itp oddities

PERSONAL REFLECTION:

I am personally very conflicted about generative AI. On one hand, I can see some benefits of it as a tool and I think it could be really interesting to train models on your own work or use it as a way to storyboard larger concepts. On the other hand, I really don’t think the benefits outweigh the negative aspects such as stealing ideas and all of the economic and environmental impacts.

I also don’t believe that tools touted for productivity are best and think that oftentimes we end up driving ourselves too hard and fast. Why must everything hinge on its ability to provide us with a greater level of speed and productivity? I find myself often at odds with this idea personally and tend to take on too much when in reality it is often times better to slow down and do less better.

⋆｡ﾟ☁︎｡⋆｡ ﾟ☾ ﾟ｡⋆

ASSIGNMENT // EXPERIMENTS WITH AI:

The prompt: Create a short video entirely created by AI-generated stills. This story should include sound, voice-over, or narration and be visually cohesive. How do you tell a story using generative media? What story do you want to tell and why?

This assignment is about experimentation, not about doing this correctly. It’s alright to go down a path that doesn’t work as long as you show documentation and explain the intent behind your concept and the thinking that brought you there.

In your post, include:

• In your post, include:
• Documentation of the project
• What is the concept?
• How did you go about creating it?
• What was successful and what was unsuccessful?
• How would you approach the project differently knowing what you know now?

⋆｡ﾟ☁︎｡⋆｡ ﾟ☾ ﾟ｡⋆

When thinking about the use of AI, one thing that comes to mind in how it can be useful is achieving things that may not have been possible before or much harder.

I decided to do two experiments since I am pretty new to using generative AI. First I would try different methods of animating the weird characters found in medieval manuscripts. Then I would put what I learned to use and create an actual video by creating a strange dreamscape comprised of creatures and dreamy landscapes with a bit of a psychedelic feel.

.⋅˚☾₊‧ ☁︎ ‧₊˚ ⋅ MEDIEVAL AI .⋅˚₊‧ ☁︎ ‧₊☾˚ ⋅.

I thought it would be fun to see if it could animate and make weird interactions with the strange characters found in medieval illuminated manuscripts. I felt that the absurd imagery might lend itself well to AI and was curious what I could get Runway to produce.

Here are the images I started with:

THE PROCESS AND RESULTS:

The first image I tried was the one of the strange person-boat thing. The results were interesting and somewhat mixed.

⋆｡ﾟ☁︎｡⋆｡ ﾟ☾ ﾟ｡⋆

Next, I went on to try and animate a cat in a crown, also with strange and mixed results. In this one I had ChatGPT help me come up with a more solid prompt for a couple. This wording helped me realize other things I could be including in my prompts to make them more effective and I actually revisited the boat image to see if I could include some of that language to make those outputs more successful.

⋆｡ﾟ☁︎｡⋆｡ ﾟ☾ ﾟ｡⋆

I tried to then animate these dueling animals which yielded very different results in that some of the animations were fairly subtle and others completely outlandish. In this version, I used RunwayAI’s tool that helps you isolate areas for animation based on different areas you color in which seemed to help a bit.I also tested between simple prompts for different models of the ai to see the differences that would yield creatively.

⋆｡ﾟ☁︎｡⋆｡ ﾟ☾ ﾟ｡⋆

Then I tried animating the monkey. For this one, I tried to animate the flowers as well to make them grow but that saw limited success despite using the animation painter which I think would help if I did a separate deep dive on dialing in the settings a bit more and exploring what the tool is actually capable of.

⋆｡ﾟ☁︎｡⋆｡ ﾟ☾ ﾟ｡⋆

Lastly, I tried out one last test on the nose flute and bird leg guys. I was curious to see if I could get the two figures to have more realistic movements since they had more full human-like bodies but also found that this was a pretty difficult task, at least more difficult than I anticipated.

⋆｡ﾟ☁︎｡⋆｡ ﾟ☾ ﾟ｡⋆

OVERALL FINDINGS:

It was HARD to get something close to what I was imagining and it did help to have ChatGPT translate my thoughts to computer speak. I had a lot of fun just changing things around slightly and seeing how it would impact the output of the animations. It was also interesting to throw in changes outside of the prompt like using the animation brush or different models of Runway AI to see how it impacted the output. There are a lot of settings I didn’t delve into as much because I was really trying to explore more of the core tools and explore those but they would probably help a lot in terms of more precise generation. I actually think I might make a whole project of this and explore this idea further and dive into animating images I find via Creative Commons.

.⋅˚☾₊‧ ☁︎ ‧₊˚ ⋅ DREAMSCAPE AI .⋅˚₊‧ ☁︎ ‧₊☾˚ ⋅.

Next, I decided to try something very different and do a dream sequence video. I felt like I could get a fairly cohesive story playing into the off-putting vibes AI-generated content can sometimes give.

I was glad I did the little mini-experiment with the medieval illuminated manuscript images because it helped me be able to edit the images for the video a little bit better.

I used a lot of specific prompting about how I wanted things to animate as well as sourced images that were similar to what I wanted to create a timeline of the dream sequence. I also brought some things into Canva/Photoshop to make slight adjustments on the background in order to try and make it fit slightly better with the original theme which was one of the hardest parts.

In the end I am fairly happy with how it turned out as a somewhat creepy psychedelic dreamscape that is loosely based around some dreams I have had before.

OVERALL FINDINGS:

I think if I had more time I would be able to create something a little more cohesive. It was hard to get everything to work together well, especially when I was sourcing outside imagery that all had different feels and compositions. I am not sure if I liked the process of creating an entire video due to this and I enjoyed the smaller animations of manipulating individual things because the cohesion wasn’t as much of an issue. On the other hand, it was much easier giving the AI more realistic looking references and I felt like I could get much more realistic and closer outputs to what I was going for than the illuminated manuscript drawings.

⋆｡ﾟ☁︎｡⋆｡ ﾟ☾ ﾟ｡⋆