Final Project – AI-generated Storyboard

Below is the process and results of using AI software Boords (https://boords.com/) to create storyboards for scenes from scripts I wrote in the past.

In total, there are three different sets of storyboards from three different scripts, separated by line segments.


Short Script 1:

Customized Character: Julian

Storyboard Series 1:

No matter how hard I try, I cannot generate the image of “squatting and petting the dog”. No matter what prompt I type, like “squatting to touch the dog”, “touches a stray dog”, “kneels down while touching a dog”, the generator refuses to show the dog. On top of that, a big car comes out of nowhere. The picture you see right now is generated by the prompt below:

Short Script 2:

Customized Character 2: Alex

This is where I found out if you customized many characters, you can only use one of them consistently. It does not work when you want to include two customized characters in one scene.

Storyboard Series 2:

This script fails at the beginning, because I cannot get the generator to put two different people in the scene. I fixed the prompt a hundred times. It just won’t put another girl into the picture. Below is the prompt for this scene.
However, when it comes to this scene, it suddenly lets me add another person beside Alex, but what’s up with the black areas… AI tends to assume Amy is Asian because Alex is Asian.
WTF. Why is there another guy? Check out the prompt below…

Now, introducing the most successful (in comparison to the first two) script-to-storyboard experiment:

Short Script 3:

Storyboard Series 3:

I don’t think any AI understands film language very well yet. If a human draws this, he will be able to frame it exactly as he or she wants to.
As you see, I cannot get the dead rabbit, and that interferes with my creative process, just like cencorship in films limits filmmakers’ creative process and artistic expression.
We can tell that the sky is significantly different, although I used the same description. The wooden cabin also looks different.
Frame 19’s action should be: The boy knocks on the door of the cabin. For some reason the generator refuses to portray anything like that. To remain consistent, I had to type in some former prompts. However, if I make any slight changes or delete anything from the first two sentences, the image turns to a completely different setting with a completely different boy at a completely different place. With the current prompt below, I only get a wooden cabin in the forrest, with no boy. Below is the prompt for frame 19.
If a filmmaker looks at frame 25, they will have no idea what is going on. What actually should be going on is typed in the invalid prompt below:
Can somebody tell me what is inappropriate about this prompt? As a result of this, I also failed to generate the boy standing behind a closed door as the script goes.
You cannot get this thing to generate a younger boy smoking. It automatically changed my character to a teenager. Also, what’s up with the cig? What’s up with the lighter? What’s in front of the window? Are those human hands?

Some of the images shown up there are ones I fixed more than 30 times.

It’s mentally challenging.

But what can I say. I don’t know how to draw.

My fault.

Anyways, here’s the essay.


Comments

Leave a Reply

Your email address will not be published. Required fields are marked *