ComfyUI Expert for Golf Course Render Enhancements
Budget: €250 – €750 EUR
Description
I need a ComfyUI expert to create a working, repeatable workflow that converts CGI renders of golf courses into photorealistic aerial images. The workflow must preserve geometry (tees, greens, bunkers, houses, roads, water, etc.) exactly, while enhancing realism through textures, lighting, and detail — similar to what Sora/OpenAI demos achieve.
Technical Requirements
ComfyUI-based workflow (not Automatic1111, not standalone Stable Diffusion UI).
Integration of ControlNet Canny + Depth to lock geometry and perspective.
Ability to process single images first, then scale to sequences of images (for video flythroughs).
Must work with SDXL 1.0
Include positive & negative prompt handling to refine realism.
Workflow should output consistent frames (important for video stitching).
Final graph should be saved and documented (clear node layout, settings explained).
Deliverables
A ComfyUI .json workflow file.
Instructions for setup (any required custom nodes, model files, or impact pack extensions).
Example outputs showing before/after (input CGI → enhanced photorealistic).
Environment
Windows 10 workstation
RTX 3090 (24GB VRAM)
ComfyUI portable build with Manager + custom nodes already available
What I’m Looking For
Someone experienced with ControlNet pipelines in ComfyUI.
Ability to balance texture realism without breaking geometry.
Clear documentation so I can re-use the workflow on my own projects.
Bonus Points
Knowledge of video consistency workflows (Deforum / FILM interpolation / Temporal ControlNet).
Experience with architectural or landscape rendering enhancement.
I need a ComfyUI expert to create a working, repeatable workflow that converts CGI renders of golf courses into photorealistic aerial images. The workflow must preserve geometry (tees, greens, bunkers, houses, roads, water, etc.) exactly, while enhancing realism through textures, lighting, and detail — similar to what Sora/OpenAI demos achieve.
Technical Requirements
ComfyUI-based workflow (not Automatic1111, not standalone Stable Diffusion UI).
Integration of ControlNet Canny + Depth to lock geometry and perspective.
Ability to process single images first, then scale to sequences of images (for video flythroughs).
Must work with SDXL 1.0
Include positive & negative prompt handling to refine realism.
Workflow should output consistent frames (important for video stitching).
Final graph should be saved and documented (clear node layout, settings explained).
Deliverables
A ComfyUI .json workflow file.
Instructions for setup (any required custom nodes, model files, or impact pack extensions).
Example outputs showing before/after (input CGI → enhanced photorealistic).
Environment
Windows 10 workstation
RTX 3090 (24GB VRAM)
ComfyUI portable build with Manager + custom nodes already available
What I’m Looking For
Someone experienced with ControlNet pipelines in ComfyUI.
Ability to balance texture realism without breaking geometry.
Clear documentation so I can re-use the workflow on my own projects.
Bonus Points
Knowledge of video consistency workflows (Deforum / FILM interpolation / Temporal ControlNet).
Experience with architectural or landscape rendering enhancement.