Spent a few days testing out Imagine - @grok's new image and video generator. And I've got a good sense of where it spikes, what the limitations are, and how to use it. Curious about how it stacks up? Keep reading 👇

Once you have an image, it's time to animate. You can write your own prompt (click "custom"), or auto-generate a video that's fun, spicy, or normal. I love the auto-prompting - many people don't know how to prompt or want to see what the model does. It takes ~17 seconds to generate a video (I trimmed here so you can see "normal" vs. "fun").

@grok You can also animate a photo that you upload. Two examples here (one meme, one real pic). The results of this are quite good - these were both first try generations.

@grok There are some real strengths to this product: - Both image and video gen are crazy fast - You get unlimited generations for $30/mo - It's on a mobile app (this is HUGE for consumer) - So good at memes or simple animations

@grok However - don't expect it to be a video world model like Veo 3. The physics, consistency, and coherency aren't at the same level. And while the native audio generation is cool, it's often a bit off (e.g. the speech is gibberish or sounds aren't synced right).

@grok A couple of things to note: 1) If you like an image you make, HEART or download it. Otherwise it gets lost in the feed. To find the images you liked, look at the little icon to the left of search bar. 2) It seems to be pretty uncensored and will generate real people, e.g. ⬇️

@grok Another tip I forgot to mention here - use the voice dictation to prompt instead of typing. Makes the whole thing much faster (and it's fun to watch the words appear on the screen!)