Greetings!

As an I.T. professional whenever there is an innovation or new technology being introduced sooner or later I always end up immersing myself in that innovation or new technology and in the case of AI image-generation Stable Diffusion I now find myself learning it for the first time. For the short time that I spent on Stable Diffusion, I found out that there are now a lot of features being introduced for it and there's a passionate community that is very happy to contribute to its advancements by creating models and extensions for free.

As of now, I am still learning about Stable Diffusion, how it works, and how to improve it by getting models and LoRAs. Installing an extension is pretty cool too! I am now able to generate decent images but not as vivid as those found on the web.

Part of my learning process about Stable Diffusion is trying out online AI image-generation tools which can blow you away with the results. For this blog, I tried an online AI image-generation tool called NightCafe. It is a very good tool for image generation. At the end of this blog I am also going to add some of the images that I generated on the Stable Diffusion that I installed on my computer they do not have a high resolution because my NVidia graphics card only got 4 gigabytes of RAM (I should have bought a machine with higher VRAM).

I used the same prompt that I got from CIVITAI on all of the images that I generated with just some little changes that changed the output a little bit but changing the presets on each batch that I generated made the images more different.

Girl with pink hair

I love anime and it's anime as my inspiration for generating these images of a girl with pink hair.

I just re-run the prompt and changed the hair accordingly.

Prompt: 1girl, medium curly pink hair, blue eyes, wearing plain white shirt, denim shorts, city, absurdres, high res, ultrasharp, 8K, masterpiece, looking at viewer
Model: SDXL 0.9
Sampling method: K_DPMPP_2M
Preset Style: Color Portrait

Prompt: 1girl, medium pink hair, blue eyes, wearing plain white shirt, denim shorts, city, absudres, high res, ultrasharp, 8K, masterpiece, looking at viewer
Model: SDXL 0.9
Sampling method: K_DPMPP_2M

Prompt: 1girl, short curly pink hair, blue eyes, wearing plain white shirt, denim shorts, city, absurdres, high res, ultrasharp, 8K, masterpiece, looking at viewer
Model: SDXL 0.9
Sampling method: K_DPMPP_2M
Preset Style: Color Portrait

Prompt: 1girl, short pink hair, blue eyes, wearing plain white shirt, denim shorts, city, absurdres, high res, ultrasharp, 8K, masterpiece, looking at viewer
Model: SDXL 0.9
Sampling method: K_DPMPP_2M
Preset Style: Color Portrait

Prompt: 1girl, long pink hair, blue eyes, wearing plain white shirt, denim shorts, city, absurdres, high res, ultrasharp, 8K, masterpiece, looking at viewer
Model: SDXL 0.9
Sampling method: K_DPMPP_2M
Preset Style: Color Portrait

Prompt: 1girl, long wavy pink hair, blue eyes, wearing plain white shirt, denim shorts, city, absurdres, high res, ultrasharp, 8K, masterpiece, looking at viewer
Model: SDXL 0.9
Sampling method: K_DPMPP_2M
Preset Style: Color Portrait

Girl with short black hair

Compared to the girl with pink hair I think this girl with black hair is more normal/real and you will see that in a bit. The only problem is that the AI included a watermark on some of the images. To avoid your images coming up with a watermark you can add "no watermark" on the prompt or "watermark" in the negative prompt. I also like this girl with black hair more because they look a little bit more real than the girl with pink hair. The makeup of this girl with black hair is less than the one with pink hair.

Prompt: 1girl, short black hair, blue eyes, wearing plain white shirt, denim shorts, city, absurdres, high res, ultrasharp, 8K, masterpiece, looking at viewer
Model: SDXL 0.9
Sampling method: K_DPMPP_2M
Preset Style: Color Portrait

Prompt: 1girl, short black hair, blue eyes, wearing plain white shirt, denim shorts, city, absurdres, high res, ultrasharp, 8K, masterpiece, looking at viewe
Model: SDXL 0.9
Sampling method: K_DPMPP_2M
Preset Style: Color Portrait

Pastel look

While browsing the internet regarding Stable Diffusion, Models, LoRAs, and extensions I came across images in pastel style. So I tried to generate images in pastel style using the same prompt. I cannot get the pastel look on two or three tries just by changing the prompt so I decided to play on the preset and it did. Check the images below.

I must admit that of all the images that I generated I really really love this pastel look because they look like a painting. The background is very nice too (which very much looks like a painting).

Note: In this batch of images that I generated I changed the preset of the last image into "Hyperreal" which seems to me that is no significant changes.

I changed the preset of this image into Hyperreal

Anime

I said earlier that I love anime. So, this post will not end without anime. The prompt that I used for generating anime images is the same as those above. The only problem is that I forgot what presets I used and the attributes of the images that I generated do not show what presets are used. Could it be that I forgot to choose a preset? I am not sure.

Using Stable Diffusion AUTOMATIC1111 on my computer

Just like I said earlier I am going to add the images that I generated using my computer where I install Stable Diffusion AUTOMATIC111. The images are not high resolution because my computer's graphics card only got 4GB of VRAM. The prompt is almost identical to the prompts that I used on the images above only but this time I also used negative prompts since it is locally installed on my computer. I also used a checkpoint that I download from CIVITAI. Here are the images...

Prompt: best quality, masterpiece, RAW photo,ultra-detailed,Korean 1girl ,looking at viewer, smile,rooftop, from above, city, casual,

Negative Prompt: 3d, 3d render, painting, digital painting,watermark, sepia, black & white,NG_DeepNegative_V1_75T, EasyNegative,(worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)), ((grayscale)),(ugly:1.33)

Model: Kawaii Realistic Asian Mix (https://civitai.com/models/83766/kawaii-realistic-asian-mix)

Sampling Method: DPM++ 2M SDE Karras

In generating the images I did not change anything on the prompt and in the negative prompt I am just clicking the generate button all the time.

image 2.png

image 14.png

image 13.png

image 12.png

image 11.png

image 10.png

image 6.png

image 5.png

image 3.png

That is all for now guys, catch you up with the next one. Wishing you all safety, good health, and abundance.

Thank you very much to all of the Hive Ecosystem Curation Teams/Individuals, I am very grateful for the upvotes/curation/support. God bless you all!

I am a Computer Engineer, blogger, farmer, gardener, father, and husband. I love countryside living, nature, and farming (rice/vegetables), and I have two decades of experience as an I.T. professional

Sort:

Trending

[-]

castleberry (74)Xenomorph last year

You got some great generations! It's cool to see the subtle changes.

$0.00

1 vote

afterglow (73) last year

Thank you. Yes, it is cool to see the subtle changes. I am amazed that AI was able to generate consistently.

maytom (72) last year

you are talented friend @afterglow

Thank you @maytom

nthtv (51) last year

Those are nice renders. Glad to see you were able to work with SDXL on a 4GB graphics card.

Have you tried ComfyUI as your front-end yet? It's a bit faster and resource friendly on low VRAM rigs.

But because ComfyUI allows total control over your workflow / pipeline, it can take time to master. But, if you're patient, it does expose you to more of a technical understanding of how Stable Diffusion works. It's quite fascinating.

Thank you. I have never tried ComfyUI as my front-end yet. But I am going to take a look at it when I have the time. Thank you for dropping by and have a nice day ahead.

Trying Out Online AI Image Generation Tool: NightCafe; Using Stable Diffusion AUTOMATIC1111 on my computer

Girl with pink hair

Girl with short black hair

Pastel look

Anime

Using Stable Diffusion AUTOMATIC1111 on my computer