<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:news="http://www.google.com/schemas/sitemap-news/0.9">
  <url>
    <loc>https://cinecourier.com/fr/one-shot-any-view-retournage-video-erreur-de-pose/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>fr</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T23:06:14+01:00</news:publication_date>
      <news:title>One Shot Any View : nouveaux trajets de caméra, erreur de pose inférieure à deux références selon les auteurs</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/it/one-shot-any-view-reshooting-video-errore-di-posa/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>it</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T23:04:16+01:00</news:publication_date>
      <news:title>One Shot Any View rigira i clip su nuovi percorsi: per gli autori, errore di posa inferiore a due baseline</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/es/one-shot-any-view-recaptura-video-error-de-pose/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>es</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T23:03:03+01:00</news:publication_date>
      <news:title>One Shot Any View: nuevas trayectorias de cámara y menos error de pose que dos métodos, según sus autores</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/fr/benchmark-physresponse-modeles-video-erreurs-physiques/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>fr</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T23:01:41+01:00</news:publication_date>
      <news:title>Benchmark PhysResponse : dix modèles vidéo ignorent les paramètres physiques, selon les auteurs</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/de/scoutcache-trainingsfreie-beschleunigung-self-forcing-longlive/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>de</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T22:57:52+01:00</news:publication_date>
      <news:title>ScoutCache: trainingsfreie Methode meldet bis zu 2,28× Beschleunigung bei Self-Forcing</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/it/scoutcache-accelerazione-senza-training-self-forcing-longlive/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>it</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T22:56:36+01:00</news:publication_date>
      <news:title>ScoutCache, metodo senza training, dichiara un'accelerazione fino a 2,28× su Self-Forcing</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/es/scoutcache-aceleracion-sin-entrenamiento-self-forcing-longlive/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>es</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T22:55:22+01:00</news:publication_date>
      <news:title>ScoutCache, método sin entrenamiento, informa de una aceleración de hasta 2,28× en Self-Forcing</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/fr/movieroll-video-longue-hierakv-publication/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>fr</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T22:53:48+01:00</news:publication_date>
      <news:title>Cadre vidéo longue MovieRoll : jusqu'à 78 % de mémoire KV-cache en moins selon les auteurs</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/it/movieroll-video-lunghi-hierakv-rilascio/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>it</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T22:52:37+01:00</news:publication_date>
      <news:title>MovieRoll per video lunghi: gli autori dichiarano fino al 78% di memoria KV-cache in meno</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/es/movieroll-video-largo-hierakv-lanzamiento/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>es</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T22:51:55+01:00</news:publication_date>
      <news:title>MovieRoll, vídeo largo: los autores indican hasta un 78 % menos de memoria de caché KV</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/de/guidance-skala-demand-score-text-zu-video-fehler/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>de</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T22:51:04+01:00</news:publication_date>
      <news:title>Studie: Ein Demand-Score sagt voraus, wann höhere Guidance-Skala fehlerhafte Text-zu-Video-Clips behebt</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/fr/fastframes-latence-interactive-diffusion-video/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>fr</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T14:40:57+01:00</news:publication_date>
      <news:title>Selon l'article FastFrames, 36 ms par image pour un modèle vidéo de 1,3B sur un B200</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/de/fastframes-interaktive-latenz-video-diffusion/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>de</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T14:39:46+01:00</news:publication_date>
      <news:title>FastFrames-Paper nennt 36 ms pro Frame für ein 1,3B-Videomodell auf einer B200</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/it/fastframes-latenza-interattiva-video-diffusion/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>it</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T14:38:39+01:00</news:publication_date>
      <news:title>Il paper FastFrames riporta 36 ms per frame per un modello video da 1,3B su una B200</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/es/fastframes-latencia-interactiva-difusion-de-video/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>es</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T14:37:41+01:00</news:publication_date>
      <news:title>El artículo de FastFrames indica 36 ms por fotograma con un modelo de vídeo de 1,3B en una B200</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/fastframes-interactive-latency-video-diffusion/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T14:36:24+01:00</news:publication_date>
      <news:title>FastFrames paper reports 36 ms per frame for a 1.3B video model on one B200</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/fr/evaluation-inpainting-video-recuperabilite-egs/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>fr</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T10:50:20+01:00</news:publication_date>
      <news:title>Des pistes oracle permettent à EGS de combler 43 % et 56 % de l'écart, selon un article</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/de/rueckgewinnbarkeit-video-inpainting-bewertung-egs/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>de</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T10:49:53+01:00</news:publication_date>
      <news:title>Studie: Mit Oracle-Tracks schließt EGS 43 % und 56 % der Lücke</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/it/valutazione-inpainting-video-recuperabilita-egs/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>it</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T10:49:22+01:00</news:publication_date>
      <news:title>Studio: con tracce oracolo EGS colma il 43% e il 56% del divario</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/es/evaluacion-inpainting-video-recuperabilidad-egs/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>es</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T10:48:39+01:00</news:publication_date>
      <news:title>Un estudio indica que EGS con pistas oráculo cierra el 43 % y el 56 % de la brecha</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/recoverability-aware-video-inpainting-evaluation-egs/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T10:47:58+01:00</news:publication_date>
      <news:title>Paper reports oracle tracks let EGS close 43% and 56% of the gap</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/de/physresponse-benchmark-videomodelle-physikfehler/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>de</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T09:15:37+01:00</news:publication_date>
      <news:title>PhysResponse-Benchmark: Zehn Videomodelle verfehlen physikalische Parameter, so die Autoren</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/it/benchmark-physresponse-modelli-video-errori-fisica/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>it</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T09:14:50+01:00</news:publication_date>
      <news:title>Benchmark PhysResponse: dieci modelli video sbagliano i parametri fisici, dicono gli autori</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/es/benchmark-physresponse-modelos-video-errores-fisicos/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>es</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T09:14:03+01:00</news:publication_date>
      <news:title>Benchmark PhysResponse: diez modelos de vídeo fallan en parámetros físicos, según los autores</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/one-shot-any-view-video-reshooting-pose-error/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T09:13:19+01:00</news:publication_date>
      <news:title>One Shot Any View reshoots clips on new camera paths; authors report lower pose error than two baselines</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/physresponse-benchmark-video-models-physics-errors/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T09:08:37+01:00</news:publication_date>
      <news:title>PhysResponse benchmark: ten video models miss physical parameters, authors report</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/h3remover-video-object-removal-without-paired-data/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-19T09:03:02+01:00</news:publication_date>
      <news:title>H3Remover, an RL-tuned video object remover, reports large gains over original H3</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/scoutcache-training-free-speedup-self-forcing-longlive/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T23:00:02+01:00</news:publication_date>
      <news:title>ScoutCache, a training-free method, reports up to 2.28× speedup on Self-Forcing</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/movieroll-long-form-video-hierakv-release/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T22:51:30+01:00</news:publication_date>
      <news:title>MovieRoll long-video framework: authors report up to 78% less KV-cache memory</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/it/punteggio-domanda-guidance-scale-fallimenti-text-to-video/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>it</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T19:56:35+01:00</news:publication_date>
      <news:title>Studio: un punteggio di domanda prevede quando più guidance scale corregge i clip text-to-video falliti</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/es/puntuacion-demanda-escala-guia-fallos-texto-a-video/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>es</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T19:55:49+01:00</news:publication_date>
      <news:title>Una puntuación de demanda predice cuándo una mayor escala de guía repara clips de texto a vídeo fallidos</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/guidance-scale-demand-score-text-to-video-failures/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T19:48:45+01:00</news:publication_date>
      <news:title>Paper: a demand score predicts when a higher guidance scale fixes failed text-to-video clips</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/lynnreal-hand-hand-controlled-video-world-model/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T18:24:39+01:00</news:publication_date>
      <news:title>LynnReal-Hand: hand-controlled video world model, with a fast VAE decoder</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/pika-h3-flex-minimax-h3-enact-pricing/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T18:10:25+01:00</news:publication_date>
      <news:title>Pika details H3 Flex, a post-trained MiniMax H3 model, and its plan prices</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/reference-forcing-long-video-drift-correction/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T15:58:03+01:00</news:publication_date>
      <news:title>Reference Forcing, per its authors, cuts drift in long AI video</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/longcat-video-avatar-1-6-arena-second-place/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T14:48:19+01:00</news:publication_date>
      <news:title>LongCat-Video-Avatar 1.6 ranks second of six in a preference Arena, its authors report</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/motion-aware-reward-model-video-editing-ve-bench/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T08:43:48+01:00</news:publication_date>
      <news:title>Motion-aware reward model cuts motion-suppressed edit picks to 0%, authors report</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/sciworldbench-video-models-science-experiments/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T06:52:38+01:00</news:publication_date>
      <news:title>SciWorldBench: video models score 1.4% to 24.8% on paired science experiments</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://cinecourier.com/en/memcrafter-memorability-text-to-video-prompt-refinement/</loc>
    <news:news>
      <news:publication>
        <news:name>Cinecourier</news:name>
        <news:language>en</news:language>
      </news:publication>
      <news:publication_date>2026-11-18T04:44:03+01:00</news:publication_date>
      <news:title>MemCrafter paper reports 0.908 memorability score and 85.00% human hit ratio</news:title>
    </news:news>
  </url>
</urlset>
