AI Video Generation Workflow

Project Overview
We need a ComfyUI workflow that can generate high-quality 720p videos in three modes:

Image to Video

Input: 1–5 static images

Output: Smooth 3–10 sec video at 24 fps

Include frame interpolation (e.g. RIFE via LTXVideo), stabilization, and subject consistency

Text to Video

Input: Text prompt

Output: 5–10 sec video at 24 fps matching the prompt

Subject-to-Reference

Input: Reference image of a person or object + context prompt

Output: 5–10 sec video of the subject in a new scene, with consistent appearance

Project Scope (Phase 1)
Workflow Design
Create ComfyUI node graph for all three video use cases

Use leading open-source video models:

LTXVideo (supports text‑to‑video, image‑to‑video)
reddit.com
+15
comfyui-wiki.com
+15
comfyui.org
+15
blog.comfy.org
+15
github.com
+15
reddit.com
+15
mimicpc.com
+2
kombitz.com
+2
reddit.com
+2
reddit.com
+2
blog.comfy.org
+2
arxiv.org
+2
arxiv.org
reddit.com
+1
reddit.com
+1

Wan2.1 for stabilization and interpolation
comfyui.org
+9
comfyui.org
+9
reddit.com
+9

RIFE frame interpolation node for smoother frame transitions

Hardware & Runtime
Target GPU: 12–24 GB VRAM (e.g., RTX 3060/4070/4090)

Performance benchmarks:

~10 sec of video generated in under 5 min

Frame interpolation completes in under 30 sec per video

Deliverables
ComfyUI workflow JSON files covering all modes

Documentation including:

Model sources with Hugging Face links

GPU VRAM usage and runtime benchmarks per 5-sec video

Node graph descriptions and prompt-engineering tips

Estimated Time (Phase 1)
Workflow design and construction: 2–3 days

Performance testing and optimization: 1–2 days

Documentation and handover: 1 day
Total: ~1 week

Required Skills
Deep experience with ComfyUI, LTXVideo, Wan2.1, and RIFE/RIFE2 frame interpolation
reddit.com
+15
comfyui.org
+15
github.com
+15
reddit.com
+14
kombitz.com
+14
github.com
+14
blog.comfy.org

Strong prompt-engineering capabilities for video continuity

Proven experience optimizing GPU performance

Portfolio with published ComfyUI video workflow examples

Budget & Timeline
Goal: Complete Phase 1 within 1 week

Budget range: USD $500–$1,000, depending on experience and deliverables

How to Apply
Please provide:

Relevant portfolio links for ComfyUI video workflows

Your GPU hardware specs

Proposed timeline

Cost estimate for initial phase

We look forward to working with an expert who can deliver smooth, realistic AI-driven video workflows using ComfyUI.