AI Prompts

[lordsmalls]

I've been playing around with google gemini because it is the only AI generator that is free and I already have a google account anyways. I wanted to get a collection of prompts everyone has been using so we can all share with each other and maybe help improve the process. Here is one that is my favorite so far. I uploaded a picture of a girl onto gemini told it to work its magic.

1/100 scale of the character in the picture, in a realistic style, in a real environment. The character is placed on a bed. the character will be barefoot. Preserve character facial features. Do not change the face of the character.

[Spindrift]

This is a very good topic, and I hope a lot of experienced AI jockeys can reply.
I been using WebUI Forge, SwarmUI and ComfyUI, all tools within the Stable Diffusion architecture and one thing I've found is that the software just doesn't understand scales or sizes very well, especially when the subject in question doesn't exist at the prompted size in reality. For example, in your prompt you specify 1/100 scale. If the woman is originally 5 ft 5 inches tall, then she would be 1/100th of 65 inches or 0.65 inches tall (1.66cm) which she certainly isn't in the image above.

I've had the same issues where I prompt "3 inches tall", and get a whole range of sizes with little consistency. This is expecially true when I have a normal sized person in the same image. I can get close to the size scale I want if its just the subject woman with maybe a giant hand or some kind of giant household item like a cup or a spoon. But again, no consistency.

For example, this is the absolute closest I've ever gotten to my prompt of "3 inches tall". As you can see, the only other thing in the image is a giant hand, no background, nothing else. In fact, its not even an entire hand.




But pull back a little to get the whole hand, and all of a suddent the subject grows in size.




With even a partial face of a normal sized person in frame the subject grows even larger.




And with the entire face and upper torso of a normal sized person in the background the subject grows even more, despite the prompt's "3 inches tall" specification remaining unchanged.




Like I said, the software doesn't seem to know exactly what we're trying to produce and confuses scales and sizes, especially when normal sized people and objects are also included in the prompt. And prompting for "photorealistic" or "hyper detailed" actually seems to exacerbate the problem. I'd really like to hear from some of the AI experts out there and receive the benefit of their experience.

[Hand-Holder]

I like the results but all pics tend to have their heads just a bit bigger than the right proportion, otherwise I think you made a great job

But remember, Grok is 100% free and allows quite a bit of daily requests, also allows quite spicy NSFW content - Stories, Pics and Videos

Grok does not have Text to Video, but it has a great Image to Video feature

META AI free version is vey good in Text to Pictures generation and does little mistakes with limbs etc, it also handles size proportion requests very well, but is fully snowflake, it is an art to make something spicy with it, like, bikini or short dress or short nightgown - yes // But lingerie - No

I don´t think META AI has requests limit, but if it does it must be huge cause I did some AI-Generation-Sprees and never got that issue

[Spindrift]

I like the results but all pics tend to have their heads just a bit bigger than the right proportion, otherwise I think you made a great job

Sharp eyes! That's another failing of the AI software. When prompted to generate shrunken woman, it seems to want to generate dolls instead.

But remember, Grok is 100% free and allows quite a bit of daily requests, also allows quite spicy NSFW content - Stories, Pics and Videos

Grok does not have Text to Video, but it has a great Image to Video feature

Looks like I'm going to have to add Grok to my growing list of AI image generators to learn. Currently trying to grasp the basics of InvokeAI.

[lordsmalls]

Used grok ai for this,

the character is shrunk down to 1/100 her original size and is placed in a cardboard box. the view will be a top down perspective

[Hand-Holder]

I´m loving Grok and if I could I would jump right away into the paid version, but I already did my eccentricities this month with the PS4

[Hug0]

I've been using Gemini 2 & 3 for the past few months, really trying to push the limits for what it can post. You would be amazed of what G2 made for me before denying my prompts, Its very aggravating. I've used Perchance AI also, but not a fan on how unstable my results have been multi-fingers & all. I do have DRAW THINGS for my MAC but I can't figure out how the Hell to get started & get it set up.

Here's a quick commission I made for someone on DA.

[Hand-Holder]

Wowwwwwwwwwwwwwwwwww, this image tells more that it looks at a first glance

A LOT of Storytelling here !!!

Urban home (window showing it)

Teenager´s room (everything is chaotic and out of place, even the bed)

Frightening perspectives - They are fresh new toys for an irresponsible careless teenager...

If the hands holding the boxes were a bit little younger than it would be perfect !!!

But than again, it can be his Father, bringing some fresh presents as a surprise to be left on the Teen´s room...