Stable Diffustion 1.5

Stable Diffusion version 1.5 is the most supported/used image generation model, with a huge community.

Its quality is not as good as FLUX but the amount of NSFW base models and LoRA available is massive.

This guide shows you what each configuration mean on the DopamineGirl Studio:

Prompt Mode

• Text Only: Using Text Prompt and Negative Prompt to tell the AI how to generate images

• Regional Prompter: Divide the image into a 2D map of regions, then assign to each region individually

Most of the time you can go pretty far with Text Only, with custom LoRA the possibility is endless.

But if what you are trying to create has a complicated composition then Regional Prompter might be a better choice.

Prompt

Describe the image you want to generate using English. Most of the time, the more detail the better image, but try not to put to much into this.

The maximum characters allowed is 20,000.

Negative Prompt

Describe what you don't want to see in the image.

For example: bad hands, teenager, deform, children, etc.

The maximum characters allowed is 20,000.

Base Model

You can think of this as themes or styles.

Most models I collect the dataset and train them myself, but for some models are downloaded from CivitAI.

Try to experiment with these models and see which one you like best.

You can also use the search engine to see what people generating to get a feel of what these models can do.

Prompt Guidance

Higher value means the AI will follow exactly your prompt. Lower value can generate more creative images but more chance something weird will show up.

• 2 - 6: Creative, but might not follow the prompt

• 7 - 10: Recommended for most prompts. Good balance between creativity and guided generation

• 10 - 15: When you’re sure that your prompt is good/specific enough.

• 16 - 20: Not generally recommended unless the prompt is well detailed. Might affect coherence and quality

The rule of thumb is: the more detail you add to your prompt, the higher value this should be.

Seed

Some parts of the AI pipeline use randomness to generate your images. This seed value controls that randomness.

Seting the value to negative (like -1, -2) means the system will pick a random positive number for you.

Keeping the same seed and everything else the same, then the output image will not change.

When generating multiple images at once, make sure to set this value to -1, or else every images will be the same.

Generate

Generate multiple images in parallel.

Make sure you set the seed to negative, otherwise all output images will use the same seed and will become identical.

Samplers

The AI will use the seed to create some random noises, and the job of the samplers is to turn this noise into beautiful images.

Try experiment with different samplers and see what works best for you.

You can click here to see the comparison between all samplers.

Scale 4X

Turn this on to double the width and height of the image, hence the name scale 4X. Usually the scale version is even better than the normal one!

Private

Turn this on so only you can see the output

Hide prompt

Turn this on so only you can see the prompt