Behind “tired eyes”

Throughout the years, one of my favorite things to do has been listening to music and disappearing into whatever scene the song decides to put in my head.

Sometimes it is very clear. A character walking through a forest. A camera floating behind someone. A strange hand appearing in the dark. Sometimes it is only a mood, or a color, or the feeling that there should be a scene there, even if I don’t know what the scene is yet. I have had this private little cinema running in my head for most of my life, and for most of my life I assumed that was where most of it would stay.

I design, draw and make interactive stuff for a living. Making a film, even a tiny one, always seemed like something that belonged to another version of me in another timeline where I have more time, more money, more patience, more people around, more technical skill. I figured maybe someday I would get to make some of these things real, probably much later in life, maybe when I was older and had earned the right to take sabbaticals to make beautiful useless things for no particular reason, but not at 35.

So the weirdest part of working with Flick was how quickly that old assumption started falling apart.

Integrating Flick into the pipeline

At first I was mostly testing. I made images, tried shots, changed prompts, looked for characters, threw things away. The interface felt almost suspiciously simple, which I liked, but it also made me wonder how far I could really take the project inside it. Then, after a while, the scattered tests started behaving like parts of a map. One image suggested another. A failed frame gave me a better idea for a scene. A character that looked wrong in one generation suddenly had the right feeling in another. I started by making my own hand-drawn storyboard, but then I realized that if I messed about in Flick, the story began to show its edges in ways I hadn't even thought of.

The chat was part of that, although I didn’t use it much in the beginning. I treated it like a help window. Later, once I realized it could respond to what was happening on the canvas, it became more useful. I could describe a feeling badly, point it toward the image, and get help turning that vague feeling into something the video model understood better. That was big. A lot of the work was still me choosing, rejecting, fixing, and trying again, but the chat helped me stay inside the project when I would normally get stuck translating a visual idea into instructions.

One of the most useful things I learned had almost nothing to do with prompting. If a generated image was close, I got much better results by saving to my computer and fixing it myself.

A hand would be wrong. A face would have the right expression but the wrong shape. A background would be beautiful except for one strange object floating where nothing should be. Sometimes the composition was exactly what I wanted, but the image had some broken little area that I knew would become worse once it started moving. So I would open it in Photoshop and do the boring old work to get it to where I wanted.

The character creation part was fascinating. It was so fun and mindblowing to turn a vague silhouette into a character I genuinely identify with. This moment might have been the most powerful to me, as it opened my eyes to the possibilities of world-building.

I think a lot of people imagine AI work as sitting in front of a prompt box until the machine gives you a finished thing. Maybe sometimes it works that way. My experience was messier and more familiar. It felt closer to making a collage, or retouching a photo, or building a comp for a client, except the comp could suddenly move if I prepared it well enough.

The rough sketches helped in the same way. They did not have to be good drawings. Some of them were only useful because they told the image where things should go: the person here, the hand there, the camera low, the body small, the light coming from behind. That was enough. A bad drawing with a clear composition could be more useful than a polished prompt with no visual anchor.

A few notes I’d share with a friend

Start with something that already has a little charge in it. A sketch you keep coming back to. A character you have drawn five different ways. A photo that feels like it wants to be a film still. A scene that arrived while listening to a song. The tool behaves better when you are not asking it to invent your taste for you.

Do not wait until the idea is fully formed. A rough drawing with a clear arrangement can be enough. The person goes here. The hand comes from there. The camera is low. The background is dark. The light is behind them. That kind of image can do more than a long prompt that sounds beautiful but gives the model nothing solid to hold.

When in doubt, ask Flick. When in doubt about who to achieve a specific still image, video animation or even about your story, go ask the Flick chat. You’ll be surprised.

Keep the almost-right images. Those are the useful ones. The perfect image almost never arrives all at once, and the bad image is usually just bad. But the almost-right image gives you something to work with. You can repair it, crop it, paint over it, use it as a start frame, or let it teach you what the scene actually wants.

When the image is close, stop prompting for a while. Take it into Photoshop, Procreate, Affinity, whatever you use. Fix the part that is bothering you. Remove the weird object. Clean the hand. Make the silhouette readable. Move the body a little. Darken the corner. Then bring it back. A few minutes of old-fashioned image work can save you from an hour of asking the machine to guess better.

Pay attention to the start and end frames. Small differences can change the whole animation. A head tilted slightly differently, a cleaner arm shape, a calmer expression, a less confusing background. These little things can decide whether the movement feels alive or broken. The image is doing more instruction than it looks like it is doing.

Save the happy accidents. Most weird outputs are disposable. Some are not. Sometimes the model misunderstands you and gives you a better doorway into the scene than the one you planned. You still have to decide what belongs, but that is part of the fun.

Is AI making me lazy?

The process did not make me feel detached from the work. It kept sending me back into the work. Back into drawing, fixing, cropping, comparing, choosing, throwing things away, keeping the strange frame that somehow had more life than the correct one. I expected the AI part to feel like the most futuristic part of the process, but a lot of the time the real progress came from very “classic” decisions made by eye, judgment and intuition.

This is also why I think some creative people might be surprised if they actually spend time with these tools instead of only arguing about them from a distance.

There is a version of the AI conversation where the artist disappears and the machine just makes the thing. I understand why people are angry at that version. I do not want that version either. But the actual experience I had was much more involved, fun and more interesting than that.

I still had to make a lot of decisions. I still had to know what looked wrong. I still had to rescue the good parts from the bad ones. I still had to bring the image into another program and fix it with my hands. I still had to choose the frame, shape the scene, and decide when something had the right feeling.

So when artsy people dismiss AI as if it is only a lazy button, I get the suspicion, but I also think they may be underestimating how much work still happens around the button.

Maybe the work changes shape. Maybe some parts get faster. Maybe some parts get weirder. Maybe the craft moves into places that are harder to describe from the outside: preparing the image, choosing the seed, repairing the output, noticing the accident, building continuity between shots, keeping a world consistent while the machine keeps trying to dream sideways.

That does not sound like less creative work to me.

It sounds like a different kind of lifting.

And for me, at least, that lifting opened something. The scenes I used to leave floating inside my had suddenly had somewhere to go. The rough drawings had another life after the page. The idea of making “useless” beautiful things does not feel so far away anymore.

That is what I mean by the floodgates opening. Not that everything became easy.

Just that, for the first time, a lot of things felt possible enough to start.