Circuits of Identity

-

 


Circuits of Identity 5:15


This one was written almost entirely by Eskritor. The idea was formed by some recent articles about gender assignment in robotics in Japan and Korea. 

The challenge on this was creating an android that didn't look like everyone elses. Image prompting is limited to a language model and it's source material training. I've become more aware of the "styles" avalable and I'm beginning to learn how to morph my own.

The next challenge was setting. Creating an environment, a "set" is a tedious process of acquiring the needed perspectives and weaving them together. I feel sometimes I'm working on a 3 camera stage set sometimes. 

Creating the hand written note in the scene was another challenge that took a lot of effort. I was able to get GPT Image 1 to generate an image of the note that I layered into another image. I was unable to get any of the other image generators to do this. All the output had distorted and/or missing letters.

GPT Image 1 has been great for strict prompt adherence at times, but it's expensive... 12 to 60 times more so than the other image generators I use. Also I find GPT Image I lacking in creativity at times.



Comments

Post a Comment

Popular posts from this blog

Metamorphose

-
Image
  Metamorphose 4:40 This took 2 full days to make and cost almost 10 dollars using Replicates cloud servers. The story was generated with a paid subscription to Eskritor using the following prompt: Write a short story about luminescent creatures running in panic and fear on a field of lava. Blue foam bubbles from fissures. Green lush vegetation grows in isolated pockets. Inspiration for this story was my recent stay on a lava field in Kalapana, Big Hawai'i USA. Edited with two open source programs: Openshot video editor Audacity sound editor An equal amout of editing was needed for both image and sound. Music was generated and cut and pasted like the images. This is my first AI driven creation. Three days ago I had no user knowledge of AI image or sound creation. I also learned how to use both open source editors during the same time. I had a crash course in learning how to use these programs. I still have a lot more to learn. I plan to make and post a new video every week. I'l...
Read more

Drained

-
Image
  Drained 3:59 This film was written with Eskritor and my newest writing tool... Meta AI with Llama 4. Meta AI is absolutely free on WhatsApp. The interactive conversations I have working out story details has been extremely more efficent than the Eskitor app.  This story was inspired from my personal experiences with heat and mosquitos while living and sometimes trying to sleep in Thailand. I wanted to try building a story with no music, just sound effects. I also wanted to try to have more control of my setting and how my character behaved. I wanted to create an actor that had more personality and realism. To help with this, I created a special character using OpenArt AI's character creation tool. Most of the sound was created with the sound effect generator stackadoc/stable-audio-open 1.0, but a few analog sound effects from Freesound were incorporated. This particular film has given me more of an opportunity to use the sound editing program Audacity. It was instrumental in...
Read more

Desperate Ride

-
Image
  Desperate Ride 4:21 This one was written with assistance by Meta AI with Llama 4. It was inspired by my recent visit to the Tallgrass Prairie Reserve in Kansas, USA. I wanted to create a film that captures a broad sense of space and has a lot of moving elements too. This was the perfect story for this. The major challenge here was the gag that has the rider falling from horse and on to the back of a buffallo. It utilized failed prompts for images that were created with layered elements in GIMP and then edited to create the action. The next challenge was the hat gag. It also require tedious setups for prompting. This film also made me aware of Vozo's lipsyncing weaknesses. Characters with beards are problematic.  Vozo is regarded as one of the most accessible AI generators for lipsyncing, using their precision mode, this was the best I could generate. I decided to not even bother for the last 2 shots of dialogue. The music was beautifully generated by Meta/music gen. It requi...
Read more