Google Flow Tutorial - How To Use Googles Flow (Complete Guide)

4 sections

  • 0:00Click 'create new project' for any new video creation; serves as the main workspace.
  • 0:07Use the scene builder to develop fully-fledged short films, with easy navigation back to main interface.
  • 0:24Allows more control by converting specific frames into videos, not just text-to-video generation.
    • This drop-down menu isn't just text to video; you also have frames to video, giving you more control.0:24
  • 0:34Add images by generating new ones or uploading, with generating images recommended for beginners.
  • 0:53Generate an image as the first frame, then animate it, providing more customization.
    • Generating the first image helps you have more control, then you can animate it for your video.0:53
  • 1:12Generate images rapidly using Google’s AI with four options at a time for variety.
    • Google’s AI generates four images at a time, offering quick options for variety.1:12
  • 1:39Choose up to four prompts to maximize variety and aesthetic appeal. Starting with fewer prompts limits options but may simplify the process.
  • 1:47Generating several images ensures better chances of matching the desired look, as initial results may not be perfect or accurate.
    • Hard to get the image correct the first time; multiple images illustrate why you need several tries.1:47
    • Updating prompts with details like make and model yields more accurate visuals.2:05
  • 2:13Modify prompts with additional details, then reuse them for consistency and efficiency. Multiple images can be used simultaneously for comparison.
    • Reusable prompts allow quick generation and adjustments without retyping everything.2:13
  • 2:46Create multiple images to select the best ones. Some may hallucinate or distort details; generating four helps get quality options.
  • 3:09Reuse existing images as prompts to quickly generate similar or related visuals, saving time during creative adjustments.
    • Generating four images is recommended because some may hallucinate or be subpar.3:09
  • 3:26While optional, end frames can be saved for future use. Adjust video settings before rendering, choosing appropriate models for quality and cost.
  • 3:42Select from three models—Fast VO, V2, or V3—based on quality needs and credits. V3 offers the best physics and audio, but at a higher cost.
    • Choose video models wisely to balance quality and credits: V3 for best physics and sound, V2 or Fast for cost savings.3:42
  • 4:22Monitor remaining credits in your profile. For cost efficiency, prefer Fast or V2 models unless high quality with sound is necessary.
  • 5:16V3 offers high quality and sound, but can add undesirable grid lines. V2 provides better physics without audio; Fast is quick but lower quality.
  • 6:13AI-generated cars look more impressive, with better physics, enhancing the realism of race footage.
    • The physics are so much better and the cars look more impressive in terms of the overall footage.6:13
    • V3 experience is probably your best bet when generating clips that need sound and vocals.6:54
  • 7:04Transitions between shots can seem abrupt or phased, but they can be improved through editing and post-production tweaks.
  • 7:18Adding sound and effects with V3 experience enriches the immersion, making the clips more engaging.
  • 7:40V3 experience is recommended for clips requiring sound and vocals, while VO fast is suitable for simple, cost-effective cinematics.
    • If you want to use camera angles, it's best to add that in the description for better control.7:50
  • 8:05Use a single prompt for initial output to avoid wasting credits; iterate to refine the quality and control over results.
    • Make sure your first prompt is just on one video to avoid wasting credits and to iterate effectively.8:40
    • Once you've generated clips, you can check them out in full screen or do further editing.9:53
  • 10:16Some clips may not play correctly or appear reversed due to camera angles or software issues; refreshing often helps resolve problems.
  • 10:22Ensure objects move in the same direction as the camera to prevent visual errors like reversed clips, especially when using cost-effective models.
    • Sometimes the clip just doesn't play or appears reversed because of the camera angle or software glitch; refreshing the browser often helps.10:22
    • Align objects to move in the same direction as the camera to avoid visual errors like reversed or inaccurate clips.10:48
  • 11:34Select and cut the useful parts of generated clips, then merge moments that work, while removing undesired transitions like cars morphing or backtracking.
    • You can cut and merge the useful parts of clips, removing the sections where objects revert or morph unexpectedly.11:51
  • 12:50Use the last frame of a clip to add prompts for continuation or to generate new scenes, maintaining scene consistency.
    • Use the last frame of a clip as a prompt to continue or create a new scene, maintaining scene continuity.12:50
  • 13:23Hover over preferred frames to save as assets for easy access, enabling iterative editing or asset reuse across projects.
    • Hover over a frame and click 'save frame as asset' to reuse specific visuals in different parts of your project.13:23
  • 13:49Extend or trim clips within the timeline to improve shot accuracy and compensate for errors without re-generating footage.
    • Extending clips ensures better shot length and quality, especially when dealing with errors or incomplete scenes.14:04
  • 14:20Delete unwanted clips from the scene builder to streamline your project and organize your visual sequence.
    • Deleting unnecessary clips in the scene builder keeps your project organized and focused.14:20
  • 14:26Combine pre-selected images and models of cars, environments, and drivers to create realistic or stylized virtual scenes.
    • Loading your own assets like cars and backgrounds offers full control over scene design and realism.14:26
  • 14:58Loading specific assets like cars and backgrounds from your library allows precise scene control, especially for models not in training data.
    • To fix resolution issues, upscale clips or download in their original size before integrating into your project.14:58
  • 16:27To handle resolution issues, upscale clips or download in original size, ensuring quality for final presentation.

Frequently Asked Questions


Perfect for students, researchers, content creators and professionals

TubeMemo helps you turn any YouTube video into clear notes, summaries, and insights