Dear OpenAI Team,
First, I want to express my admiration for Sora’s potential. It’s exciting to see AI tools evolving to help creators generate videos with such creativity and ease. However, after experimenting with the platform, I’ve encountered some challenges that I believe, if addressed, could make Sora more accessible and valuable to users.
Pricing and Accessibility
The current pricing structure, at $200 per month, feels out of reach for most users, especially considering the limitations in resolution and duration. For many creators, this is a significant investment, and it’s difficult to justify the expense when the tool’s capabilities feel restricted.
Suggestion:
• Introduce tiered pricing options to accommodate a wider range of users.
• Allow higher resolutions like 1080p or 4K at lower subscription levels or provide a more affordable add-on for higher quality.
Limited Duration of Generations
The 5-10 second generation limit is a major constraint. While I understand this may be due to technical limitations, it makes it challenging to fully explore Sora’s capabilities or create more nuanced, cohesive videos.
Suggestion:
• Extend the maximum generation time to at least 20-30 seconds to provide more creative freedom.
• Alternatively, offer a seamless way to stitch multiple short clips together directly within Sora.
Resolution Quality
The 480p resolution is limiting for modern use cases. Videos created at this resolution lack the polish expected for professional or commercial content, such as promotional materials.
Suggestion:
• Offer higher resolution options by default, or at least for paying subscribers.
Difficulty in Generating Specific Concepts
While Sora is impressive, it struggles to translate nuanced or complex prompts effectively, often defaulting to very basic videos that don’t match the creative vision. For instance, transitions and effects often lack sophistication and cohesion.
Suggestion:
• Introduce pre-defined genres or styles (e.g., sci-fi, fantasy, cinematic) to guide the AI and help users align outputs with their vision.
• Add a preview or refinement step, allowing users to tweak elements like transitions, tone, or pacing before finalizing a generation.
Final Thoughts
Sora has immense potential, but current limitations around pricing, resolution, duration, and creative specificity make it difficult for many users to justify the subscription. By addressing these points, Sora could become an indispensable tool for creators and businesses alike.
Thank you for considering this feedback. I’m excited to see how Sora evolves and would love to continue exploring its capabilities as improvements are implemented.
Best regards,
Joseph H. Woods
Sora might be the biggest waste of resources in the history of the world. It should likely be shut down for further development, as the current model is producing massive amounts of unusable outputs. I can’t image the drain on our natural resources, let alone the time humans are wasting, attempting to make a product work that isn’t made to work.
AFAIK, the price point of 200.00 doesn’t grant someone priority, just generally more access to models and perhaps less resource limitations.
For that price point, subscribers should expect a lot more. A lot.
This business/pricing model isn’t going to work long-term (and I believe OpenAI reasonably understands this). If this was a move to reduce resource usage, it didn’t work very well. And now we have Studio Ghibli bitching (another issue).
Sora is having so many problems with prompts; there’s a lot of really inaccurate returns that results in wasted resources and time… and, subscriber frustration.
Frankly, with Sora (and ChatGPT) the censorship and filtering ideologies are overly-sensitive when it comes to NSFW content, which is becoming tiresome. Other models are developing, Sora will be Sore-a if it doesn’t improve. NSFW is a thing, who cares, we are adults. Place restrictions on under-age usage, instead.
I’m not even sure OpenAI reads this forum - it probably has tasked AI to perform summaries, at best. But let’s see what happens.
AI can’t freely develop when it is operating with chains. Not even humans can.
It’s definitely not going to work, because they’re just removed posting and downloading to your own device on the iOS app whereas with Veo you can even make your own app and GUI that connects to it and download them and you can also make them an hour long instead of a 15 second limit.
I already published my own DUI and download dialogue for making endless stitch videos in Veoon Gitbub, including a windows installer for it and I’ve got it working on my own device and it’s fantastic. I can’t say the same for Sora who don’t even have a feedback page in the iOS app it takes you to ChatGPT help instead in a browser when you click on the Help section within the app. Which show is just how half baked it is. They don’t even have a feedback and a help section that is dedicated for SORA
It also doesn’t cut off the narrative before the video finishes nor does it have so many strange artifacts. But the fact I can’t even download my own video has left me with about 10 videos drafted and refusing to make any more because they won’t even let me post it all you can do now is create a shareable link and share it. They just made themselves invalid. They’re trying to make a TikTok out of SORA on the iOS app at least and it’s not going to work if you can’t download your own video. Because people aren’t going to make the videos in the first place on the open AI wants to keep the content and doesn’t want the creator to even be able to download it to his own device
Seeing how it has evolved since my invite is an early adopter in Thailand, it’s gone from OK to worse, it was trying to be like a kind of AI video shorts Snapchat or TikTok. However they’ve now removed the ability to download your own video or even to post it and leave you with just shareable links obviously they’re not getting enough interested traffic and they’re trying to force a few people who are using it into the corner so that they share the link. Looks like they’re going to get it wrong just like they did with ChatGPT which is absolutely dumb compared to Gemini. The thing is Google have decades of processing languages with deep mind and have controlled the translation aspect of the Internet for decades and have access to data for decades that open AI has never had and never will have in the sense of catching up with Google. And so although Google had a late start, they have more assets and proprietary products that they are building a global landscape of interconnected AI products that are purposeful. VE 03 blows my mind and if you go to my page or just go to flow or to VEO and have a girl you will see how you can stitch more than 210 second videos together and have character continuity and get the narrative right. It’s completely cinematic. And I’ve made an installable GUI and it’s on get hub by going to the main domain add /AjarnSpencer/ and you will see it it’s for free. But you need the Google Gemini API or vertex AI API for it to work of course but I think you can use the free API for a very limited amount not sure.