Exploring Sora by OpenAI: Promising, but Not Quite There Yet (Watch the Videos)

I recently got the chance to try Sora by OpenAI, and it left me with mixed feelings. There’s potential, but also plenty of room for improvement.

Here’s the deal: Sora works well with short, specific prompts. Give it a simple idea, and it delivers decent results. But if you throw in a detailed prompt filled with specifics and variables, the output doesn’t quite hold up.

One feature I was eager to test was attaching an image for Sora to use as a base for generating a video. While the feature exists, it’s far from polished. The results often miss the mark, even when the instructions seem straightforward.

After exploring the work of others and my own creations, here’s what I’ve observed:

• The best-generated videos are short (5 seconds) and focus on one main element with a complementary background.
• Asking for multiple elements, characters, or complex details doesn’t yield good results.
• Using an image as a starting point for Sora to generate a video doesn’t consistently produce desirable outputs.

In its current state, Sora reminds me of the early versions of MidJourney. The potential is clear, but it hasn’t reached its peak yet.

One surprising strength, though, is text generation. When I explicitly asked Sora to include text in a video—like adding my company’s name in the background—it handled it well. A small win, but a win nonetheless.

Plans and Features: Sora is available in both paid plans, with the Pro subscription offering longer videos, higher resolutions, and an increased cap on video generations. It supports three aspect ratios: 9:16, 16:9, and 1:1, which makes it versatile for various platforms.

Ideal Use Case: Right now, I think Sora shines in creating short b-rolls to support other content. A quick, 5-second clip with a clean focus can enhance your videos in creative ways.

Sample Video: I tested Sora by prompting it to create a video of a submerged Dubai city underwater, complete with sharks and turtles. The result? A 5-second clip at 480p resolution. Not bad, but definitely not mind-blowing.

Transparency Matters: If you share a Sora-generated video on LinkedIn, it’ll carry a C2PA (Content Credentials) badge. This badge lets viewers know where the video was created, whether AI was involved, and sometimes even the app or device used. It’s a move toward transparency, and I think it’s a step in the right direction.

Final Thoughts: Sora isn’t a game-changer yet, but it has its use cases. If you’re looking to add a bit of creativity to your content or need quick, simple clips for b-rolls, it’s worth experimenting with.

Have you tried Sora yet? What do you think? Let’s discuss!