Sign In
homemodelsimagesvideos3D Modelshubsarticlescomicschallengesupdatesshop

80S BROADCAST SCREENSHOT WORKFLOW

6

rolenzio's Avatar

rolenzio

May 11, 2025

(Updated: 5 months ago)

workflows
vhs effectanalog80s80s style
80S BROADCAST SCREENSHOT WORKFLOW

Creating 1980s TV Show Screenshots with ComfyUI: My Workflow

Introduction

In this article, I’ll share my personal ComfyUI workflow for transforming any image into a realistic 1980s television screenshot, complete with analog textures, VHS glitches, and that nostalgic charm we all love. The foundation is a streamlined image-to-image process that uses two LoRA models—and the best part? The prompt doesn’t have to be written by hand.

🧠 Automatic Prompting with Florence 2

Instead of manually writing a prompt, I use Florence 2, a powerful vision-language model that can analyze the original input image and generate a detailed descriptive prompt automatically. This saves time and ensures the prompt accurately reflects the image content while integrating seamlessly into the vintage aesthetic pipeline.

🛠 The Core of the Workflow: Image-to-Image

The main process uses image2image in ComfyUI. You start with any image—whether a portrait, landscape, or object—and pass it through a stylization pipeline that includes:

  • Florence 2 for prompt generation

  • Two loaded LoRA models that apply retro effects

  • Controlled noise and CFG values to maintain structure but stylize the visual appearance

🎛 The Two LoRA Models I Use

1. My Custom LoRA

I created a LoRA model trained on authentic 1980s broadcast footage. I

📎 Link to my LoRA: [https://civitai.com/models/1554151/80s-broadcast-screenshot-flux-lora]

2. [Author Name]’s LoRA

To enhance the style, I combine my model with the LoRA by [Name], which adds additional broadcast artifacts and period-accurate effects.

📎 Link to dobrosketchkun’s LoRA: [https://civitai.com/models/71125/vhs-footage-with-poor-quality-and-glitches-sd15-flux]

🧩 Activating the LoRA Models

To trigger the style effects correctly, make sure your auto-generated prompt (from Florence 2) includes the keywords:

  • VHS FOOTAGE 80s BROADCASR

These keywords can be manually appended to the Florence-generated prompt if needed, or set by default in your node chain.

🔁 Workflow Steps

  1. Load the source image into the ComfyUI input node.

  2. Use Florence 2 to analyze the image and generate a descriptive prompt.

  3. Append "VHS FOOTAGE, 80s BROADCAST" to ensure both LoRAs are activated.

  4. Run the image2image pass with denoise (around 0.7)

6