A Case Study with Barbara Feldon from Get Smart Using ComfyUI

Case Study: Barbara Feldon with ComfyUI

In this creative project, we explore various capabilities of ComfyUI through the iconic actress Barbara Feldon, best known for her role as Agent 99 in the classic 1965–1970 TV series Get Smart. We utilize advanced tools such as FLUX, Hunyuan3D, and AnimateDiff to process, enhance, and animate Barbara’s visuals. This case study demonstrates a complete AI-assisted creative workflow from image restoration to video generation.

Part One: Image Processing

Original Photographs of Barbara Feldon

Here we display original photos of Barbara Feldon, which serve as the basis for subsequent transformations and enhancements. These images were selected for their clarity and representative facial angles, ideal for training LoRAs and exploring editing workflows.

Barbara Feldon Original 1 Barbara Feldon Original 2 Barbara Feldon Original 3 Barbara Feldon Original 4
Barbara Feldon Original 5 Barbara Feldon Original 6 Barbara Feldon Original 7 Barbara Feldon Original 8
Barbara Feldon Original 9 Barbara Feldon Original 10 Barbara Feldon Original 11 Barbara Feldon Original 12

FLUX and Its Capabilities (2023)

FLUX is a powerful modular toolkit released in mid-2023 by Black Forest Labs (BFL) for ComfyUI. It enables advanced inpainting, outpainting, upscaling, animation, and LoRA training, built atop Stable Diffusion technology.

  • 1.1: Inpainting and Outpainting with FLUX
    Flux Fill uses advanced in/outpainting models...
    Barbara Feldon Inpainting 1 Barbara Feldon Inpainting 2 Barbara Feldon Inpainting 3 Barbara Feldon Inpainting 4
  • 1.2: FLUX Upscaler (2023)
    Enhances resolution and image quality...
    Barbara Feldon Original 3 Upscaled Example
  • 1.3: FLUX LoRA Character Consistency Training
    Uses upscaled photos to train a LoRA for consistent facial identity...
    • Barbara Feldon LoRA 1 Barbara Feldon LoRA 2 Barbara Feldon LoRA 3 Barbara Feldon LoRA 4
    Barbara Feldon LoRA 5 Barbara Feldon Selfie Barbara Feldon LoRA 7 Barbara Feldon LoRA 9
  • 1.4: 3D with Hunyuan3D-2 (2024)
    Converts images to detailed 3D meshes...
    Barbara Feldon 3D

    A two-stage 3D generation system from Tencent, featuring shape generation (Hunyuan3D-DiT) and texture synthesis (Hunyuan3D-Paint).

  • 1.5: FLUX Depth Canny + ToonCrafter (2023)
    Applies depth and edge data for cartoon-style interpolation...
    • Animated Barbara Feldon 1
    Animated Barbara Feldon 2
  • 1.6: CatVTON (2023)
    High-fidelity virtual try-on system merging diffusion with NeRF-like 3D reasoning for realistic garment overlays.

    • Since there's only a single image, no carousel. Let's ensure no forced cropping:

    Barbara Feldon CatVTON

Part Two: Video Generation & Animation

In this phase, we apply generative and animation tools to bring Barbara Feldon to life through video sequences, dance animations, and stylized performances. These tools build on the LoRA and processed images above for consistent identity and artistic fidelity.

  • 2.1: ReActor (2023) — Face Swap Module

    Allows face swapping in videos using real or generated reference portraits...

  • 2.2: Sonic (2023) — Lip-Sync Portrait Animation

    Generates accurate lip movements from audio plus a portrait still...

    Vertical Barbara Feldon
  • 2.3: Hunyuan IP2V — Image-to-Video Engine

    Converts static images into dynamic scenes guided by prompts & style references...

    Vertical Barbara Feldon
  • 2.4: Hunyuan Video (2024) — Video-to-Video Engine

    Tencent’s advanced Video-to-Video solution guided by prompts & LoRA styles, supporting higher resolutions and dynamic 3D backgrounds...