Tried just the word romance, to see what this would do:
I think it’s stuck on Asian features because of what I was trying before that:
2 Likes
I don’t know what AI software MidJourney uses. One article suggests its a mix of Stable Diffusion and Dall-E2. But I seem to have better luck with MidJourney than the other software I’ve experimented with.
More image samples / lotto inspirations…
Update: Found this reddit forum with MidJourney examples. Might be of use.
2 Likes
I’ve been doing a lot of these silhouette on teal images lately, trying to come up with something good for the cover. I saved hundreds of them, that’s how many came out awesome, but I wanted it to fit the story too.
I’ve finally settled on this one. I didn’t feel like fixing it in an editor so I did a couple of things in Canva (mainly the moon and the birds had to be fixed) and put the cover together. I think it looks good. I hope it doesn’t look too much romancy since it’s not a romance story (though if readers ship these two, I won’t mind).
Since this is a sequel, I’m going to update the first book’s cover too to go with this one. I hope one of the hundreds of images I’ve accumulated will fit. Bigger challenge is changing the title. I’m still brainstorming it.
Gasp. Damn. I missed it before. There’s an extra leg. I’m going to have to fix it in an editor after all. 
2 Likes
The crurrent ones I’m doing in a poll to figure out my cover.
1 Like
Btw, why do you screenshots? You can just download it from this screen (without having to do the ads).
1 Like
Because it’s a hassle to me?
Next screen:
So, publishing it makes it useable for others, which I’m mostly fine with, but I’m generating them as fast as I can, when I’m searching for something. That and sharing doesn’t put the file in my control.
These are the options for download:
Honestly, a lot of different files for photos don’t automatically go in the photo phile where I can just pull it up on my phone. Sometimes I have to search for where the download went. So, a lot of extra steps when a screenshot gives me what I need. I’m going to have to do a minor edit or two before I’m done anyway.
1 Like
I don’t go to the next step though.
On this screen, I expand it to see what it looks like and if I like it, I download it right from here.
1 Like
Clicked it. Says saved. Not in my photo file.
And I have never figured out how to access my files like I did on my S7.
Trying to get it in here it shows that it’s under Dream Art:
I’d be able to get them off the phone by connecting it to my computers. But the screenshots I can easily move around and Even edit a hair on my phone.
So, u til I figure out how to easily access a saved picture in the phone photo editor, saving through the program ain’t an option.
1 Like
Did you try using the browser instead? Works the same but saves in downloads.
1 Like
It’s an app on my phone. So I never tried it.
And just did.
So, when I do it on the app, it will still be screenshots. But I’m not always going to remember to pull up the website. Lmao
1 Like
I’m not sure what your phone can and can’t do but my phone can save webpages like they’re apps.
Here I have both. Dream as an app and as a browser (the one that has a little chrome icon).
NC is Night Cafe btw, also a browser version.
I saved WackyWriters the same way too.
That’s Canva on Chrome. I actually want to save it like this too because only on the browser version I can download an individual page.
1 Like
It did for wacky, but then the wacky button quit acting right and I had to get rid of it, so I just pull up Chrome directly. Never did figure how to do that on purpose, though. Lmao
1 Like
Ah, nmd, just did it. Me and my deliberately not learning the extent of my software.
2 Likes
Wow this took a ton of editing without it looking like I did any editing, since the starting AIgen was good.
3 Likes
Very cool image.
But if that’s a cover, i think you can improve the font.
1 Like
I probably could. I didn’t want to fight with it further. lmao
2 Likes
But really, its about what fonts read well as thumbnails with that busy a background. That limits things.
1 Like
I found the reddit forum for Stable Diffusion, and through that found a website / forum for sharing Stable AI art and their prompts / models. Some of the art is quite good…almost identical to some professional Anime artists I know of (intentional or not)…The displayed images are better and more accurate than what I usually produce with Mid Journey (and with no hyper-twitchy / false positives aplenty prompt censorship). Am now considering switching to Stable Diffusion for image generation.
On that note, I found some videos for downloading / installing Stable Diffusion onto a personal computer.
One more reason to upgrade my ageing system, I guess.
Recent Mid Journey experiments
^Copied over a SD prompt list from the civitai website to check compatability.
SD prompt used in MJ, without the 'banned' entries
girl, cyberpunk augmentation, cyberware, cyborg, carbon fiber, chrome, implants, cyber plate armor, dark atmosphere, dark night, (black short disheveled hair:1.1), black eyeshadow, beautiful detailed glow, detailed, Cinematic light, intricate detail, highres, rounded eyes, detailed facial features, high detail, sharp focus, smooth, aesthetic, extremely detailed, industrial factory background, slim body, stylish pose
2 Likes
For some reason, I really like the cars.
2 Likes
I’ve considered doing that, but not my computer because it doesn’t have enough power. Apparently it’s quite cheap to use an AWS instance for it. Could cost just a few cents per generation. Allegedly.
But I would try it only if I wanted to train my own model. The idea is that you can feed it a series of images and I’m guessing keywords that go with them and then it can create similar images in that style or of that person/object.
So that’s the use I’d expect out of it.
Yes, you wouldn’t have to worry about the censorship as much either but there are limitations too. 2.1 is more accurate than 1.5 for example, but they’ve also sensored the training images a lot more and I’m seeing a difference in quality of the… hmm how do I put it, the feeling? The images evoke fewer feelings from me. They feel colder, dead.
But I wouldn’t expect my own model to give me higher quality results overall.
Honestly, I’m pretty impressed with your car images. I haven’t been able to create anything like that with Stable yet.
As to your character portraits, the only thing I’d change is to make them more dynamic, to show their personality. But I imagine that Midjourney must be capable of that too.
1 Like