [2026-07-22]. https://www.morganstanley.com/insights/articles/ai-in-media-entertainment-benefits-and-risks.",
+        "[17] World Economic Forum. Artificial Intelligence in Media, Entertainment and Sport [R]. Geneva: World Economic Forum, 2025.",
+        "[18] Autodesk. 2025 State of Design & Make [R]. San Francisco: Autodesk, 2025.",
+        "[19] Prime Focus Limited. Investor Presentation: January 2026 [R]. Mumbai: Prime Focus Limited, 2026.",
+        "[20] MovieLabs. Industry Forum Spring Summit 2026 Presentation Deck [R]. Los Angeles: MovieLabs, 2026.",
+        "[21] McKinsey & Company. Value Creation in the Metaverse: The Real Business of the Virtual World [R]. McKinsey & Company, 2022.",
+        "[22] Citi GPS. Metaverse and Money: Decrypting the Future [R]. New York: Citigroup, 2022.",
+        "[23] Accenture. Metaverse: Evolution, Then Revolution [R]. Accenture, 2023.",
+    ]
+    for text in new_references:
+        paragraph = doc.add_paragraph(text, style="Normal")
+        paragraph.paragraph_format.left_indent = Cm(0.72)
+        paragraph.paragraph_format.first_line_indent = Cm(-0.72)
+        paragraph.paragraph_format.space_after = Pt(4)
+        paragraph.paragraph_format.line_spacing = 1.2
+        for run in paragraph.runs:
+            format_run(run, size=9.5)
+
+
+def update_toc_reference_page(doc: Document) -> None:
+    for paragraph in doc.paragraphs:
+        if paragraph.text.startswith("8 参考资料\t"):
+            paragraph.text = "8 参考资料\t19"
+            return
+
+
+def update_revision_history(doc: Document) -> None:
+    table = doc.tables[0]
+    if any("补充第7章市场调研" in cell.text for row in table.rows for cell in row.cells):
+        return
+    cells = table.add_row().cells
+    values = ["3", "补充第7章市场调研", "V1.3", "项目组", "2026-07-22", "增加市场规模、应用、竞品和降本证据"]
+    for cell, value in zip(cells, values):
+        cell.text = value
+    for cell in cells:
+        cell.vertical_alignment = WD_CELL_VERTICAL_ALIGNMENT.CENTER
+        for paragraph in cell.paragraphs:
+            paragraph.alignment = WD_ALIGN_PARAGRAPH.CENTER
+            for run in paragraph.runs:
+                format_run(run, size=8.5)
+
+
+def validate_docx(path_bytes: bytes) -> None:
+    path = os.fsdecode(path_bytes)
+    with zipfile.ZipFile(path) as archive:
+        bad = archive.testzip()
+        if bad:
+            raise RuntimeError(f"DOCX ZIP校验失败：{bad}")
+        document_xml = archive.read("word/document.xml")
+        for marker in ("7.1 调研口径与市场边界", "图6", "[23] Accenture"):
+            if marker.encode("utf-8") not in document_xml:
+                raise RuntimeError(f"DOCX内容校验失败，缺少：{marker}")
+        if document_xml.count(b"<w:drawing") < 6:
+            raise RuntimeError("DOCX内容校验失败，嵌入图片数量不足6张。")
+
+
+def localize_existing_figure5() -> None:
+    ASSET_DIR.mkdir(parents=True, exist_ok=True)
+    configure_chinese_fonts()
+    dneg = ASSET_DIR / "figure5_dneg_capabilities_cn.png"
+    create_dneg_capability_chart(dneg)
+
+    doc = Document(os.fsdecode(TARGET_BYTES))
+    if len(doc.inline_shapes) < 5:
+        raise RuntimeError("未找到图5对应的嵌入图片。")
+    shape = doc.inline_shapes[4]
+    blip = shape._inline.graphic.graphicData.pic.blipFill.blip
+    image_part = doc.part.related_parts[blip.embed]
+    image_part._blob = dneg.read_bytes()
+
+    for paragraph in doc.paragraphs:
+        if paragraph.text.startswith("图5  DNEG"):
+            paragraph.text = "图5  DNEG端到端媒体内容处理能力（根据Prime Focus/DNEG官方材料重绘[19]）"
+            paragraph.style = doc.styles["Caption"]
+        for run in paragraph.runs:
+            replaced = run.text.replace(
+                "将Stereo Conversion与电影/剧集VFX",
+                "将立体转换（Stereo Conversion）与电影/剧集视觉特效",
+            ).replace("国际头部VFX服务商", "国际头部视觉特效服务商").replace(
+                "由单一VFX服务扩展", "由单一视觉特效服务扩展"
+            )
+            if replaced != run.text:
+                run.text = replaced
+    for table in doc.tables:
+        for row in table.rows:
+            for cell in row.cells:
+                for paragraph in cell.paragraphs:
+                    for run in paragraph.runs:
+                        replaced = run.text.replace(
+                            "Stereo Conversion、VFX、虚拟制作",
+                            "立体转换、视觉特效、虚拟制作",
+                        )
+                        if replaced != run.text:
+                            run.text = replaced
+
+    doc.save(os.fsdecode(TEMP_BYTES))
+    validate_docx(TEMP_BYTES)
+    os.replace(TEMP_BYTES, TARGET_BYTES)
+    print("localized_figure5=", TARGET_BYTES.decode("gbk"))
+
+
+def polish_existing_document() -> None:
+    doc = Document(os.fsdecode(TARGET_BYTES))
+    heading_index = next(i for i, p in enumerate(doc.paragraphs) if p.text.strip() == "8 参考资料")
+    for paragraph in doc.paragraphs[heading_index + 1:]:
+        p_pr = paragraph._p.get_or_add_pPr()
+        num_pr = p_pr.find(qn("w:numPr"))
+        if num_pr is not None:
+            p_pr.remove(num_pr)
+    update_toc_reference_page(doc)
+
+    note = next(t for t in doc.tables if t.cell(0, 0).text == "内部商业测算建议")
+    previous = note._tbl.getprevious()
+    if previous is None or previous.tag != qn("w:p"):
+        spacer = doc.add_paragraph(style="Normal")
+        spacer.paragraph_format.space_after = Pt(4)
+        note._tbl.addprevious(spacer._p)
+
+    doc.save(os.fsdecode(TEMP_BYTES))
+    validate_docx(TEMP_BYTES)
+    os.replace(TEMP_BYTES, TARGET_BYTES)
+    print("polished=", TARGET_BYTES.decode("gbk"))
+
+
+def main() -> None:
+    if not os.path.exists(TARGET_BYTES):
+        raise FileNotFoundError(TARGET_BYTES.decode("gbk"))
+
+    ASSET_DIR.mkdir(parents=True, exist_ok=True)
+    configure_chinese_fonts()
+    growth = ASSET_DIR / "figure3_global_em_growth.png"
+    applications = ASSET_DIR / "figure4_application_map.png"
+    cost = ASSET_DIR / "figure6_cost_evidence.png"
+    create_growth_chart(growth)
+    create_application_map(applications)
+    create_cost_evidence(cost)
+
+    dneg = ASSET_DIR / "figure5_dneg_capabilities_cn.png"
+    create_dneg_capability_chart(dneg)
+
+    if not os.path.exists(BACKUP_BYTES):
+        shutil.copy2(TARGET_BYTES, BACKUP_BYTES)
+
+    doc = Document(os.fsdecode(TARGET_BYTES))
+    build_chapter(doc, growth, applications, cost, dneg)
+    update_references(doc)
+    update_toc_reference_page(doc)
+    update_revision_history(doc)
+
+    doc.core_properties.subject = "基于多阶段深度估计、视图变换与视频修补的双目视频生成系统"
+    doc.core_properties.comments = "第7章市场调研于2026-07-22补充。"
+    doc.save(os.fsdecode(TEMP_BYTES))
+    validate_docx(TEMP_BYTES)
+    os.replace(TEMP_BYTES, TARGET_BYTES)
+
+    print("updated=", TARGET_BYTES.decode("gbk"))
+    print("backup=", BACKUP_BYTES.decode("gbk"))
+    print("assets=", ASSET_DIR)
+
+
+if __name__ == "__main__":
+    if "--polish-existing" in sys.argv:
+        polish_existing_document()
+    elif "--localize-existing" in sys.argv:
+        localize_existing_figure5()
+    else:
+        main()

2d_to_3d_market_tech_reports/citi_metaverse_and_money_2022.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08f86b65df3f78520c4354b7b689e05e1a1bfe9329977584398d1cd0b52bf7de
+size 8769353

2d_to_3d_market_tech_reports/deloitte_future_of_media_and_entertainment_2024.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9fa4b12841cbeb60c77e0906a93506315dc6b74c495fc3e4b852dfffe02f69b5
+size 2384582

2d_to_3d_market_tech_reports/deloitte_state_of_gen_ai_enterprise_2024.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14e93daf0e9672010758fc7d4c549b37781e4581d01caa1e49e92d6fc47b027d
+size 16791883

2d_to_3d_market_tech_reports/deloitte_virtual_production_future_content_creation_2024.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6e55185ce1afe51b2756f3cc3637317a5fb46cea2fe2fce6c3999129d674e8e
+size 617111

2d_to_3d_market_tech_reports/key_charts/README.md

@@ -0,0 +1,50 @@
+# 关键图表页索引
+
+这些 PNG 均由本目录中的原始 PDF 按页导出，未改写图中数据。使用时应在图注中标注机构、报告名、年份和 PDF 页码。
+
+## 01_market_size
+
+| 文件 | 原报告页 | 内容 |
+|---|---:|---|
+| `pwc_2025_p05_global_em_revenue.png` | 5 | 全球 E&M 收入 2020-2029，2029 年 3.51 万亿美元，CAGR 3.7% |
+| `pwc_2025_p07_revenue_structure.png` | 7 | 全球 E&M 广告、连接、消费者收入结构 |
+| `pwc_2025_p16_ott_market.png` | 16 | 香港 OTT 收入 2024-2029；仅作区域示例 |
+| `pwc_2026_p01_global_em_2030.png` | 网页打印 1 | 全球 E&M 2030 年 4.2 万亿美元标题和摘要 |
+| `pwc_2026_p05_sphere_immersive.png` | 网页打印 5 | Sphere 2025 年 7.81 亿美元收入及 4D 放映应用 |
+| `pwc_2026_p06_cinema_live_experiences.png` | 网页打印 6 | 影院和现场体验收入趋势图 |
+| `pwc_2026_p08_global_ott.png` | 网页打印 8 | 全球 OTT 2025 年 2266 亿美元、2030 年 3040 亿美元 |
+| `mckinsey_2022_p37_metaverse_impact.png` | 37 | 元宇宙 2030 年 4 万亿-5 万亿美元潜在经济影响及用例 |
+| `citi_2022_p12_metaverse_tam.png` | 12 | 不同渗透率假设下的元宇宙 TAM 情景表 |
+| `accenture_2023_p10_metaverse_value.png` | 10 | 企业预期的 4.2% 收入、约 1 万亿美元价值和高管调查 |
+
+## 02_cost_value
+
+| 文件 | 原报告页 | 内容 |
+|---|---:|---|
+| `autodesk_2025_p07_digital_transformation_roi.png` | 7 | 数字化转型改善幅度调查图 |
+| `autodesk_2025_p38_media_ai_impact.png` | 38 | M&E 的 AI 扰动与收益对比图 |
+| `autodesk_2025_p46_cost_control.png` | 46 | 成本控制是受访企业首要挑战，33% |
+| `deloitte_2024_p07_virtual_production_cost.png` | 7 | 重拍、VFX、差旅、外景、合成和抠像成本说明 |
+| `netflix_2026_p12_ai_cost_case.png` | 12 | 17 分钟 AI 增强画面，2 倍速度、半成本案例 |
+| `morgan_stanley_2025_p01_cost_estimate.png` | 网页打印 1 | 媒体约 10%、电视/电影最高 30% 的 GenAI 降本估计 |
+| `morgan_stanley_2025_p03_tv_film_savings.png` | 网页打印 3 | 前后期、VFX、虚拟组件和减少重拍的解释 |
+| `movielabs_2026_p146_jiostar_case.png` | 146 | JioStar 云编辑平台前后状态和 30%+ 成本改善 |
+| `movielabs_2026_p148_jiostar_outcomes.png` | 148 | 30%+ 成本改善、85% 更少 IT 人员、零 CAPEX |
+| `wef_2025_p10_ai_creative_value.png` | 10 | AI 内容创作价值链和创意生产力指标 |
+| `wef_2025_p14_ai_productivity_cost.png` | 14 | AI 提效降本流程图和编辑工作时间指标 |
+| `wef_2025_p15_ai_content_creation.png` | 15 | 视频对象处理和资产管理案例 |
+
+## 03_competitor_product
+
+| 文件 | 原报告页 | 内容 |
+|---|---:|---|
+| `primefocus_2026_p04_business_and_products.png` | 4 | Prime Focus/DNEG 的创意、AI/技术和制作三条业务线 |
+| `primefocus_2026_p09_stereo_conversion_capability.png` | 9 | DNEG 端到端内容生态中的 Stereo Conversion |
+| `primefocus_2026_p14_immersive_business.png` | 14 | 主题乐园与数字演出业务和应用案例 |
+| `primefocus_2026_p17_content_tam.png` | 17 | 企业内容转型超过 1300 亿美元的公司市场口径 |
+
+## 使用说明
+
+- `official_web_print` 文件是官方网页的本地打印件，便于归档，不应描述为厂商发布的原生 PDF 报告。
+- 图表可用于内部项目文档；如文档对外发布，应确认原报告的转载和图表使用条款，优先采用“据报告数据重绘并注明来源”的方式。
+- 具体数字、页码和限制条件见上一级目录的 `market_evidence_matrix.csv`。

2d_to_3d_market_tech_reports/market_evidence_matrix.csv

@@ -0,0 +1,19 @@
+"evidence_id","evidence_level","theme","organization","report","year","exact_claim","page","chart_file","project_use","limitation","local_report","official_url"
+"M01","C-宏观市场上限","全球娱乐传媒市场","PwC","Global Entertainment & Media Outlook 2025-2029: Hong Kong Summary","2025","全球娱乐与媒体收入预计由2024年的2.93万亿美元增长至2029年的3.51万亿美元，预测期CAGR为3.7%。","PDF第5页","key_charts/01_market_size/pwc_2025_p05_global_em_revenue.png","说明项目面向的电影、OTT、广告和沉浸式内容处于数万亿美元内容经济中。","这是整个E&M市场，不是2D转3D技术自身市场；不能直接写成本项目TAM。","pwc_global_entertainment_media_outlook_2025_2029.pdf","https://www.pwchk.com/en/tmt/entertainment-and-media-outlook-2025-2029.pdf"
+"M02","C-宏观市场上限","全球娱乐传媒市场","PwC","Global Entertainment & Media Outlook 2026-2030","2026","全球E&M收入由2025年的3.5万亿美元增至2030年的4.2万亿美元，CAGR为3.4%，2030年相较2025年新增约6000亿美元收入。","网页打印PDF第1-2页","key_charts/01_market_size/pwc_2026_p01_global_em_2030.png","用于市场总盘和数字内容需求增长的最新背景。","网页打印件仅为官方网页的本地留档；4.2万亿美元包含广告、连接和消费支出等广泛板块。","pwc_global_entertainment_media_outlook_2026_2030_official_web_print.pdf","https://www.pwc.com/gx/en/issues/business-model-reinvention/outlook/insights-and-perspectives.html"
+"M03","B-相邻可服务市场","OTT视频市场","PwC","Global Entertainment & Media Outlook 2026-2030","2026","全球OTT收入由2025年的2266亿美元预计增至2030年的3040亿美元，CAGR为6.1%，新增774亿美元。","网页打印PDF第8页","key_charts/01_market_size/pwc_2026_p08_global_ott.png","支撑长视频、流媒体片库立体化和空间视频版本制作的下游分发需求。","OTT收入并不等同于3D/空间视频收入，只能作为下游内容市场。","pwc_global_entertainment_media_outlook_2026_2030_official_web_print.pdf","https://www.pwc.com/gx/en/issues/business-model-reinvention/outlook/insights-and-perspectives.html"
+"M04","C-宏观市场上限","沉浸式与空间内容","McKinsey & Company","Value Creation in the Metaverse","2022","2030年元宇宙在消费和企业用例中的潜在经济影响约为4万亿至5万亿美元。","PDF第37页，Exhibit 11","key_charts/01_market_size/mckinsey_2022_p37_metaverse_impact.png","说明空间内容、VR/AR硬件、数字媒体和现场娱乐的长期需求上限。","这是情景化经济影响估计，技术、监管和社会不确定性很高；不是市场收入，也不是2D转3D TAM。","mckinsey_value_creation_in_the_metaverse_2022.pdf","https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/value-creation-in-the-metaverse"
+"M05","C-宏观市场上限","沉浸式与空间内容","Citi GPS","Metaverse and Money: Decrypting the Future","2022","设备无关的广义元宇宙到2030年TAM情景为8万亿至13万亿美元，潜在用户约50亿；报告图表给出不同数字经济渗透率假设。","PDF第3、8、12页，Figure 3","key_charts/01_market_size/citi_2022_p12_metaverse_tam.png","用于呈现个人消费设备、AR/VR和沉浸式互联网的远期情景。","范围包含电商、社交、游戏、基础设施等，口径极宽且为情景推演；不得作为本项目直接市场规模。","citi_metaverse_and_money_2022.pdf","https://www.citigroup.com/global/insights/metaverse-and-money_20220330"
+"M06","C-宏观市场上限","企业与消费沉浸式市场","Accenture","Metaverse: Evolution, Then Revolution","2023","具有元宇宙战略的企业预计未来三年4.2%的收入来自相关新产品、服务或商业模式，对应约1万亿美元；3200名受访高管中89%认为元宇宙对未来增长重要。","PDF第10页","key_charts/01_market_size/accenture_2023_p10_metaverse_value.png","支撑企业端和消费端对空间内容及沉浸式体验的需求预期。","数字来自高管预期调查，不是已实现收入，也不是2D转3D专项市场。","accenture_metaverse_evolution_before_revolution_2023.pdf","https://www.accenture.com/content/dam/accenture/final/accenture-com/document/Accenture-Metaverse-Evolution-Before-Revolution.pdf"
+"M07","B-直接下游市场","全球影院市场","PwC","Global Entertainment & Media Outlook 2026-2030","2026","全球电影票房预计以3.2%的CAGR恢复增长，到2030年达到395亿美元；影院将继续投资家庭环境难以复制的大画幅、高规格视听体验。","官方网页正文；网页打印PDF第6页含影院趋势图","key_charts/01_market_size/pwc_2026_p06_cinema_live_experiences.png","为3D电影版本、影院大画幅和高规格视听体验提供直接下游市场依据。","395亿美元是全部电影票房，不是3D电影票房，也不是立体转换服务收入。","pwc_global_entertainment_media_outlook_2026_2030_official_web_print.pdf","https://www.pwc.com/gx/en/issues/business-model-reinvention/outlook/insights-and-perspectives.html"
+"M08","B-直接应用案例","沉浸式场馆","PwC","Global Entertainment & Media Outlook 2026-2030","2026","Las Vegas Sphere在2025年报告收入7.81亿美元，节目包括《绿野仙踪》4D放映和演唱会，其母公司正在美国其他地区及迪拜规划扩张。","官方网页正文；网页打印PDF第5页","key_charts/01_market_size/pwc_2026_p05_sphere_immersive.png","说明沉浸式大屏、4D放映和现场体验已经形成实际收入与扩张案例。","这是单一场馆的整体收入，不是2D转3D项目收入；应用相关性高但不可直接外推。","pwc_global_entertainment_media_outlook_2026_2030_official_web_print.pdf","https://www.pwc.com/gx/en/issues/business-model-reinvention/outlook/insights-and-perspectives.html"
+"V01","A-直接影视案例","制作时间与成本","Netflix","FQ2 2026 Earnings Call Transcript","2026","纪录片系列American Experiment使用17分钟AI增强画面；Netflix称这部分相较此前方案制作速度为2倍、成本为一半，并称生成式AI已用于约300部作品，主要集中在后期。","PDF第12页","key_charts/02_cost_value/netflix_2026_p12_ai_cost_case.png","为自动化后期可缩短周期、降低单位内容成本提供大厂真实项目案例。","这是17分钟AI增强画面的单一案例，并非2D转3D或全片成本基准；不能外推为所有项目均节省50%。","netflix_q2_2026_earnings_call_transcript_ai_cost.pdf","https://s22.q4cdn.com/959853165/files/doc_financials/2026/q2/Netflix-Inc-_Earnings-Call_2026-07-16T00_00_00_English-1.pdf"
+"V02","A-直接影视成本基线","重拍与VFX成本","Deloitte","The Future of Content Creation: Virtual Production","2024","高预算电影重拍通常占最终制作成本5%-20%甚至更多；高预算科幻/奇幻电影VFX可占总预算20%；虚拟制作可减少重拍、合成、抠像、差旅和外景成本。","PDF第7页","key_charts/02_cost_value/deloitte_2024_p07_virtual_production_cost.png","用于建立传统制作和后期的成本痛点，并解释自动深度、视图生成和修补的降本逻辑。","报告讨论虚拟制作，不是2D转3D；可作为相邻工作流成本基线，不能宣称本项目已实现相同比例节省。","deloitte_virtual_production_future_content_creation_2024.pdf","https://www.deloitte.com/content/dam/assets-zone3/us/en/docs/industries/technology-media-telecommunications/2024/us-tmt-the-future-of-content-creation-virtual-production.pdf"
+"V03","B-行业分析估计","影视GenAI降本","Morgan Stanley","How AI Benefits—and Threatens—the Entertainment Industry","2025","Morgan Stanley Research估计GenAI可使媒体行业成本降低约10%，电视和电影最多约30%；影视内容资本支出约占总费用50%，数字/虚拟组件还可减少重拍及前后期成本。","网页打印PDF第1-3页","key_charts/02_cost_value/morgan_stanley_2025_p01_cost_estimate.png; key_charts/02_cost_value/morgan_stanley_2025_p03_tv_film_savings.png","支撑AI影视制作的行业级降本空间，并说明后期、VFX和虚拟组件是主要价值环节。","这是研究机构的行业估计，覆盖广义GenAI而非2D转3D；最多30%不是已验证的统一基准。","morgan_stanley_ai_entertainment_costs_2025_official_web_print.pdf","https://www.morganstanley.com/insights/articles/ai-in-media-entertainment-benefits-and-risks"
+"V04","B-相邻制作案例","云端后期降本","MovieLabs","Industry Forum Spring Summit 2026 Presentation Deck","2026","JioStar/Postudio案例称云端虚拟编辑比本地部署成本低30%以上，所需IT人员减少85%，实现零CAPEX和100%利用率。","PDF第146、148页","key_charts/02_cost_value/movielabs_2026_p146_jiostar_case.png; key_charts/02_cost_value/movielabs_2026_p148_jiostar_outcomes.png","说明模块化、云化、批量化媒体处理可降低基础设施与运维成本，适合映射项目的长视频批处理和服务化部署。","案例衡量的是云端编辑平台，不是立体转换算法；应作为工程部署价值旁证。","movielabs_industry_forum_spring_summit_2026.pdf","https://movielabs.com/prodtech/Industry-Forum-Spring-Summit-2026-Presentation-Deck.pdf"
+"V05","B-行业组织综合证据","AI内容生产效率","World Economic Forum","Artificial Intelligence in Media, Entertainment and Sport","2025","报告称文生图GenAI可使人类创意生产力提高约25%；LLM可能减少编辑人员72%的工作时间；自动资产管理案例的相关运营成本最多可下降75%。","PDF第10、14、15页，Figures 2-3","key_charts/02_cost_value/wef_2025_p10_ai_creative_value.png; key_charts/02_cost_value/wef_2025_p14_ai_productivity_cost.png; key_charts/02_cost_value/wef_2025_p15_ai_content_creation.png","用于说明内容生成、编辑、跨帧对象处理和资产管理的自动化价值。","各数字来自报告引用的不同研究或案例，任务和口径不同；不可合并成一个项目ROI。","wef_ai_in_media_entertainment_sport_2025.pdf","https://reports.weforum.org/docs/WEF_Artificial_Intelligence_in_Media_Entertainment_and_Sport_2025.pdf"
+"V06","B-跨行业调查","数字化ROI与成本压力","Autodesk","2025 State of Design & Make","2025","5594名行业领导者调查显示，多数受访组织认为数字化转型带来的客户满意度、生产力或创新改善达到50%以上；33%将成本控制列为首要挑战，M&E被描述为AI扰动和收益最明显的行业。","PDF第7、38、45-46页","key_charts/02_cost_value/autodesk_2025_p07_digital_transformation_roi.png; key_charts/02_cost_value/autodesk_2025_p38_media_ai_impact.png; key_charts/02_cost_value/autodesk_2025_p46_cost_control.png","为媒体制作数字化投资、AI采用和成本控制优先级提供大样本调查图表。","50%+指受访者对改善幅度的分档回答，不是严格财务核算的项目ROI；调查覆盖Design & Make多个行业。","autodesk_state_design_make_2025.pdf","https://adsknews.autodesk.com/app/uploads/2025/04/2025-SDM-Report-%E2%80%93-Final.pdf"
+"P01","A-直接竞品能力","Stereo Conversion产品能力","Prime Focus / DNEG","Investor Presentation January 2026","2026","DNEG将Stereo Conversion列入端到端媒体内容处理生态，并与VFX、虚拟制作、沉浸式体验、后期和发行并列。","PDF第9页","key_charts/03_competitor_product/primefocus_2026_p09_stereo_conversion_capability.png","直接证明大型VFX厂商仍将2D/3D立体转换作为商业能力，可作为本项目竞品和产业验证。","投资者材料仅展示能力版图，没有披露算法架构、精度、处理时长或单项目成本。","primefocus_investor_presentation_jan_2026.pdf","https://www.primefocus.com/wp-content/uploads/2026/01/PFL-Investor-Presentation-January-2026-1.pdf"
+"P02","B-竞品公司市场口径","内容转型可寻址市场","Prime Focus / DNEG","Investor Presentation January 2026","2026","公司将企业内容转型的全球机会描述为2030年超过1300亿美元，业务包括媒体资产管理、生成式AI内容创作和营销编排。","PDF第4、17页","key_charts/03_competitor_product/primefocus_2026_p04_business_and_products.png; key_charts/03_competitor_product/primefocus_2026_p17_content_tam.png","可说明头部内容服务商正从单一VFX扩展到AI内容生成、管理和分发平台。","这是公司投资者口径，且汇总了多个第三方市场；不是Stereo Conversion单项TAM。","primefocus_investor_presentation_jan_2026.pdf","https://www.primefocus.com/wp-content/uploads/2026/01/PFL-Investor-Presentation-January-2026-1.pdf"
+"P03","B-相邻应用落地","主题乐园与数字演出","Prime Focus / DNEG","Investor Presentation January 2026","2026","DNEG披露其为环球主题乐园制作屏幕内容，并称Disney与Comcast主题乐园及景点2025年收入超过400亿美元；公司正制作两场全长数字演唱会，FY26及以后收入超过6000万美元。","PDF第14页","key_charts/03_competitor_product/primefocus_2026_p14_immersive_business.png","支撑2D转3D/空间内容在主题乐园、数字演出和沉浸式大屏中的应用方向。","数字来自企业投资者材料，部分为关联大盘或项目管线，不等于本项目可获得收入。","primefocus_investor_presentation_jan_2026.pdf","https://www.primefocus.com/wp-content/uploads/2026/01/PFL-Investor-Presentation-January-2026-1.pdf"
+"P04","B-行业趋势","视频广告制作","PwC","Global Entertainment & Media Outlook 2025-2029: Hong Kong Summary","2025","报告指出GenAI正在通过压缩成本和时间改变视频广告制作，AI采用早期通常先在后台运营中体现为降本增效。","PDF第27页","","支撑广告、品牌内容和短视频批量立体化的商业方向。","定性判断，没有给出2D转3D专项节省比例。","pwc_global_entertainment_media_outlook_2025_2029.pdf","https://www.pwchk.com/en/tmt/entertainment-and-media-outlook-2025-2029.pdf"

2d_to_3d_market_tech_reports/mckinsey_value_creation_in_the_metaverse_2022.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6717312ce0e99d0d8f002111044564e65c328c80dba3afcc695f7eb2154459eb
+size 7380236

2d_to_3d_market_tech_reports/morgan_stanley_ai_entertainment_costs_2025_official_web_print.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3a0b864f3666809d9e901a1f302e8aad3b14e4eedd281cd9cd6a93ae7387655
+size 120818

2d_to_3d_market_tech_reports/movielabs_2030_vision_evolution_media_creation.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ba40c3e845dec7e29436d5972df9f9a2f668f4ee752b3e08d98fc974cc334b1
+size 18426397

2d_to_3d_market_tech_reports/movielabs_industry_forum_spring_summit_2026.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef5a0dd1075029dd5ad6abcc833fd99f1e750d962712090ed3bb5588cd9f6fcc
+size 71896338

2d_to_3d_market_tech_reports/movielabs_interoperability_media_creation.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbfcae32b751a3a9f376656ccbda20a23cae1d57f4b056af87894a48a5b08a76
+size 21518582

2d_to_3d_market_tech_reports/movielabs_software_defined_workflows.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9212f14b180f67c648ada0eb5f1af1c9a8a84be6b7013e2e03ee3bebeaeaba27
+size 15017060