Sora vs. Midjourney — The AI Struggle for Dominance
Beyond unbelievable, what bothered me a bit about the OpenAI’s Sora is that the quality of the video images is impressively superior to those generated by DALL-E.
This leads us to believe that OpenAI is already able to match the quality of MidJourney.
Lets dissect the strengths, weaknesses, and nuances of TensorFlow and PyTorch, and witness the clash of algorithms unfold in this epic AI faceoff.
For images, OpenAI is in catch up mode, maybe Dale-4 will get closer to MJ v6 but by that time we may see v7, but competition is also good.
David often talks about Midjourney becoming a world engine in his podcast. For those who don’t know him, he is the computer scientist and the founder of Midjouney. Imo, both in terms of company vision and quality, Midjourney is the only real competitor to Sora.
On one side, Sora’s utilization of TensorFlow stands as a testament to its technical prowess. The framework, with its robust ecosystem and extensive documentation, empowers developers to craft sophisticated models with relative ease. However, as we delve deeper, we must acknowledge its shortcomings, particularly in image generation, where it appears to lag behind Midjourney.
Here is a twitter thread that shows all the prompts with videos alongside comparisions using Sora and Midjourney:
Meanwhile, Midjourney, under the visionary leadership of David, continues to push the boundaries of AI innovation. With discussions on the company’s podcast hinting at grand aspirations, Midjourney’s vision of becoming a “world engine” sets it apart. David’s insights into the future of AI, coupled with Midjourney’s advancements in image generation, paint a compelling picture of a formidable competitor to Sora.
What looks revolutionary is Sora’s recent videos is the background, it’s the way the subject and the background are equally evolving, the people walk through streets with other people in the background walking through the streets, in the cat video it’s the cat walking while the foliage moves away from the camera… I thought this tech was about slightly moving the subject a little while the background remained the same so it was like an animated gif, or a looping gif with rain in the background you know the ones, instead Sora looks like it’s making entire scenes rather than wavy images.
Despite this, Midjourney maintains a considerable lead of 12–18 months in this particular development, with prospects only set to improve moving forward. The company’s forward-looking vision sets it apart from its competitors, a perspective perhaps gleaned from their podcast discussions.
Ultimately, in this clash of algorithms, the question remains: who will emerge victorious? Will it be the established giant, Sora, or the ambitious upstart, Midjourney? Only time will reveal the answer as the AI landscape continues to evolve, shaped by the relentless pursuit of excellence and the quest for technological supremacy.
Thanks so much for reading, Feedback is much appreciated.
I’ll see you in the next one ;)