Skip to content

Latest commit

 

History

2 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

The Forge

Production layer on top of ComfyUI. Image, video, audio.

Built for game art, pre-vis, graphic novels, and sound. You design a workflow in the node graph once. You run ten thousand of them from a script.

workflows/*.json      pipelines, exported as API format
presets/*.json        house style — prompt fragments and settings
jobs/*.jsonl          the work, one line per output
forge.py              submit, poll, retrieve, name, file

A job file can be reviewed, diffed, and written by another program. The expensive part of a graphic novel is not one render. It is a thousand consistent ones.

Install

From the ComfyUI terminal (Comfy Desktop → >_ in the bottom-left):

powershell -ExecutionPolicy Bypass -File "$env:USERPROFILE\Downloads\gf-forge\bootstrap.ps1"

Run it again any time. It resumes interrupted downloads and skips what is already there.

.\bootstrap.ps1 -Skip video          # image + audio only
.\bootstrap.ps1 -Only audio          # sound stack only
.\bootstrap.ps1 -DryRun              # print what it would fetch, and how big

Models

Image

model licence why
FLUX.1-schnell fp8 Apache-2.0 Default for concept art and covers. Four steps, no CFG. Fits 12 GB.
SDXL base 1.0 CreativeML Open RAIL++ Workhorse. ControlNet and IP-Adapter ecosystem — consistency.
ControlNet Union SDXL Apache-2.0 Depth, canny, lineart, openpose, scribble, tile. Pose control for the same character in a new panel.
IP-Adapter Plus SDXL + CLIP-ViT-H Apache-2.0 Style from reference images. Point it at existing Ghost Front plates.
4x-UltraSharp + RealESRGAN x4 GPL / BSD Tile-upscale a 1024 plate to 4096.
SAM + GroundingDINO Apache-2.0 "Cut the figure out of every cell of this sheet."

Video

model licence why
Wan 2.1 T2V 1.3B Apache-2.0 Open text-to-video that fits 12 GB.
Wan 2.1 I2V (14B, offloaded) Apache-2.0 Take a finished panel and move it.
SVD XT 1.1 Stability Community Short I2V with decent camera coherence.
AnimateDiff v3 + motion LoRAs Apache-2.0 Loops, directed camera motion.
RIFE + FILM MIT / Apache-2.0 8 fps → 24/60. Generate few frames, interpolate the rest.

Audio

model licence why
ACE-Step v1 3.5B Apache-2.0 Music with lyrics and structure. Scored cues, not just loops.
Stable Audio Open 1.0 Stability Community Trained on Freesound. SFX first.
MMAudio MIT Video-to-audio / foley on the cuts.
MusicGen Melody CC-BY-NC (check before commercial use) Melody-conditioned.

Custom nodes: ComfyUI-Manager, VideoHelperSuite, KJNodes, Impact Pack, rgthree-comfy, ComfyUI_essentials, Frame Interpolation, ControlNet Aux, IPAdapter Plus, Segment Anything, ACE-Step, MMAudio.

Use

python forge.py run workflows/txt2img_flux.json \
    --set prompt="a ruined Normandy farmhouse at night, painted plate, 16:6" \
    --out art/plates/

python forge.py batch jobs/chapter_plates.jsonl --out art/plates/

python forge.py sheet --ref art/hawke_ref.png --poses poses/walk8.json \
    --out art/sheets/hawke_walk/

python forge.py audio workflows/music_ace.json \
    --set prompt="slow funeral march, solo horn, muffled timpani, 1944, 52 bpm" \
    --set seconds=90 --out audio/cues/

python forge.py audio workflows/sfx_stable.json \
    --set prompt="M1911 pistol shot, close, outdoor, night, no reverb tail" \
    --n 10 --out audio/sfx/m1911/

python forge.py video workflows/i2v_wan.json \
    --set image=art/panels/p014.png \
    --set prompt="slow push in, ash falling, the figure does not move" \
    --out film/panels/

forge.py queues over HTTP, retries, manages seeds, writes deterministic filenames and a sidecar .json per output. --contact builds a contact sheet for a batch.

House style

presets/ghostfront.json is the look: flat three-value gouache cel, above-right key, indigo shadows, #2A2A2E ground, no outlines, division colours (oxblood FLESH, gunmetal-and-brass STAHL, radium green CHEMIE, arc-light blue GLOCKE), sheet geometry the cutter expects — 160×160 cells, 8 columns, feet on y=148, mass on x=80.

Add your own under presets/.

About

Production platform on top of ComfyUI - image, video and audio driven from scripts, not the node graph.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages