{"id":3970,"date":"2024-02-16T08:59:25","date_gmt":"2024-02-16T00:59:25","guid":{"rendered":"https:\/\/www.1ai.net\/?p=3970"},"modified":"2024-02-16T08:59:25","modified_gmt":"2024-02-16T00:59:25","slug":"%e9%87%8d%e7%a3%85%ef%bc%81openai%e5%8f%91%e5%b8%83%e6%96%87%e7%94%9f%e8%a7%86%e9%a2%91%e6%a8%a1%e5%9e%8bsora%ef%bc%8c%e4%b8%80%e6%ac%a1%e5%8f%af%e7%94%9f%e6%88%901%e5%88%86%e9%92%9f%ef%bc%81","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/3970.html","title":{"rendered":"Big news! OpenAI releases Sora, a human-generated video model that can generate 1 minute of video at a time!"},"content":{"rendered":"<p>In the early morning of February 16,<a href=\"https:\/\/www.1ai.net\/en\/tag\/openai\" title=\"[View articles tagged with [OpenAI]]\" target=\"_blank\" >OpenAI<\/a>The official website released the innovative<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e6%96%87%e7%94%9f%e8%a7%86%e9%a2%91%e6%a8%a1%e5%9e%8b\" title=\"[Sees articles with tags]\" target=\"_blank\" >Vincent video model<\/a>\u2014\u2014<a href=\"https:\/\/www.1ai.net\/en\/tag\/sora\" title=\"[See articles with [Sora] label]\" target=\"_blank\" >Sora<\/a>.<\/p>\n<p>Judging from the effects of Sora generated videos displayed by OpenAI on its official website, the quality of the generated videos, resolution, text semantic restoration, video motion consistency, controllability, details, colors, etc. are excellent!<\/p>\n<p><strong>In particular, it can generate videos up to 1 minute long<\/strong>It is more powerful than mainstream products such as Gen-2, SVD-XT, and Pika.<\/p>\n<p>On September 21, 2023, OpenAI released the text-based graph model DALL\u00b7E 3. Together with the current Sora and the previous voice model Whisper, ChatGPT already has four multi-modal functions: text, image, video, and audio. Is AGI still far away from us?<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-3971\" title=\"640-141\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/02\/640-141.png\" alt=\"640-141\" width=\"554\" height=\"229\" \/><\/p>\n<p><strong>Sora Brief Introduction<\/strong><\/p>\n<p>At present, the field of Wensheng video has been unable to generate high-quality long videos due to reasons such as inter-frame dependency processing, training data, computing resources, and overfitting.<\/p>\n<p>The biggest technical breakthrough of Sora is that it can generate a 1-minute video while maintaining quality, which is very rare in the industry. This also once again demonstrates OpenAI&#039;s strong technical research and development capabilities in the field of large models.<\/p>\n<p>Sora is a diffusion model that generates videos by starting with a video of static noise and then gradually transforming the video by removing the noise in multiple steps.<\/p>\n<p><strong>Sora uses the same Transformer architecture as ChatGPT and uses the restatement technology in DALL-E 3. It is a method for generating highly accurate descriptive subtitles for visual training data.<\/strong>Therefore, Sora accurately restores the user&#039;s text prompt semantics during the video generation process.<\/p>\n<p>In terms of functions, in addition to text-generated videos,<strong>Sora can also generate videos from images and accurately animate image content. It can also extract elements from videos and extend them or fill in missing frames.<\/strong>, the functions are very comprehensive.<\/p>\n<p>OpenAI will release the Sora paper later, and &quot;1ai&quot; will bring you a more in-depth technical interpretation.<\/p>","protected":false},"excerpt":{"rendered":"<p>In the early morning of February 16, OpenAI released the innovative text-generated video model - Sora - on its official website. From the effect of Sora-generated video shown by OpenAI on its official website, it is very good in terms of generating video quality, resolution, text-semantic restoration, video action consistency, controllability, details, color, etc.! In particular, it can generate videos of up to 1 minute in length! Surpassing mainstream products such as Gen-2, SVD-XT, Pika, etc., it's a king's bomb right out of the gate. On September 21, 2023, OpenAI released the text-generated graph model DALL-E 3. Together with the current Sora and the previous speech model Whisper, ChatGPT already has 4 multimodal functions of text, image, video, and audio... is AGI still far away from us?<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[148,146],"tags":[190,1249,1248],"collection":[],"class_list":["post-3970","post","type-post","status-publish","format-standard","hentry","category-headline","category-news","tag-openai","tag-sora","tag-1248"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/3970","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=3970"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/3970\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=3970"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=3970"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=3970"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=3970"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}