Ever stumbled upon a picture online and wished you could reimagine it through the eyes of DALL-E or Midjourney, but got stuck on the prompt part?
Well, that’s where my latest creation, the Image Recreator GPT, jumps in to save the day! It’s a nifty tool I built using ChatGPT, and here’s the lowdown on its cool features:
- Simply upload a picture, and voila! You get three variations.
- Each generated image comes with its own ID. This makes tweaking them later a breeze.
- You’ll also get a Midjourney prompt for each image.
- Got an image with copyrighted characters? No problem, you can upload those too!
Check out the interface here:
Let me walk you through how this works. I tried it with a Pixar-style animation showing a little boy and a dragon enjoying ice cream on a park bench. After uploading and submitting, Image Recreator whipped up three images, each with its own identifier and Midjourney prompt, like so:
The three images? One in 3D cartoon style, another in 3D Pixar style, and the last in 3D Disney style. The Pixar one was spot-on, though the boy and dragon’s poses were slightly off from the original. No biggie—I just tweaked it with this prompt:
Prompt: Image x2 is good. However, it would be better if the boy’s gaze were directed towards the little dragon rather than the camera while hold the ice cream. Additionally, the little dragon should have its tongue sticking out while gazing at the ice cream.
You can also use the generated
gen_id “Kih3zgDbZlpdIMRl” for specific tweaks. Remember, the more detailed your prompt, the better the adjustment.
And yes, Image Recreator can navigate copyright waters. For instance, check out what it did with this Spider-Man image: Feeling peckish? Enjoy these bonus fried chicken photos:
If you find Image Recreator as handy as I do, add it to your ChatGPT. You can find it here:
I’ve also developed another GPT that empowers DALL-E 3 to significantly enhance its capabilities: