AI General Sharing Thread

Let me hop in, I made you several (that's the beauty of AI B):rock-right:)







Which one do you prefer? :D

Bonus in vector style

Somehow I knew you'd be the first to hop on the idea. :) Happy holydays to the both of you.

Thank you. And don't forget the cat :cutesmile:

I tinkered around a vector image prompt and got some pretty nice results.
(waifu pictures)









AI used as a filter is pretty potent I think.

@Indinera and @JohnDoeGames What do you use to generate your AI content, if you don't mind me asking?

I have slowly been doing a little research so I can start to explore AI generated images. I would like to generate placeholders that reflect what I would like to have in a future game. That will allow me to do more practical experiments with different mechanics and concepts, such as 3D views and such.

Rather than spend hundreds of hours searching for free images to experiment with, I'd rather generate images that are more closely reflective of what I want. That seems much more efficient.

Not that either of you are probably concerned about it, but for the sake of others who might read this and decide to judge me on it, I'm not committed to actually using the AI generated images in a game at this time. I am more curious about getting ideas and thoughts into something visual that I can experiment with. Much later I can decide if I want to find a human artist to replicate the placeholders.

I'm looking for something that is reasonably affordable, fairly easy to use, can be text prompt driven, and is mostly or completely unrestricted for content generation (think NSFW). I would mostly be doing text prompting, but I could see myself taking images of things I know I like and trying to prompt AI to make changes to meet my needs and vision.

I'm also interested in something that lets you make changes to the image that is created and that lets you generate additional images that carry over the same characters. Like if the AI creates a character that I absolutely love, I would want to be able to generate more images of that same character consistently.

I'm not against buying software to download on my computer or using a browser interface, but I also don't want to spend a crap ton of money either since I'm just exploring right now.

I did generate a couple of images with this site yesterday and I was very impressed by the quality.

Hunyuan Image 3.0: Tencent's New Open-Source Text to Image Model Free Online - EaseMate AI
I haven't learned much about that site, but one problem I saw was that I didn't see how I could edit an image. One prompt generated a very cool image, but I could see a few minor mistakes that ideally I could "chat" with the AI to point out problems that needed to be fixed. But maybe the software isn't at that point yet.

For instance, the AI generated an image of a D&D party of 4 characters and it was awesome except one character was apparently supposed to be holding a war hammer across her shoulder and the handle actually seemed to be going through her shoulder. Pretty much everything else was spot on.

Also, I didn't see how I could generate additional images of the same characters with that software. But I only just started looking at that site.

Anyway, I'm just curious what you both are using and if you have any advice on where I can learn more about AI image generation and any good sources that I can play around with or any communities that how good discussions and guidance on people exploring and trying to learn.

I am very willing to read tutorials, watch videos, etc. if you have any recommendations on learning to use the AI image generation.

Also, feel free to send me a PM in response if you don't think this thread is appropriate for this sort of conversation or if you want your response to be more private.

Thanks in advance and I hope that 2026 is off to a great start for you!

I'm using Gemini.
It will censor NSFW content but you can still do kinky stuff if you're smart lol

For videos I'm using Grok but I generally prefer to generate pictures for some reason (Grok tends to be OTT kinky, too).

These for instance were achieved (for entertainment purposes obviously) using Gemini to alter my own source material.






And one a little bit late lol



Generally speaking you can either prompt from nothing (just text) or use a base image and edit it to fit whatever you like (my favourite method as you retain more control on the final result and it's a lot more fun).

Prompt refining generally goes through a sequence of "try and fail" until you get exactly what you want.

@Indinera Thanks for that! I hadn't tried Gemini yet. I just did with the following prompt:

"An image of a D&D style adventurer group with a Barbarian, Druid, Wizard, Cleric, and Rogue. The party members are children ages 10 to 14, 3 boys and 2 girls, all holding appropriate weapons for their D&D classes, in a ready to fight pose, and the setting is a crypt."

And this is the result:



Wow! That is a pretty cool result!

One of my ongoing novel trilogies is exploring how D&D type adventurers get their start. What would it be like if an adventuring party met and started their journey as children instead of adults.

I would want to fine tune this image to be more similar to my story characters, with more specific details of each character, but this looks like a pretty good book cover to me.

Very, very cool.

And I would think that giving this to a human artist with key notes about what needed to be changed would make the commission process so much easier.

EDIT: Holy Crap! I just followed up with the following with Gemini:

"Can you make another image with those exact same characters only this time they are sitting inside a tavern, having a meal, and resting from an adventure?"

And this was the result:



Those characters look pretty similar to me.

Just incredible. My mind is blown.

EDIT EDIT: Although I see that it randomly duplicated the Rogue character at the table. LOL!

AI actually does a lot of cool things.
Which one did you use?

I just used Google Gemini (Fast). After generating a few images of that same party and starting to get more mistakes, like missing party members or merged images, I then tried (Thinking) and the result was insanely better.

Here is another prompt that was giving the (Fast) option problems:

"Please create an image of that same party, traveling down a forest path,
looking very tired and worn out, and it is close to dusk."

Here is the result of the (Thinking) option:



I could have a lot of fun with this. I need to look at what you get by upgrading. I'm just using regular old Goggle Gemini for these.

I did discover that Gemini won't generate an image of that party actually fighting if anyone is injured. It said it could generate images of minors but not with that content. I only asked for the Barbarian to be injured while they are fighting a bear. After removing the injured part, this is what it came up with using the (Fast) option.



Notice the Bear holding a staff? LOL!

I've only ever used (Fast) for my images. I like that you get so many attempts with it. To be fully honest I used (Thinking) once or maybe twice and it didn't give very inventive results, so I dropped it. On the other hand the other one has more often than not left me satisfied or at least entertained. Needless to say I've never felt the need to upgrade either.

About "minors", I get that reply often (even though I've never used a minor's pic ever), it tends to see minors everywhere even with a pic of a 25-30 years old lady lol so I think it's just served as an excuse to not do something it doesn't want to do. Wounds might be one of those things even though I've never had trouble to generate pretty gory pictures.




Try to ask Gemini to remove it lol

I mostly use various generators from the Perchance web site since everything there is free, I have used other sites but they either got taken down or have limited use. I have also downloaded Automatic 1111 and Comfy UI, but my results are not as good and take a long time due to my graphics card not being strong enough for that. When it comes to cleaning up and enhancing the images, I tend to look up free online tools or use Paint.net to fix as many of the errors as possible. I do provide the links in the descriptions when I upload my images to Deviant Art, so people can try those tools out.

Yeah, I use gemini too. Not so much for images, though. I like to upload my writings and coding, to talk about my creations with AI in a live-chat. For this I also use Gemini or NotebookLM. Both services by google. I use notebookLM to generate audo or video material I can download and listen to while on the go. It may sound weird, but I get so many idea's in the bus, while I am listening to an AI generated podcast about my stories or my games.

Note, if you ask either AI to talk about your work, tell them what you want: Honest and critical feedback. If you don't say that, they will tell you everything is amazing and you're a genius, no matter what you made.

So, for generating images: Gemini.
For discussing my work: NotebookLM.

Oh, and whatever you do, never ever, even if your life depends on it, especially when your life depends on it, ask AI for advice. They will send you straight to your death, as long as the answer they provide makes you happy. (If they give advice you don't like, just tell them it is stupid, and they will advice the complete opposite. AIs only care about providing you with an answer that makes you think the AI did a good job.)

Thanks @LazerRay ! I'm going to check out that site too.

And thanks @JohnDoeGames for your input!

Out of sheer, morbid curiosity, I asked Gemini to convert the party into pixel sprites for an RPG Maker MZ game. Then I asked it to make 10 pixel sprites in different poses for the Barbarian for RPG Maker MZ. Then I asked it to make a walking sprite for the Barbarian for RPG Maker MZ.

I didn't give it any specific dimensions or anything and I haven't tested these. But I think the results are pretty interesting anyway. Especially for just playing around, not trying to fine tune anything, and not giving Gemini more specific requirements. I like the random green bear in the back of the first image. LOL!

EDIT (1/4/2026): Added a SPOILER so the thread loads faster.





Very interesting indeed.

This looks pretty good to me. Keep sharing your stuff.

Day 2 of exploring what Google Gemini can do. I haven't upgraded to a monthly subscription so I'm limited on the number of times I can use the Thinking or Pro option. But I definitely see a difference in overall quality of the images. It seems like the Fast option is more prone to mistakes.

I'm also figuring out how to fine tune descriptions to modify the images.

I have been trying to get Gemini to make a walking sprite sheet, but it hasn't understood any of my descriptions so far. It certainly gets some poses that look correct, but I would probably have to cut and paste from multiple attempts to get a valid sprite sheet.

Here are just a few samples that I made today with experimenting with different things.

EDIT (1/4/2026): Adding a SPOILER so the thread loads faster.



I think these come out pretty darn good.

And I have been experimenting with other AI generators and you can DEFINITELY see a quality difference depending on what you use.

I am terrible at visualizing so this is really great at helping me see concepts and what I am thinking about as a starting point.

The fairy's cute and I like the one with the black cat too.


Isn't Gemini the best?

@ZombieKidzRule These character features look great
I tried, but the AI Im using doesnt seem to understand pixel art

Google Gemini seems to work very well and it is free. It does have some restrictions that are annoying but the safety filters are important.

I’m still experimenting with Gemini and learning how to interact with it.

Give it a try and see how it does with pixel images.

I found this in an ad... he did some really interesting things with the 2k3 monsters

Day 3 of experimenting with Google Gemini. I focused more on Pixel Art generation to show @LuigiNumber1 what it can do.

I am currently working on generating a bunch different portraits of the same character to reflect not only emotion, but also the results of status effects from conditions. I am going to finish up the portraits and then put them together in multiple Faces sheets so I can do some experiments with them in MZ.

The portraits look really nice and they are super quick to generate.

Anyway, here is a SPOILER again to make it so the thread pages loads more quickly. These are all Pixel Art style generated images by Google Gemini. They are a mix of regular assets and icons.


I only hope that I will someday be able to use Aseprite to make Pixel Art even half a good as these.

i liked the skeleton