Author: Aaron Belchamber

Google’s Veo3 Video Production – What’s the Trial, How Much Does It Cost?

June 8, 2025 Aaron Belchamber No Comments

I’ve heard more and more about the impressive production quality of Google’s Veo 3 so I asked Gemini:
How does video production with Veo3 work regarding fair use and resale of generated clips? Is everything that is created in Veo3 owned by google? There is a “try now” button on the website, what is the trial period, what can you use from the trial period and what are the trial’s limitations? Finally, please summarize a budget cost for using it. If I am making 30 second high quality TV commercials that may require between 5 and 15 scenes, how are costs broken down?

…Unsure what Veo3 is? Check this summary out from Greg Isenberg (unfortunately there’s some bad language mid-way through — which, pet peeve of mine why do people feel the need to just drop f-bombs? I spent 5 years in the US Army and yes, foul language is often used, but KNOW your audience man! Lots of teenagers and kids watch your videos to learn something, school teachers might want to show your video but guess what they can’t because a guest needs to drop f-bombs for the sake of dropping f-bombs.) Anyway…

Veo3, as a generative AI video tool from Google, is a rapidly evolving technology with implications for content creation, ownership, and usage. Here’s a summary of the information gathered:

Fair Use and Resale of Generated Clips & Ownership:

The ownership and copyright of AI-generated content, including that created with Veo3, is a complex and highly debated topic. There isn’t a definitive, universally accepted legal framework yet, and current interpretations can vary by jurisdiction.

  • Current Legal Landscape (US): The U.S. Copyright Office has generally stated that works created solely by AI, without sufficient human authorship, are not copyrightable. This means if you simply input a prompt into Veo3 and it generates a video, the output itself might not be eligible for copyright protection. However, there’s an ongoing discussion about what constitutes “enough” human intervention to grant copyright. Some argue that creative prompting, editing, and integration of AI-generated content into a larger human-conceived work could establish copyright.
  • Google’s Stance (Implied): While Google’s official terms of service for Veo3 specifically regarding ownership aren’t immediately clear in the search results, the general sentiment in discussions is that AI systems are designed to produce “transformative works” to avoid direct copyright infringement of their training data. If you’re looking to replicate your own existing copyrighted works or use specific copyrighted elements (like a famous character), you would need to contact Google directly for potential contractual agreements, which are not generally available to the public.
  • Resale and Fair Use: Given the ambiguity around copyright, the resale and fair use of Veo3 generated clips are also in a gray area. If the output isn’t copyrightable by you, your ability to legally control its resale or claim fair use for others’ usage of it would be limited.
  • Recommendation: For professional use, especially for TV commercials, it is highly recommended to consult with a legal expert specializing in intellectual property and AI to understand the latest interpretations and ensure compliance. Relying solely on AI-generated content for commercial use without clear ownership could expose you to risks.

Trial Period and Limitations:

Access to Veo3 is primarily granted through Google’s paid AI plans. As of recently, it has become more accessible:

  • Google AI Ultra Plan: This is the highest tier of access, currently priced at $249.99 per month. It may offer a discounted rate for the first three months (e.g., $125/month). This plan provides the highest limits and exclusive access to Veo3.
  • Google AI Pro Plan: Veo3 access has been expanded to this lower tier, priced at $20 per month. Subscribers to the Pro plan get “limited access” to Veo3 in Flow and a “10-pack of trial video generations” to test it out in the Gemini app. This suggests a limited trial within the Pro plan.
  • Student Advantage: Eligible students (18+) may get Google AI Pro for free for an extended period (e.g., 15 months) by signing up by June 30, 2025, and verifying enrollment. This effectively provides a free trial with limited Veo3 access.
  • API Access (for developers): Accessing Veo3 through the Vertex AI API (Google Cloud’s enterprise AI platform) might offer a “free trial period, limited only by your credit allocation and the platform’s usage tiers.” This requires more technical knowledge.

Trial Limitations (based on available information):

  • Limited Generations: The Google AI Pro plan explicitly mentions a “10-pack of trial video generations.”
  • Resolution and Duration (API): The API currently supports 16:9 aspect ratio, 720p resolution, 24 FPS, with an 8-second maximum duration per clip. It’s likely similar limitations apply to direct usage.
  • Prompt Interpretation & Audio Issues: Even in paid versions, users have reported “hit-or-miss” prompt interpretation and occasional issues with audio generation.

Budget Cost and Cost Breakdown for TV Commercials:

Based on date of this article, 6/8/2025, Veo3’s pricing is primarily subscription-based, with costs varying based on the plan you choose and your usage.

  • General Budget Cost:
    • Google AI Pro: $20 per month (limited Veo3 access).
    • Google AI Ultra: $249.99 per month (or $125/month for the first 3 months) for highest access.
  • Cost Breakdown for 30-second TV Commercials (5-15 scenes):
    • Per-Second Cost (Estimated): One user on Reddit mentioned Google Veo 2 pricing at $0.50 per second. While this is for Veo 2, it gives a potential idea of a per-second cost model if that applies to Veo3 credits. Another discussion mentioned that $250 for Ultra might yield “up to 83 Veo 3 videos a month,” which suggests a credit-based system where different generations consume credits.
    • Commercial Scenario: For a 30-second commercial potentially requiring 5-15 scenes, each scene would likely be a short clip (e.g., 2-6 seconds per scene).
      • If each scene is, for instance, an 8-second generation (the maximum current output for the API), and you need 15 scenes, that’s 15 generations.
      • Assuming a credit-based system within the monthly subscription, your primary cost would be the monthly subscription fee (e.g., $249.99 for Ultra) that grants you a certain number of credits or generations.
      • The actual cost per commercial would then depend on how many generations are needed to get the desired clips (accounting for re-renders due to prompt interpretation issues), and whether the allocated monthly generations/credits are sufficient.
      • One user claimed to make an ad that would typically cost $50K-$150K in LA, for “3 hours and $7.25 in AI credits,” implying a per-credit cost, but this was a specific instance and likely part of a beta or test.

Summary of Costs for TV Commercials:

  • Your primary cost will be the monthly subscription to a Google AI plan (Pro or Ultra).
  • The number of usable generated clips you can create for that fee will depend on your plan’s limits (number of generations/credits) and the efficiency of your prompting (how many attempts it takes to get a good result).
  • While the technology can dramatically reduce traditional production costs (actors, locations, crew), the exact “cost per commercial” with Veo3 is best thought of as a portion of your monthly subscription plus the human effort involved in prompting, refining, and potentially editing these AI-generated clips into a cohesive commercial.
  • Currently, the cost breakdown is more about the subscription tier you choose rather than a direct per-scene or per-second charge for the final commercial, although internal credit consumption might exist.


Unsure About if You Can Publish Responses From AI? Ask the AI!

June 8, 2025 Aaron Belchamber No Comments

Prompt:

When I ask Gemini a question, am I allowed to publish the answer on my website within an article if I cite Gemini as the source? What am I allowed to publish from Gemini and what are the restrictions?

Gemini Response

It’s great that you’re thinking about proper attribution and usage when publishing content generated by Gemini!

Here’s a general guideline on what you can publish and the restrictions, keeping in mind that specific terms can sometimes vary or evolve. For the most precise and up-to-date information, you should always refer to the official Google Terms of Service and any specific terms that might apply to Gemini or generative AI tools.

General Understanding and Best Practices:

What you are generally allowed to publish (with proper attribution):

  • Direct Answers to Your Questions: You can typically publish the factual information, creative content (like poems, stories, code snippets, etc.), or summaries that Gemini provides in response to your prompts.
  • Originality of Output (Generally): The output generated by Gemini is generally considered unique content created in response to your specific prompt.
  • Attribution: It is a best practice and often a requirement to cite Gemini (or “Google’s Gemini AI”) as the source. This is crucial for transparency with your audience and acknowledges the tool used in content creation.1 A simple line like “Generated by Google’s Gemini AI” or “Answer provided by Gemini AI” is usually sufficient.

What are the restrictions and important considerations?

  1. Accuracy and Verification:
  • Crucial Responsibility: While Gemini aims to be accurate, you are solely responsible for verifying the accuracy of any information before publishing it. AI models can sometimes generate incorrect, misleading, or outdated information.2 Do not publish anything without fact-checking.
  • No Guarantee of Accuracy: Google’s terms will generally state that they do not guarantee the accuracy or completeness of the output.
  1. Harmful, Illegal, or Unethical Content:
  • Prohibition: You are strictly prohibited from publishing any content generated by Gemini that is illegal, harmful, hateful, discriminatory, infringing, or promotes violence or self-harm.
  • Compliance with Laws: Your use and publication of Gemini’s output must comply with all applicable laws and regulations.3
  1. Sensitive Information:
  • Do Not Input Private Data: Avoid inputting any personally identifiable information (PII), confidential data, or sensitive company secrets into Gemini.4
  • Do Not Publish Private Data: Never publish any personal or confidential information you might have inadvertently or deliberately included in a prompt or that Gemini might generate.5
  1. Copyright and Intellectual Property (Evolving Area):
  • AI and Copyright is Complex: The legal landscape around AI-generated content and copyright is still evolving. While the output is generated for you, the extent to which you can claim full copyright ownership over purely AI-generated content (without significant human creative input/editing) is still debated in some jurisdictions (e.g., the U.S. Copyright Office has indicated that works generated solely by AI are not copyrightable).
  • Your Responsibility: You are responsible for ensuring that the content you publish, even if generated by AI, does not infringe on existing copyrights or intellectual property rights of others.
  • Plagiarism: Avoid presenting AI-generated content as your own original thought or research without proper disclosure.6
  1. Misrepresentation of AI:
  • Transparency: Do not misrepresent the content as solely human-generated if AI played a significant role. Be transparent about its origin.
  • No Impersonation: Do not use Gemini to impersonate individuals or entities.
  1. Scalability and Automated Use:
  • Refer to the terms of service for any limitations on automated or high-volume use of the API, if you’re using it in that manner.

In summary, yes, you can generally publish answers from Gemini on your website within an article, provided you:

  • Cite Gemini as the source.
  • Verify the accuracy of the information.
  • Ensure the content is lawful, ethical, and appropriate.
  • Understand the evolving nature of copyright for AI-generated material.

Always refer to Google’s official terms of service for the most definitive and up-to-date guidance.


SheepIt! A Free Blender Render Farm Service

March 30, 2025 Aaron Belchamber Leave a comment
SheepIt

As long as you’re okay if your Blender project is potentially shared with hundreds of other users on the internet, SheepIt is a great way to get your Blender renders completed quickly and for free.

I used it recently for a rush animation job.  The animation was for the debut of a live streaming broadcast.  Unfortunately I had to make some last-minute changes and I was running out of time fast!

My Blender render was taking a lot longer than I was hoping.  It was a 30 second show intro in HD 1920×1080 at 24 FPS.  That’s 720 frames, which usually wouldn’t be so bad. The scene had a lot of reflections and motion blur and I tried every trick that came to mind to lower the render time.  I lowered my render passes from 256 to 16 without much noticeable artifacting but still could not get the average frame render time below 5 minutes.

With the clock ticking I started my animation render, even using Flamenco with 3 machines on my local render farm, I was averaging about two minutes per frame.  720 frames at 2 minutes would take… A full 24 hours!

Yikes!  I had less than 3 hours before I had to deliver the animation and in that time I had to lay music and a transition over it.  Luckily, I had run up millions of credits on SheepIt so instead of lending my machine to render other peoples’ Blender projects, it was my turn!

I logged into SheepIt! and followed the simple instructions to submit a project and was happy to see within 15 minutes some monster processing power was being dedicated to render my project!  Within less than an hour, I got my PNG sequence fully rendered and ready for download.  SheepIt gave me so much time back I was no longer in “rush mode” at all.

If you expect you might need to leverage a render farm and you’re okay with the public openness about it, I suggest starting a free SheepIt account and lend your computer to build up some points that you can later redeem for your own projects later.  I’m certainly glad I was curious about SheepIt and was pleasantly surprised how things turned out.

If you’re not ready to render anything you can start banking credits on SheepIt by joining.  You will then download their client and it will take take of the rest. 

Though Blender rendering can be resource intensive, you can even restrict the use of your computer’s CPUs and memory in case you still want to be able to use your computer.  It’s also easy to pause or exit out of.