Scottlin Dang's Gen AI Portfolio

Live Portrait

Use Live Portrait in ComfyUI to transfer facial movement and expression from video to still image.

Flux in ComfyUI
Open source AI applications and models [ComfyUI, CogVid, Flux dev] + Adobe Illustrator + Davinci Resolve
Images : ComfyUI Flux - Videos : CogVid - Editing in : Adobe Illustrator + Davinci Resolve
Scottlin Dang's GenAI Portfolio Home Page (special edition)

A personal customized lora: FLUX scottl2n Lora Model

Using AI technologies and applications to train personal Lora model of me and turning myself into any character out of my imagination. Generative AI technologies and applications assist artists and designers to accelerate their creative process.

Flux Gym: open-source AI application used to train images into Flux lora model.

ComfyUI : open-source AI application used to generate and modify images.

Adobe Illustrator : traditional digital graphic design application used to compose images and typography.

 

Comfy UI - Animate Diff

Explore the integration of IPadapter and Animatediff within ComfyUI to produce dynamic videos from static images.

Model: DreamShaper 8.0.

IPadapter and CLIP Vision models: IP-adapter sd1.5 (Vit-G), Clip vision G.

AnimateDiff model: mm-sd-v1.5-v2.

Ksampler: euler.

RAIA - Repurposed AI Arts Videos

Project RAIA or Repurposed AI Arts is a hobbies of mine where I can express myself through Gen AI videos, music and poetry

Generate images: from Stable Diffusion A1111 (combine with WaveLoopBack Script to generate sequential series of images).

Use ComfyUI: to generate videos from text (AnimateDiff workflow combine with steerable motion node and Prompt travel).

Use Text to Speech: to generate voice reading poetry.

Use Davinci Resolve: to put images sequence to video, and shorts videos together, adjust frame rate, correct colors, generate subtitles, and create motion graphic for logo. 

For more AI videos - [RAIA-Repurposed AI Arts] @Raiad3sign
Instagram Tiktok Youtube
Gen AI open source applications + Davinci Resolve for Video Editing Experiment

In this project, I aim to explore a diverse range of video editing techniques, seamlessly integrating both traditional methods and cutting-edge GenAI technologies. My goal is to enhance efficiency and streamline the editing process.

Gen AI open source: Stable Diffusion Automatic 1111 and ComfyUI.

Text to Speech open source: Applio RVC Fork.

Lip sync open source: Sadtalker.

Video edit: Davinci Resolve

Vision Igniter

For this project, I intend to delve into a wide array of video editing techniques, seamlessly blending conventional methods with state-of-the-art GenAI technologies, including music generation. My objective is to improve efficiency and simplify the editing workflow

Gen AI open source: Stable Diffusion Automatic 1111 and ComfyUI.

Music Gen: Udio.

Lip sync Gen: Hedra.

Video edit: Davinci Resolve

Depth Map in Stable Diffusion
Using Depth Map Diffusion to generate depthmap from image and control camera movement in 3D environment.
camera circle motion
camera dolly-zoom-in motion
camera swing motion
camera zoom-in motion
depth map
camera circle motion
camera dolly-zoom-in motion
camera swing motion
camera zoom-in motion
Depth Map
camera circle motion
camera dolly-zoom-in motion
camera swing motion
camera zoom-in motion
Depth Map
Creatures of the Night

Stable Diffusion A1111

SDXL model: copaxTimelessxlSDXL1_v42.

Lora model:  Mythical-creatures-lora-13sdxl.

Sampler: DPM++ 2M SDE Karras

Dive into my mythical creatures series, where Stable Diffusion A1111 meets Lora. Witness captivating visuals that bridge the gap between reality and imagination.

Steerable Motion

Steerable motion is a ComfyUI node for batch creative interpolation. 

In this project, my goal is to delve into the fascinating world of motion graphics and animation by experimenting with node-based morphing techniques. By seamlessly transitioning from one image to another, I aim to uncover unexpected and captivating artworks that emerge from blending different artistic styles.

ComfyUI: Steerable Motion Node

Stable Diffusion A1111: are used to generate images combining with pre-trained lora for style and editing in prompt for composing.

Comfy UI prompt styler workflow

In this project, I leverage the SDXL prompt styler and IPadapter to explore style variations, transforming realistic photographs into diverse styles of paintings and drawings

Gen AI open source:  ComfyUI.

models:  SDXLFaetastic V2.4., Ip adapter SDXL Vit-h.

Comfy UI Controlnet Union workflow

In this project, I experiments on SDXL  Controlnet Union to explore style variations, transforming the same subject into diverse styles of paintings and drawings

Gen AI open source:  ComfyUI.

models:  SDXLDreamshaper, Controlnet Union

Text to Gif Workflow in ComfyUI

Animation using interpolation parameters 

SD15 models: Dream Shaper V8.

VAE: BerryMix.

K sampler: Euler Ancestrial.

Upscale model:4x Foolhardy Remacri.

Cosmetic Product Mockup

Use Lora to Generate Cosmetic product mockups in Stable Diffusion A1111

SD15 models: Dream Shaper V8.

Lora: cosmetics_v2.

K sampler:DPM++ SDE Karras.

 

SAME Style - DIFFERENT Subjects -OR- SAME Subject -DIFFERENT Styles

Use loras, prompts, and wildcards to Create series of images in similar style of different subjects or same subject in different styles.

In this project, I explore how consistent styles can be applied to different subjects or how the same subject can be portrayed in various styles. Loras, prompts, and wildcards play a crucial role here.