Google's VEO 3 has a lot to say... (Tutorial + Flow Examples)

Gabe Michael
22 May 202509:01

TLDRThis tutorial explores Google's VO3 within the Flow ecosystem, highlighting its unique features like automatic sound effects and camera movement options. The presenter tests various prompts, such as a 1980s robot stargazing and an alien scene, noting that some features switch to V2. Despite limitations, the results are impressive. The cost is detailed, with V3 generations costing around $3 per 8-second clip. The presenter also showcases a text-to-image project with speech capabilities, demonstrating VO3's potential for creative storytelling.

Takeaways

  • 🤖 Google has launched VEO 3 (VO3) within its new Flow ecosystem, introducing enhanced video generation capabilities.
  • 🎥 VO3 allows users to input creative prompts, such as a '1980s robot stargazing on a suburban roof,' to generate stylized video clips.
  • 🔊 VO3 now automatically includes sound effects in generated videos, even when they’re not explicitly prompted.
  • 🎬 The platform supports camera movements like dolly, pan, tilt, and orbit—but currently only when using V2, not VO3.
  • 🖼️ Users can generate videos from still frames and add scene instructions, although uploading custom assets is not yet available.
  • 🧪 The 'ingredients to video' feature lets users mix different elements (characters, scenes, etc.) without specifying categories.
  • 💰 Access to VO3 requires the Google AI Ultra plan, which costs $125/month for the first 3 months, increasing to $250/month.
  • 💳 Each VO3 generation consumes 150 credits (approx. $3 per 8-second clip), and users receive 12,500 credits monthly under Ultra.
  • 📉 VO3 sometimes defaults back to V2 depending on feature compatibility, affecting visual fidelity and feature access.
  • 🧠 The script concludes withVEO 3 Features Overview a powerful poetic recitation (Rudyard Kipling’s 'If'), highlighting VO3's potential for rich storytelling with voice and music integration.

Q & A

  • What is Google's VEO 3 and what is its main purpose?

    -Google's VEO 3 is part of its new ecosystem, Flow, which aims to provide advanced video generation capabilities, combining text-to-video features with AI-powered tools. It allows users to create videos with automatic sound effects, camera movements, and other dynamic elements based on simple prompts.

  • What unique feature did the author discover while using VEO 3?

    -The author discovered that VEO 3 automatically adds sound effects to generated videos, even without the user requesting it in the prompt. This feature stands out in the new system and adds an extra layer of depth to video creation.

  • What is the camera movement feature in VEO 3?

    -VEO 3 includes a feature that allows users to select camera movements such as dolly in, dolly out, tilt up/down, pan left/right, orbit, and more. However, this feature was not yet available in the V3 version, and the author had to switch to V2 to use it.

  • Why was the author unable to use V3 for certain tasks?

    -The author was unable to use V3 for certain tasks because the camera movement features are currently not supported in V3. As a result, the system switched to V2, which allowed the cameraVEO 3 Features Overview movements to be applied.

  • What does the 'ingredients to video' feature in VEO 3 do?

    -The 'ingredients to video' feature allows users to combine different elements, such as characters, scenes, and objects, into a single video. Users only need to provide a description of the scene, and the system automatically assembles it without needing to specify details like character or style.

  • What is the significance of the Google AI Ultra plan for VEO 3?

    -The Google AI Ultra plan is necessary for accessing VEO 3. Users on the Google AI Pro plan do not have access to VEO 3 unless they upgrade to the Ultra plan, which provides more advanced features such as access to Flow and VEO 3.

  • What is the cost structure for using VEO 3 on Google AI Ultra?

    -The Google AI Ultra plan costs around $124 to $125 per month for the first three months, after which the price increases to $250 per month. Each V3 generation costs 150 credits, and with 12,500 credits per month, this amounts to about $3 for each 8-second clip.

  • What challenges did the author face while using VEO 3 for video generation?

    -The author faced challenges related to the current limitations of VEO 3, such as not being able to use camera movements and running out of credits. Additionally, the system sometimes skipped over certain elements, like the Easter Bunny in one of the test shots.

  • How does the author feel about the potential of VEO 3?

    -Despite some limitations, the author is impressed with VEO 3's potential, especially its ability to generate complex and creative scenes, such as a robot and a woman stargazing or a text prompt turning into a well-executed video.

  • What poem did the author include in the final project, and why?

    -The author included the poem 'If—' by Rudyard Kipling in the final project, possibly to test the platform's text-to-video capabilities, demonstrating how the system can handle poetic text and convert it into video content. The result was a fascinating mix of text and visuals that aligned with the emotional tone of the poem.

Outlines

00:00

🤖 Exploring VO3 and AI Video Generation in Google's Flow

This paragraph details a comprehensive walkthrough of Google's newly released VO3 video generation model within its Flow ecosystem. It begins with a whimsical reference to the woodchuck tongue twister, blending humor with technical exploration. The narrator introduces the simple interface of Flow and initiates a prompt involving a 1980s robot and a stargazing scene, highlighting that unexpected audio effects are now embedded automatically. The discussion moves into working with frame selection and testing prompts using pre-existing images, including a vivid scene with a glowing egg and a tentacled creature. It then shifts focus to camera movement options such as dolly, pan, tilt, and orbit, explaining that VO3 does not currently support these features, leading to a switch back to the older V2 model. The narrator experiments with scene creation by adding elements like aliens and an Easter Bunny in a suburban backyard, encountering limitations in rendering and generation accuracy. Despite technical hiccups, including the AI skipping certain elements, the narrator remains optimistic. They further describe a creative video transition from Earth to an alien planetVO3 Flow AI Walkthrough, echoing the initial robot and woman scene. Finally, a nostalgic romantic vignette set in a 1984 high school hallway ties back to the woodchuck phrase, demonstrating the model’s ability to render text-based prompts visually with high fidelity.

05:01

💸 Pricing, Credits, and a Poetic AI Showcase

This paragraph shifts to practical considerations of using Google’s AI video platform. It outlines the credit system and subscription pricing: 12,500 credits per month with VO3 generations costing 150 credits each—roughly $3 per 8-second video. The current subscription offer is $124–$125/month for the first three months, increasing to $250/month thereafter. Access to VO3 is exclusive to the Google AI Ultra plan, while lower-tier plans like AI Pro grant limited access to Flow and Whisk but exclude VO3. The paragraph transitions into a showcase project testing the combination of text-to-image generation with embedded speech. A powerful poetic monologue, likely adapted from Rudyard Kipling’s “If—,” is used to illustrate the platform's capacity to integrate literary depth with multimedia visuals. The passage explores themes of perseverance, integrity, humility, and resilience. It ends on a reflective, inspirational note, enriched by background music, underlining the potential of this AI tool not just for technical experiments but also for emotive storytelling.

Mindmap

Keywords

💡VEO 3

VEO 3 refers to the latest version of Google’s AI video generation model, which is part of the Flow ecosystem. It allows users to generate high-fidelity video clips from text prompts. In the script, VEO 3 is used to demonstrate new video features, such as embedded sound effects and advanced scene building, although it is not yet compatible with some camera movements.

💡Flow

Flow is the new ecosystem introduced by Google that houses VEO 3 and other creative tools like Whisk. It acts as the user interface or environment where users input prompts and interact with different generation models. The video describes Flow as a central platform where prompts are entered and assets are managed.

💡Prompt

A prompt is a piece of text entered by the user to instruct the AI model on what kind of image or video to generate. For example, 'a 1980s robot sitting on top of a suburban home stargazing' is used in the video as a sample prompt to create a visual scene. Prompts are crucial in guiding the AI's output in VEO 3.

💡Camera Movements

Camera movements such as 'dolly in,' 'tilt up,' and 'pan left' refer to dynamic changes in the camera angle or position during a video scene. The script explains that while these movements are selectable, they are not yet supported in VEO 3 and default to an earlier version (V2). This highlights the current limitations and future potential of VEO 3.

💡Credits

Credits are the internal currency used on Google's AI platform to generate content. Each video generation in VJSON Code CorrectionEO 3 costs 150 credits, and the user mentions running out of their monthly allocation. This introduces a financial aspect to using the tool, where users must manage their credit usage carefully.

💡Ingredients to Video

'Ingredients to Video' is a feature within Flow similar to Whisk, where users can add various visual elements (like characters or objects) and combine them into a scene using a prompt. The video demonstrates this by combining aliens, an Easter Bunny, and a backyard setting to create a cohesive narrative.

💡Scene Builder

Scene Builder is a tool within Flow that allows users to scrub through video clips and select frames to use as assets. This provides a more granular level of control over video creation. In the script, it's used to illustrate how users can save specific frames and incorporate them into projects.

💡Google AI Ultra

Google AI Ultra is a subscription tier that provides access to VEO 3 and other advanced tools in the Flow ecosystem. The narrator mentions needing this plan to access VEO 3, distinguishing it from the Pro plan, which only offers limited tool access. This shows the tiered nature of access within Google's AI platform.

💡Sound Effects

Sound effects are automatically generated audio elements that accompany the video clips, enhancing the immersive quality of the scenes. The user notes that these were not prompted but were still included, showcasing a new feature of VEO 3 where audio is contextually added based on the visual prompt.

💡Text-to-Video

Text-to-video refers to the technology that allows video generation from written prompts. VEO 3 exemplifies this by taking descriptive language and turning it into moving images. The script shows various use cases, including robots stargazing and an Easter Bunny in a backyard, emphasizing the creative potential of this AI capability.

Highlights

null

A unique feature in VEO 3: sound effects are automatically included with video prompts, without being requested.

VEO 3's ability to handle video frames is demonstrated with user-friendly tools for adjusting and reviewing each frame.

Exploration of camera movements like dolly in, dolly out, pan left/right, and more, to enhance video scene dynamics.

Integration of camera movements with prompt-generated content, but currently only functional within V2, not V3.

The Scene Builder tool allows users to scrub through clips, save frames as assets, and generate new video sequences.

Test projects reveal that VEO 3 performs well, even though it requires switching to V2 for certain advanced features.

Prompt-driven scene generation produces visually compelling results, as seen in a 1980s robot and woman stargazing scene.

A sequence demonstrating the power of VEO 3: a high school hallway with a robot and a girl discussing a famous tongue twJSON Code Correctionister.

The ease of creating short video sequences from simple text prompts, showcasing VEO 3's potential for creative storytelling.

The integration of text-to-image and speech features within Google’s ecosystem adds significant depth to video generation.

Users are introduced to Google AI’s subscription-based pricing model, with different access tiers for V3.

VEO 3's text-to-image features allow for custom visualizations, making it easy to create stunning imagery from text descriptions.

Advanced scene composition, such as flying over a suburban backyard with aliens and an Easter bunny, demonstrates the tool’s flexibility.

The subscription cost for Google AI Ultra, which grants access to VEO 3, is outlined with a breakdown of credit usage and costs.