Filmora 15 : Le réglage IA pour des vidéos parfaites

This tutorial explores several lesser-known but powerful features within Filmora, focusing on enhancing video creation and optimization for various platforms. The first key feature demonstrated is **automatic reframing**. This is particularly useful for social media content creators who need to adapt their videos for different aspect ratios, such as horizontal for YouTube, square, or vertical for platforms like Instagram or TikTok. The process is shown by taking a horizontal video and converting it to a vertical format. Users can right-click on the imported video, select "Auto Reframe," and then choose the desired aspect ratio (e.g., 9:16). The software allows for manual adjustment of the frame within the reframing window. The speaker highlights using the spacebar for pausing and playing the video to precisely adjust the frame as the action unfolds, especially in clips with camera movement. Filmora then smooths out the movement to ensure the subject stays within the reframed area. Once satisfied, the reframed video can be exported. The tutorial then moves to **AI-powered content extension**, addressing a scenario where a video needs to be longer than the original footage. In this case, a 25-second clip needs to be extended to at least 30 seconds for YouTube. The speaker explains that simply slowing down an already slow-motion video isn't ideal. Instead, Filmora's AI tools are employed. The first step involves capturing a still image from the original video using the snapshot tool. This image is then used as a reference in the "AI Image" feature. A prompt is entered to describe a desired scene, building upon the reference image. For example, the prompt requests the jogger to be finishing her run, appearing warm, and drinking cool water from a plastic bottle, still against the graffiti-covered wall. After generating the AI image, the speaker demonstrates how to remove the "Generated by AI" watermark using another AI tool, "AI Object Removal." This refined image is then used in the "Image to Video" AI feature. A new prompt describes the action of the jogger drinking water, and the AI generates a short video clip. This AI-generated video clip is then integrated into the timeline, extending the original footage to the desired length. The speaker also touches upon adjusting the duration of the AI-generated clip and removing any watermarks from it. The tutorial then introduces **AI Music Generation**. Two methods are presented. The first, "AI Ambient Music Generation 2.0," analyzes the timeline content to create adaptive background music. However, the speaker finds this method to be repetitive and less engaging. The preferred second method involves using the "AI Music" feature. Here, users can specify genre (e.g., funk), mood (e.g., happy), and even provide a descriptive prompt about the video's content. The AI then generates music based on these parameters. The speaker demonstrates generating a 34-second funk track with French lyrics, expressing pleasant surprise at the quality and the inclusion of lyrics. This AI-generated music is then downloaded and added to the project, with the excess audio trimmed to match the video duration. A significant portion of the tutorial is dedicated to **AI Subtitle Generation**. The speaker emphasizes the importance of subtitles for social media engagement, especially on platforms where users might watch with sound off. Filmora's "AI Subtitle" feature allows for automatic subtitle creation from spoken dialogue. The process involves importing the video, selecting the relevant clip on the timeline, choosing the transcription language (French in this case), and then clicking "Generate." The AI analyzes the audio and creates a new subtitle track on the timeline. Users can then double-click on this track to access a dedicated subtitle editing window. Here, they can review and correct any transcription errors, adjust text formatting (font, size, color, alignment), and apply animations. The speaker demonstrates customizing the appearance of subtitles by changing the fill color to yellow, adding a semi-transparent black background with rounded edges, and then applying these settings to all subtitles with a single click. This feature is highlighted as a powerful tool for accessibility and broader reach. Finally, the tutorial reveals a **"Fake 4K" export trick** to improve video quality on platforms like YouTube, especially when viewed on larger screens. The speaker explains that YouTube's default compression codec (AVC1) can degrade video quality. By exporting the video in a "fake 4K" resolution (3840x2160) using the HEVC (H.265) codec and a high bitrate, YouTube is prompted to use a superior compression codec (VP9) that preserves more detail. This results in a better viewing experience, even if the original footage was 1080p. The process involves going to the export settings, selecting HEVC as the encoder, choosing 3840x2160 resolution, setting the quality to "Superior," and manually adjusting the bitrate (e.g., to 50,000 kbps). The speaker also advises keeping the color space as SDR for broader compatibility. The ability to save these custom export settings as a preset is also mentioned, streamlining future exports. Throughout the tutorial, the speaker reiterates that this is only a glimpse of Filmora's capabilities, hinting at future tutorials covering more advanced features like automatic video sequencing, dead-time removal, and sound enhancement. The video concludes with a reminder about available affiliate links and a call to action for likes and comments.