How to make very realistic photos
Making realistic photos is very simple with AI Runner, but it fully dependent on the model and embeddings you choose.
In this example I used the following settings
- Model: nextphoto_v10
- Embeddings: veryBadImageNegative and EasyNegative
- Scheduler: DPM++ 2M Karras
- Size: 512x768
- Prompt:
a photo of a woman eating at denny's
- Negative prompt:
verybadimagenegative_v1.3, easynegative, cartoon, fake, not real, illustration, drawing
- Steps: 20
- Scale: 7.5
- Seed: 1574226477
- Samples: 1
This produces the following image
Next I navigated to Filters > RGB Noise
and increased the sliders slightly for each channel. This produces a nice grain effect
Lastly, I navigated to Filters > Blur > Gaussian Blur
and increased the slider to 0.03. The final image looks more like film
To use custom models and embeddings:
- Create a folder to store your custom models (for example
C:\Users\<your_user_name>\stablediffusion\custom_models
) - Create a folder to store your custom embeddings (for example
C:\Users\<your_user_name>\stablediffusion\custom_embeddings
) - Download your models and embeddings to these folders
- Next run AI Runner and navigate to
Settings > Path preferences
in the toolbar at the top of the AI Runner application and choose the model folder you just created forModel Path
and the embeddings folder forEmbeddings path
- Click ok to close the Path Preferences window
Your custom model will show up in the Model dropdown, and your embeddings will show up under the embeddings tab.
Use the names that show in the embeddings tab for the embeddings you want to trigger.
Update In v1.17.5 I have introduced the ability to apply the filters mentioned in this post (RGB noise and blur) with the correct settings for a film-like effect with the click of a button.
Just generate your image, then navigate to Filters > Film
and a small amount of grain and blur will be added to your image for a more realistic effect.
Files
AI Runner with Stable Diffusion | AI Art Editor and Custom Chatbots
AI Art using Stable Diffusion on your own PC
Status | In development |
Category | Tool |
Author | Capsize Games |
Tags | ai, art, artificial-intelligence, Asset Pack, Generator, Meme, Pixel Art, stable-diffusion, tool, User Interface (UI) |
Languages | English |
More posts
- AI Runner v3.066 days ago
- Overcoming Challenges: The Journey to Building AI RunnerSep 01, 2024
- AI Runner Development Preview: Interrupt LLMs, Stable Diffusion and Text-to-spee...Mar 08, 2024
- AI Runner v3.0.0 Dev Preview: Draw / generate in "real time"Feb 29, 2024
- Talking to my evil chatbotFeb 29, 2024
- v3.0.0 Voice to Text PreviewFeb 27, 2024
- Upcoming version informationFeb 02, 2024
- AI Runner now FREEJan 06, 2024
- v2.3.0 Milestone progressNov 25, 2023
- v2.3.0 progress updateOct 31, 2023
Comments
Log in with itch.io to leave a comment.
Thanks for the explanation! Thats helps so much
I am glad you found this useful. I’ll add more mini tutorials soon.