I got to thinking about this today. Occasionally I will use Bing AI's image generator to make images for different projects I am working on. Earlier today I made a series of California images and they came out pretty good. Of course there are the typical AI glitches like on one image my bear had a fifth leg and AI can't seem to spell "California" correctly to save its life.
But I got to thinking about printing them out on cardstock, adding some gems or stickles or something and making up my own sets of embellishments. Like a lot of us here I often don't care for many of today's scrapbook collections and I struggle with finding embellishments that work for the topics of my pages. It seems like this could be a good solution. Of course many people are morally opposed to AI and I understand why but I like the idea of using this to help me in my creative endeavors.
Here's a couple of the ones I made.
Last Edit: Apr 14, 2024 20:23:43 GMT by chaosisapony
Thanks ladies! I'm having fun, I've made a couple other states and I am going to be doing a location in a retro style that I can use my Silhouette to cut out in a postage stamp format.
If anyone wants to try I've been using Bing AI. It's free and very easy.
Okay, this is so much fun! I can see that I am going to waste spend a lot of time doing this. I am always on the lookout for flower stickers, and I just got AI to make me some.
For the first lot, I asked for "a bouquet of gerberas sticker in pink, orange, yellow"
It wasn't quite what I wanted, so for these ones, I asked for "hand drawn bouquet of daisies sticker in aqua, turquoise, pink, pale yellow"
The second ones are closer to what I wanted.
I also did some watercolour flowers, a bunch of balloons with a personalised message for a friend, a cartoon quokka, and a cartoon Santa on the beach.
grammadee for this one I put in the instructions "a cartoon gnome in red and white hockey outfit with Dee written on the front." I didn't even ask for a maple leaf on the uniform, but it must have assumed due to me asking for red and white.
Oh Oh Oh - you are bad peas!! I need to learn to do that!! I often download images to my Silhouette and use the print and cut feature - but that doesn't let me make my own! Like rugby stuff - for girls.
How do I make this work with my Scan n Cut? If you find any tutorial links please let me know,.
The only thing I am able to do is take the image into photoshop and remove the background, then save as png. When I import it into Silhouette Studio it gives me automatic cut lines. If I don't save as a png then all Silhouette will recognize is a big square box. Sorry, I have never worked with the Scan N Cut machine to know about the differences.
Okay, this is so much fun! I can see that I am going to waste spend a lot of time doing this. I am always on the lookout for flower stickers, and I just got AI to make me some.
For the first lot, I asked for "a bouquet of gerberas sticker in pink, orange, yellow"
It wasn't quite what I wanted, so for these ones, I asked for "hand drawn bouquet of daisies sticker in aqua, turquoise, pink, pale yellow"
The second ones are closer to what I wanted.
I also did some watercolour flowers, a bunch of balloons with a personalised message for a friend, a cartoon quokka, and a cartoon Santa on the beach.
Ok those came out awesome! I never thought about adding the word "sticker" into the prompt but I am assuming that is why you got that nice, tailored white outline. How perfect!
grammadee for this one I put in the instructions "a cartoon gnome in red and white hockey outfit with Dee written on the front." I didn't even ask for a maple leaf on the uniform, but it must have assumed due to me asking for red and white.
grammadee - did you say something about a gnome on a stump?
Love the bear! Our bird feeders have been just fine for the last year and now twice in one week we've been hit. We need to get them up higher. Need to figure out how to make some images for that.
Yes! I posted about this a while back, I signed up as a beta user of Dall-E. Which I guess is the tech that Bing uses now?!
I really like your CA montage, that's what I was trying to do with it too, like make things more niche.
I haven't played with it in a bit, so going to see if I can get some better results. I like the idea of the sticker too!
I've also been trying to look into AI sorting/storing to manage stash inventory, but haven't found anything right just yet.
Awesome! Do you have any prompt tricks to share? I'm trying really hard to get it to omit words. Like, for example, many of my California images the tech is actually putting the word "California" on the image. Only, it can't spell it right, so it looks ridiculous but I'm not having any luck just getting it to not use words.
I'm trying really hard to get it to omit words. Like, for example, many of my California images the tech is actually putting the word "California" on the image. Only, it can't spell it right, so it looks ridiculous but I'm not having any luck just getting it to not use words.
I'm having trouble with that too. I wanted to have three different coloured balloons, one balloon with 'Happy' on it, one with 'Birthday' on it, and one with 'Jeanine' on it. Out of four images, it got one correct. The others said: Happy - Happy Birthday - Jeanine Happy - Birthday - Happy Jeanine
Yes! I posted about this a while back, I signed up as a beta user of Dall-E. Which I guess is the tech that Bing uses now?!
I really like your CA montage, that's what I was trying to do with it too, like make things more niche.
I haven't played with it in a bit, so going to see if I can get some better results. I like the idea of the sticker too!
I've also been trying to look into AI sorting/storing to manage stash inventory, but haven't found anything right just yet.
Awesome! Do you have any prompt tricks to share? I'm trying really hard to get it to omit words. Like, for example, many of my California images the tech is actually putting the word "California" on the image. Only, it can't spell it right, so it looks ridiculous but I'm not having any luck just getting it to not use words.
I understand why it can’t get fingers and legs and the like right, I think, but I’m perplexed by how it’s spelling California wrong if it’s getting that from the prompt. Is it like the fingers - it keeps duplicating letters? Perplexing… I really don’t understand much about AI…
I’m perplexed by how it’s spelling California wrong if it’s getting that from the prompt. Is it like the fingers - it keeps duplicating letters? Perplexing… I really don’t understand much about AI…
I’m not an expert in this area but I gather, analyze, and write reports on aggregated data.
AI outputs relying on machine learning models are subject to faulty output due to crowdsourced data. Advanced algorithms can only do so much when massive amounts of unstructured data are continuously received as input.
In other words, laymen are training machine models. Consequently, ai is subject to human errors. There’s obviously more nuance to it, but hopefully I’ve conveyed the idea.
Awesome! Do you have any prompt tricks to share? I'm trying really hard to get it to omit words. Like, for example, many of my California images the tech is actually putting the word "California" on the image. Only, it can't spell it right, so it looks ridiculous but I'm not having any luck just getting it to not use words.
I understand why it can’t get fingers and legs and the like right, I think, but I’m perplexed by how it’s spelling California wrong if it’s getting that from the prompt. Is it like the fingers - it keeps duplicating letters? Perplexing… I really don’t understand much about AI…
I'm confused by that too. I'm spelling it correctly in my prompt. The big things I notice are letters repeating and vowels being omitted. So I have gotten a lot of "Calfornia" and "CCaliiforna"
I’ll try to better explain why you may get errors:
Machine learning models work by making predictive outputs based on aggregated data. We (the users) train ai through our inputs (prompts). Thus, the model uses our input and all similar inputs to generate output.
Think of all the people entering a prompt related to California. Now imagine how many may have misspelled California or written prompts that are California adjacent (maybe nothing to do with the state but spelled similarly by the person writing the prompt). That’s an example of how you may get data errors in the output.
Think of prompts entered into ai as input that hasn’t gone through quality control because they’re entered by everyday people.
*disclaimer - This doesn’t apply across the board. There are different ways of entering and analyzing data. I’m trying to give a general breakdown to (hopefully) add insight to why you may get wonky results.
I have been messing around with it and I have found if you put "" around the word you want in the image, it puts it in there correctly spelled. So far. When I put in orange tractor with dad on it a field with pine trees in the background - it brought up an orange tractor with an old guy and kid on and near the tractor When I put it in as orange tractor with "dad" on it a field with pine trees in the background - it had the word dad on the tractor like I wanted.
I'm trying really hard to get it to omit words. Like, for example, many of my California images the tech is actually putting the word "California" on the image. Only, it can't spell it right, so it looks ridiculous but I'm not having any luck just getting it to not use words.
I'm having trouble with that too. I wanted to have three different coloured balloons, one balloon with 'Happy' on it, one with 'Birthday' on it, and one with 'Jeanine' on it. Out of four images, it got one correct. The others said: Happy - Happy Birthday - Jeanine Happy - Birthday - Happy Jeanine
I had the same thing, I was trying to put a phrase in the sticker and it did do it (I used quotes around the phrase) but some of the letters came out a bit weird looking. Not sure if you can somehow set like the font or size?
I do not know how to get the word out of it though, maybe try something like "solid image of California" or "the shape of California"? I wonder if maybe not capitalizing it would help?
Now I have to play with it more...
I know Canva has a lot of AI functions now in the upgraded version, which I do not have, but maybe poke around on their FAQs to see if they have any tips?
I've been using ChatGPT to help me plan a trip itinerary and a few times it has returned results of things that are closed or no longer exist. Kind of reminds me of when maps doesn't really understand some of our crazy jug handles and roadways in NJ.
Ah, love that last sticker though - but goodbye digital designers/artists. I also sometimes wonder about how it "borrows" images from other art. I duno?!
I have been messing around with it and I have found if you put "" around the word you want in the image, it puts it in there correctly spelled. So far. When I put in orange tractor with dad on it a field with pine trees in the background - it brought up an orange tractor with an old guy and kid on and near the tractor When I put it in as orange tractor with "dad" on it a field with pine trees in the background - it had the word dad on the tractor like I wanted.
That tractor came out great!
I tried adding quotes but it didn't seem to change my outcome.