Video breakdownPro

Create 50 - 100 Viral Podcast-Style Short Videos per Month Using AI Image and Video Generation

Source: Spanish-language tutorial showing a full workflow to produce fake-podcast short-form video clips at scale using ChatGPT, Google Gemini, and Google Flow

1. Generate a podcast-set image using ChatGPT or Google Gemini. Open either tool and paste the prompt below to get a prompt you can feed to the image generator.Use the generated image prompt in ChatGPT's image tool (left sidebar > 'Images') or in Gemini's 'Create Image' feature. Download whichever result you prefer in full size.

   Write a detailed image-generation prompt (in English) that instructs an AI image tool to create a vertical (9:16) photo of a [man/woman] seated in a modern podcast studio setup with professional lighting, microphone, and dark background, suitable for short-form video content on TikTok, Reels, or YouTube Shorts.

   This prompt asks ChatGPT or Gemini to write you a ready-to-use image prompt. Replace [man/woman] with your preferred subject. Paste the output into the image-generation tool of your choice.
2. Generate 10 video ideas - each split into two 8-second clips - for your chosen niche. Paste the prompt below into ChatGPT or Gemini.Each idea should include the exact spoken line (dialogue) for each clip. Controversial or thought-provoking lines drive comments and shares, which boosts viral reach.

   Give me 10 short-form video ideas for the niche of [YOUR NICHE, e.g. personal development / pets / finance]. Each video must contain exactly two clips of 8 seconds each. For every clip provide: (1) a full cinematic video-generation prompt describing the same man already shown in my reference image, seated in a modern podcast studio, do not re-describe his appearance; (2) the exact dialogue line he appears to say - make every line bold, controversial, or provocative to encourage viewer comments. Format clearly as: Video 1 - Clip 1, Video 1 - Clip 2, etc. All text in English.

   Replace [YOUR NICHE] with your topic. The output gives you ready-to-paste cinematic prompts for the video generator AND the spoken line for each clip. Use one video's two clip prompts per session in Flow.
3. Open Google Flow and create a new project. Click 'New Project', then click the '+' (add) button and choose 'Upload Media' to upload the podcast-set image you downloaded in Step 1.Once uploaded, click 'Add to Request' so the image appears as the visual reference for your first video clip.
4. Generate Clip 1. With your image attached, paste Clip 1's cinematic prompt from Step 2 into the Flow prompt field. Set the format to 9:16 video, choose the 'Lite' model (costs 10 credits instead of 25), then click the send/generate arrow.The Lite model produces an 8-second clip at 10 credits. Wait 30 - 60 seconds for rendering. Preview the result; if acceptable, move to Clip 2.

   Vertical 9:16 podcast studio, same man from the reference image, he looks directly at camera and says: '[PASTE EXACT DIALOGUE LINE FROM CLIP 1 HERE]'. Cinematic lighting, shallow depth of field, professional podcast microphone in foreground.

   Replace the bracketed dialogue with the exact line from your Step 2 output for Clip 1. Keep all other details so Flow maintains visual consistency.
5. Generate Clip 2 as an extension of Clip 1. Click on the Clip 1 result inside Flow, then click the '+' (extend) button and choose 'Extend'. Paste Clip 2's cinematic prompt (same studio, new dialogue line) and click generate.Using the 'Extend' feature keeps the same visual environment and character across both clips, making the two-part video look seamless.

   Same podcast studio and same man, continuous scene. He now says: '[PASTE EXACT DIALOGUE LINE FROM CLIP 2 HERE]'. Maintain identical lighting, background, and camera angle as the previous clip.

   Replace the bracketed text with the Clip 2 dialogue from Step 2. The 'Extend' workflow tells Flow to continue from the existing clip rather than starting fresh.
6. Download the combined ~16-second video from Flow by clicking the video thumbnail and selecting 'Download'.Flow stitches both clips together. Each two-clip video costs 20 credits total. At 1,000 credits per billing cycle you can produce up to 50 videos (or 100 videos if you use single 8-second clips instead of pairs).
7. Edit the video in any video editor (the tutorial uses CapCut). Import the Flow file, then: (a) stretch the video slightly vertically to hide any watermark at the bottom; (b) play through and use the 'Split' tool at the exact moment the first clip ends to remove any glitchy transition frame; (c) optionally add a text overlay with the controversial quote; (d) export and upload to TikTok, Reels, orThe hard cut between clips can look like a natural podcast excerpt cut - lean into that aesthetic rather than trying to hide it. Leaving the last syllable slightly cut off makes it feel like a clip pulled from a longer show.