If you find yourself here, please take a few seconds to clap this blog. Up to 20 claps allowed per person. When a new challenge theme is released, I find it can go many different directions for me: ~ I have several ideas and have a hard time deciding a direction. ~ I already have an image that fits the theme and I decide to just refine it. ~ I have an idea I'm happy with and just go with it. ~ I have no ideas or nothing I'm really happy with. The last case scenario is the subject of this blog and I'm going use the challenge 'Magic in Everyday Life' that is in the voting stage as I write this. In this case, my immediate first thought was to create a representation of 'me' relaxing as my daily chores were being performed around me. However, I thought this might be too popular of a theme, and I decided to try for something different (often the kiss of death in these challenges, I know). Given that I found myself voting that direction every time one of those images came up, I probably should have stuck with the idea. Instead, I turned to my preferred AI assistant, Chat GPT. I start by telling it what the theme is and ask for ideas. This first batch will either be completely rejected because I know they will be all too common, or they will spark an idea for something I do want to create. DISCLAIMER: Just because there were images with the below ideas in voting, does not mean the creator used AI for the idea. In addition, there is nothing 'wrong' with using AI to help with image ideas, after all, it's what this blog is about. This time I rejected the first batch: ~ The Laundromat Between Worlds (saw at least 3, maybe 4 in voting) ~ A Little Magic in Her Morning Coffee (saw similar in voting also) ~ The Seamstress Who Stitches the Seasons (we just did memory weaver) ~ The Midnight Grocery Run (actually completely overlooked this one and now I want to try it!) ~ The Woman Who Sweeps Away the Night (saw something similar in voting) ~ A Rainy Day, Inside Out (saw a couple with this idea in voting) ~ The Gardener of Impossible Things (another overlooked idea I might have liked) ~ When the Books Come Alive (saw a couple variations like this as well) Just for fun and to break up the wall of text, I created a quick image that is a mashup of a couple of the ideas AI gave me for this challenge. Then a quick image generation for the midnight grocery run tells me this is way more magic than I wanted for everyday magic. If I had decided to explore this, I would have had to make some major changes. Same goes for the gardener of impossible things. I really like the image, but it stretches beyond everyday magic in my head. I told AI what I thought the issue was with the ideas it gave me: The theme is a touch of magic to everyday life and these ideas are revealing entire worlds. AI then shifted gears and gave me a batch of ideas revolving around a cat playing with yarn. Some ideas were still way too fantastical like knitting spring, catching a moonbeam, or transforming the living room. I did decide to try one idea revolving around the cat knitting itself a friend. While this was super cute, I just wasn't sure it was really on theme. Perhaps it would have done better than my current entry, but we'll never know. It's at this point that I had my own idea and I shared with Chat GPT: Let's try a completely different idea. Maybe a city street in fall, possibly light rain. An old woman sits at cafe table. Subtle witch vibe, a knowing smile, small magic trails from her to a couple. She has just used a bit of magic to help this couple discover each other. At this point, I worked with Chat GPT to develop and refine the prompt, and this is the first generation of the idea. I'm actually quite happy with the image. To me, this is a little bit of magic in everyday life and an idea I didn't expect to see too much of. The goal is to just make minor changes at this point, and I ended up running two more generations. Below was my submission. Currently, it looks like I didn't hit the sweet spot for the majority of voters, but lately I've come to accept that. My goal now is to submit an image I like. I don't always caption an image, but when I do, I either have an idea straight away or I used AI to get the ball rolling. Rarely, if ever, do I use something straight from Chat GPT. In this case, I read this list of ideas and it sparked my own: Giving Fate a Magical Nudge. Thank for taking the time to read my blog. I really hope you enjoyed it. I would be very grateful if you could take a moment to clap this blog (up to 20 claps per users) and if you feel inclined, share your thoughts in the comments.
Comments (8)
beastxyz
I voted for yours - it had a refreshing narrative: using magic to help others fall in love. Definitely one of my favs in the challenge so I am sad that it did not do well. I am guessing if it was a playful younger woman it might have gotten votes, but this is just personal guesswork. I would have submitted the cute cat if I had thought of it - that was pretty creative too.
panos
fiirst of all: The images ib this blog are AMAZING. Now, as to the matter of AI assistance - not just to help in contests, in anything creative. I don't know. If you go to someone (whether AI or person), with the seed of an idea and that someone fleshes out for you.. to what degree the final result is yours? Food for thought.
base451121
Very interesting These my dear... thnks a lot!!
TTRPG_Player
BeautifuL! Need to read later...
Sealine
Totally valid point. I used to avoid AI help with ideas and prompts completely. Then I started using it as described above mainly for challenges when I was feeling really frustrated with them but still compelled to compete. Then I found it could save me a lot of credits. I can't see images in my mind, so everything would start with a vague concept, generate, see what I could do to improve, generate, over and over until I was happy. Now I can discuss an idea, listen to AI, take what I want, leave what I don't all before I ever see an image. The end product, definitely not all mine, but will I still use it as a tool, absolutely. I also find that if I feed my base idea into new models like GPT, the end product is very similar to the AI assisted image so really, I guess I've been doing it for a while anyway. (And yes, sometimes I do feel like a fraud, but I also remind myself that a large portion of people creating AI images are doing this and have been doing this. We never truly know how much of what is created is the person and how much is AI, whether model input or AI chat input.)
Sealine
Thank you and appreciate the vote 😊 At one point, it looked like I maybe would make top 20. That is no longer the case. Fortunately, I have learned not to stress about the results.