# Sprite DX - Stage 2 - Generate Frames

**SpritedDX**’s vision is to make a tool that will create *animated and working platformer characters* from text prompts.

In yesterday’s [post](https://blog.sprited.app/sprite-dx-stage-1-character-generation), we created Comfy workflow for Stage 1 to generate character reference.

Today, we strive to build Stage 2. Stage 2 will animate the character using video gen AIs.

---

## What I will do

I will create a Comfy workflow that will generate animated frames of run cycle.

---

## Prerequisite

* **Fal.ai**: You need Fal.ai account and need some credits added to your account.
    
    * You can use other providers like *Scenario*, but setup is going to be more manual.
        
* **Comfy UI**: I’m using ComfyUI to create the workflow.
    

---

## Ingredients

Reference image character image generated in [Stage 1](https://blog.sprited.app/sprite-dx-stage-1-character-generation).

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1753820120409/555dc277-8bab-4bda-b282-127102ac5139.png align="center")

---

## Comfy Workflow

First, we use the **ComfyUI Manager**’s **Custom Node Manager** to install [ComfyUI-fal-API](https://github.com/gokayfem/ComfyUI-fal-API).

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1753829836564/9aa0bdf3-e0db-4531-88b6-85316ab65a21.png align="center")

You have to follow the [instructions](https://github.com/gokayfem/ComfyUI-fal-API?tab=readme-ov-file#configuration) to configure it with your Fal AI API keys.

We can then add `Seedance Image-to-video (fal)` node and type in some prompt.

Flow Summary

* `Load Image` node to load the reference image
    
* `Seedance Image-to-Video` node to generate the video from the reference image and a prompt. This outputs a video URL.
    
* Now, use use `Load Video from URL (fal)` node to load the videos into frames.
    
* Finally, we use `Video Combine (VHS)` node to construct a gif or mp4.
    

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1753830155544/c664c6f4-145e-43d4-a0a1-0228216d5144.png align="center")

---

Still need to do some more work to tune the prompts and make the character move in place. Logging off now. I will continue later.

**Update**: Had to read through the Seedance 1.0 Paper to understand the prompt format. I have a pretty good solution here: [https://blog.sprited.app/designing-multi-shot-prompts-for-sprite-animation-generation](https://blog.sprited.app/designing-multi-shot-prompts-for-sprite-animation-generation)

—Sprited Dev 🌱