Category: 05

Sneeze—VQGAN+CLIP

Prompts
“magical giant village in tree tops” and “medieval cats fighting”

Using the Google Colab notebook was interesting. I have used Google Colab notebook before but only briefly and unsuccessfully for another project. I just remember being very confused about the concept of running pre-made code from others and not understanding that all you had to do was press the play button on the code snippets. It was very easy to complete and the most difficult part was waiting for the images to render. I thought it was cool that you could either give it a base image to work off of or just have it generate everything from scratch. Both of my prompts were generated from scratch by the colab notebook.

Sneeze—TextSynthesis

Narrative Device

“The old woman in the store was giving out fortunes when she caught sight of me. “You’re going to be rich,” she said. “You’ll find a lot of money.” But I didn’t believe her. I only had a few dollars. “You’ll get a headache,” she said again. I was too tired to argue. So I bought a fortune cookie and went to bed.”

Narrative Device takes in two themes and generates a whole paragraph out of them. I thought that this was really easy to use since it only needed the user to come up with two subjects, but sometimes the outcome would exclude one of them. When I used headache and fortune cookie, the device would only sometimes give me a paragraph that was related to headaches. Besides that, generating text on the same themes creates remarkably different storylines, which is impressive. I think that this device creates a semi-logical plot with a much more cohesive story compared to InferKit.

InferKit Tool

“Upon opening the door, Dorthy found that she was not in her room. She had transported to the land of ice and was about to slip and fall.
Determined to walk on, she insisted that she could handle it. The Mice that were near her allowed her to use their icy paths to reach her destination. They were also helpful in avoiding new dangers of their own.”

The InferKit Tool has a different dynamic from the Narrative Device. It requires the user to give it a starting phrase or sentence or sentences to build the story. I think that this gives the user more control, which can be nice, but providing fewer words results in a random story. When I experimented with giving it a phrase, the story went to a different subject all together. Providing more words makes the story a lot more cohesive, but then that requires the user to work more.

 

Solar-TextSynthesis

Blobfish, Diner: For the last few weeks, the blobfish have been congregating at the diner. They sit in a circle around the fryer, cooking their food. Some of them stare at the fryer with envy, while others watch them with skepticism. But everyone seems to be contentedly slurping their food.

There was a parasite in the rainforest, that looked like it was made out of a bunch of humans. The forest had believed that the parasites were killing the trees, but it turned out that the trees had been going crazy, having some strange dreams. Their dreams were filled with skulls and bones and other things that shouldn’t exist. That was probably why the forest had sent the giant trees to find these humans.

Reflection: It was very humorous to experiment with random words and themes. The stories that are generated often go against my expectations and play around with different texts to create imaginative and surprising stories.

Sneeze—ArtBreeder

Using ArtBreeder was quite fun! I think that at the start, I was confused on how the tool worked since there were not instructions and the UI is not intuitive. Eventually, I understood how it worked and from there I was able to play around with the parents and genes. I enjoyed how the user could alter the weight of a certain trait and also have a chaos gene which adds in a bunch of random traits to create really interesting images.

Sneeze—PixtoPix

While using this tool, I had fun trying to come up with cute or wacky designs of cats, shoes, or buildings. However, the outcomes were not as satisfying for me. This was especially apparent in the edges2cats tool because the AI was misinterpreting certain aspects of my drawings since they were not standard to cats. For instance, my super large eyes cat has some fur in his eyes, which was not ideal, and my toothy cat doesn’t have as much personality as he was intended to. Although there are limitations to these tools, its very impressive that the AI is able to detect features from drawings, but I think that there are some issues with drawings that have more animated or cartoony features.