Courses

Graphics

Merida

comfyui - open terminal, cd comfyui, then python main.py
wan wrapper tools on comfyui video
Foocus
Automatic 1111

Image Prompts

https://openart.ai/community/AqkSeee40nBncNncCO6a

https://huggingface.co/Gustavosta - prompt generators

style, backgrounds, editing, faces, emotions, lip synching

'Please create an image'
- Description
- Character
- Pose
- Emotion
- Fashion
- Setting
- Style - the style of the photography or art
- Artist (paint, photographer)
- Lighting
- Camera Angle

Consistent Characters

https://ailemonacademy.ck.page/
Consistent Characters

Fooocus, Google Co-labs
Fooocus - consistent faces

Face Swap Image Input (prompt)
Text Prompt Engineering, word order, girl's name, repetition
Fixed Seed
Training


https://www.youtube.com/@iammiameow/videos - consistent characters via bot-GPT creation in ChatGPT 4.

'Consistent Character GPT,' can access through ChatGPT 4

Character Consistency
- Prompt Detail
- Face Swap
- Custom GPTs
- Software
- Sprite Sheets
- Training

Characters, ArtFlow

https://www.youtube.com/watch?v=UdkQX88clIM

https://www.youtube.com/watch?v=Sc2f2oTaR7U

https://www.youtube.com/@DigitalSuccessAI/videos

https://intercom.help/leonardo-ai/en/articles/8584503-creating-consistent-characters

https://characteraigenerator.com/

1. image refs into Gemini, which calls Nano Banana
1024 2x2 grid different scenes
1. image and style reference (midjourney)

2. seed

3. comic book with number of frames
4. Train a Loras

Consistent Characters

Image Editing

PhotoPea - Online photo editor

https://www.remove.bg/
PNG with transparent background

Adobe Background Remover

- Small Size (.jpg)
- Raster Image (.png)
- Vector Image (.eps)

Upscale

In-Painting

Open Source

ComfyUI (GitHub) for modular workflows.
Diffusers (Hugging Face) for Python-based inference.

Fooocus
Fooocus - GitHub File Repository

Juggernaut XL

https://github.com/easydiffusion/easydiffusion
https://easydiffusion.github.io/

Slow generation → Update to Nvidia Driver 531 Speed Optimization:
Use LCM-LoRA: 4-8 steps instead of 20-50
Lower resolution: Generate at native res, upscale later
Batch generation: More efficient than sequential
Use fp16: Half precision (torch.float16)
xFormers: Memory-efficient attention

Quality Optimization:
SDXL Refiner: Use for final detail pass
High-res fix: Generate small, upscale with SD
Negative prompts: Critical for SD 1.5, less for FLUX
ControlNet: Add pose/depth guidance
LoRAs: Fine-tune specific styles (1-2 max)
Local Installation
Stable Diffusion 1.5/2.1: Easily at full precision
Stable Diffusion LCM-LoRA
https://stability.ai/license
https://platform.stability.ai/

https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/stable_diffusion.ipynb https://colab.research.google.com/drive/1Ab5bnDxRh69UMV8UMkz_rGWRkO9-vSsd

- rename fooocus and data folders -

Image FX by Google:

Not yet available in Thailand, can try free VPN services:
https://aitestkitchen.withgoogle.com/tools/image-fx

https://huggingface.co/black-forest-labs
https://huggingface.co/spaces/black-forest-labs/FLUX.1-Kontext-Dev
FLUX.1 [schnell] - 12B parameter, image generator
Start with Flux.1 [schnell] on Replicate or Poe.com to test without local setup.

https://huggingface.co/latent-consistency/lcm-lora-sdxl
SDXL (with --medvram), DeepFloyd IF (if RAM allows)
. Segmind Stable Diffusion (SSD-1B)

https://huggingface.co/dreamlike-art/dreamlike-photoreal-2.0

    Merida

Chrome Extensions (tools)
Awesome Screen Recorder
1. Consistent Characters

2. Censorship issues

https://huggingface.co/meituan-longcat/LongCat-Image Start the Image Gen Tool Locally: Launch your chosen tool in the background (e.g., via terminal/command prompt). For A1111: python webui.py --api (runs a web server with API). For ComfyUI: python main.py --listen (enables API). For Fooocus: python entry_with_update.py --listen (enables server mode). This makes the tool accessible via HTTP on localhost. PhotoPea - Online photo editor

In capcut, can import videos, and capture individual start and end frames. So, you might use a viral video, capture a frame, and upscale the image, then use that frame image as a base to create an ai image, then use that image as the start and end frames for our own video.
+ Change language, change soundtrack, and change background.
https://3d-models.hunyuan.tencent.com/
3D modeling
3D Graphics
Take screenshots of cropped areas of my screen, chrome extension captures.

Singularity Tutorials

Atomic Gains

CapCut

ChatGPT PLUS - need paid plan to use Dall-E 3

-----

Canva - uses Stability.ai
Canva Magic Studio

Can use Andover.edu or UC Berkeley to get .edu email, and can sign-up as a student or educator, and get canva free or lower cost.

https://www.youtube.com/@metricsmule/videos

------

Unstable Diffusion - X-rated

Bing Image Creator - needs Microsoft hotmail signin

-------------

Ideogram - can add text spelled correctly to an image, for logo, T-shirt design

DreamStudio - often blurs sensual content

Lime Wire - use the Blue Willow Model, limited free usage

GenCraft - can use trained models if upgrade

Google Gemini - temporarily can't create images of people

Use generated images as sprite sheets, for animation.

Adobe Creative Suite

Adobe Creatve Suite - 'premier pro' and in fact, many of their mainstays, have incorporated elements of Runway ML and Pika Labs into their products. Thus, buying a subscription to the Adobe Creative Suite ensures a long-lasting and developing tie-in to modern tools.

Photoshop Generative Fill: Adobe Creative Cloud subscription. add new elements to existing pictures, remove objects, and generate brand-new scenes. The feature uses Adobe’s own Firefly AI model so you’ll get different results from any of the other image generators.

Adobe Audio

Adobe Video Animation

SOURCES - ChatGPT can source from Wikipedia Commons or Public Domain works.

The Metropolitan Museum of Art - must check the box: 'Open Access.' Each image will list 'pubic domain' under the image at the left.

Adobe Stock Images - with paid subscription, have access to copyright-safe images.

3D modeling, 3D Graphics

https://3d-models.hunyuan.tencent.com/
WordTrance.com