Video için yapay zekâ promptları

Isolate all human speech from the background. Remove wind noise, microphone hum, and room echo. Normalize all speakers to a target loudness of -14 LUFS and a true peak of -1.5 dB.

Transcribe the audio track. Identify and create an edit list of all instances of the words 'um,' 'uh,' 'like,' 'you know,' and 'so.' Also identify all silent segments longer than 0.7 seconds. Do not delete them; present the list for review.

Separate the source audio into three distinct stems: one for dialogue, one for background music, and one for ambient sound effects. Export each as a 24-bit WAV file.

Analyze the dialogue track for harsh 'p' and 'b' sounds (plosives) and sharp 's' sounds (sibilance). Apply a de-esser and a dynamic EQ to reduce these artifacts without making the speaker sound lispy or muffled.

Using the provided transcript, create a new video sequence that only includes the sentences marked with '[[KEEP]]'. Ignore all other text.

Analyze this long take of a speaker. Create a new sequence that removes all silences and repeated phrases, resulting in a tight, energetic jump-cut edit.

I have 5 takes of the same monologue, labeled 'take_1' through 'take_5'. Analyze all five for speech clarity, pacing, and fewest filler words. Create a new sequence using only the best take.

Analyze the footage of the basketball game. Identify every time a player successfully makes a basket. Place a marker 1 second before each successful shot.

Transcribe all dialogue and generate an SRT file. Format captions to CEA-608 standards, with a maximum of 32 characters per line and a maximum of 2 lines per caption. Ensure no caption stays on screen for less than 1 second or more than 7 seconds.

Generate word-by-word animated captions for this 9:16 video. Use the font 'Proxima Nova Bold', size 90. Highlight the active word in yellow (#FFC700) and keep all other words white. Position the captions in the vertical center of the frame.

Transcribe the English dialogue, then translate it into Spanish (Latin America). Burn the Spanish subtitles into the video in the lower third. Use a semi-transparent black background box for readability.

Scan the transcript and identify all key technical terms, company names, and product names. Create a list of these terms with their corresponding timestamps to be used for creating on-screen callout graphics.

Create a 15-second, 4K, seamless loop. Abstract moving gradient of blue and purple hues, soft focus, slow, meditative movement. Cinematic quality.

Generate a 7-second, 4K, 24fps photorealistic shot. A chef's hands, viewed from above, meticulously plating a dessert in a high-end restaurant. Shallow depth of field, warm lighting. Match the style of the 'Chef's Table' series.

Analyze the dialogue in my timeline, which is about remote work and collaboration. Find and suggest 10 high-quality, royalty-free stock video clips that match these themes. Prioritize clips with natural-looking people and no obvious "stock photo" feel.

Take this 480p, 4:3 aspect ratio archival clip and upscale it to 1080p using AI. Enhance detail, reduce noise and compression artifacts, and use frame interpolation to convert the frame rate to 29.97 fps.

Analyze the color grade of Clip A (reference). Apply the same color palette, contrast curve, and saturation levels to Clip B. Pay special attention to matching the skin tones of the subject.

Create a color grade with a "summer blockbuster" look. High contrast, teal in the shadows, warm orange and gold in the highlights and skin tones. Saturated but not unnatural.

Create a mask that tracks the speaker's face throughout the clip. Invert the mask and slightly desaturate the background to make the subject pop.

Analyze this 60-minute podcast video. Identify 5-7 of the most engaging, self-contained moments suitable for YouTube Shorts and TikTok. Reframe each clip to a 9:16 aspect ratio, ensuring the primary speaker remains centered at all times. Add animated captions.

Analyze this full 25-minute episode. Identify the most emotionally impactful and visually interesting shots. Create a 60-second trailer with a clear narrative arc: hook, rising action, and call-to-action.

Take the final 16:9 sequence and export three versions: 1. A 9:16 version for TikTok/Reels with burned-in animated captions. 2. A 1:1 square version for an Instagram feed post. 3. A 4:5 vertical version for an Instagram feed post. Intelligently reframe the action for each aspect ratio.

On all new footage added to the 'Ingest' folder: 1. Create 1080p ProRes 422 Proxy files. 2. Generate and attach a transcript to each clip. 3. Analyze audio and apply 'Universal Dialogue Cleanup'. 4. Move the original source files to the 'Source_Media' folder and the proxies to the 'Proxies' folder.

Scan all clips in the project bin. Identify every clip where the keyword 'AI' is spoken in the transcript. Create a new sequence containing all of these clips in chronological order.

Prepare this project for archival. 1. Consolidate all used media into a single folder. 2. Generate a final 'As-Broadcast' script from the main sequence's captions. 3. Create an EDL (Edit Decision List). 4. Export a clean version (no graphics/text) and a dirty version (with graphics/text) in ProRes 4444. 5. Zip the entire project folder.