
METHODOLOGY
I took a few artistic liberties with this piece -
1. DALL-E takes about 10 seconds, and generates 10 images. For pacing & storytelling purposes, I picked one image and showed it instantly rather than including the 10 second wait time
2. When an existing image was modified (e.g. when the birds and sun were added in the beginning) I did this by first generating the art, then uploading it back into DALL-E with the "edit" tool, erasing portions of the image, and letting it generate over the erased portions
3. I also added specific phrases (not pictured in the video) in order to generate the image I wanted. For example, in the first scene, I asked for "a woman standing in a field on an overcast day, oil painting" but in the video I simplified it to "a woman, oil painting" for storytelling purposes
4. In some cases, but not all cases, when you request images of a human, DALL-E will generate diverse options of people with different ethnicities. When it didn't do this by default, I refined the search to add specific terms in order to get the diversity I wanted for the piece. For example, I simplified prompts from "asian woman, oil painting" to "woman, oil painting" in the video for storytelling purposes
Credits --
Director: Karen X Cheng
AI: instagram.com/openaidalle
Art account:
Song "Feeling Good" by Nina Simone (suggested by Omar Beuchat)
Sound FX David Bass
#artificialintelligence #ai #creative #video
LOW BUDGET FILMMAKING ➡️
Follow me on IG for more videos
IG:
WHO AM I
Hey, I'm Karen X Cheng.
I’m a creative director and I love sharing my creative ideas and experiments!