Alexander Garzon
6d3d2fca46
BYOK OpenAI gpt-image generation and editing nodes for ComfyUI — calls go directly to the OpenAI API with your own key (no proxy, no credit markup). - Two nodes: "Imagent: OpenAI Image" (text-to-image) and "Imagent: OpenAI Image Edit" (edit / inpaint / multi-reference compose). - Models gpt-image-2 / gpt-image-1.5 / gpt-image-1 with a per-model dynamic UI (comfy_api DynamicCombo): size, background, custom WxH, and input_fidelity are revealed only for the models that support them. - Edit node accepts an auto-growing set of reference images (up to 16) plus an optional inpaint mask; generate supports batches up to 8. - API key via OPENAI_API_KEY or a local config.json. Requires ComfyUI >= 0.23.0.
89 lines
4.5 KiB
Python
89 lines
4.5 KiB
Python
"""ComfyUI node: image edit / inpaint via OpenAI images.edit.
|
|
|
|
Like node_generate, `model` is a DynamicCombo with a nested size combo. Reference
|
|
images use an auto-growing input (up to 16). Behavior lives in build.run_edit().
|
|
"""
|
|
from __future__ import annotations
|
|
|
|
from comfy_api.latest import IO
|
|
|
|
from . import build
|
|
from . import params as params_mod
|
|
from .node_generate import _gpt_image_2_inputs, unpack_size
|
|
|
|
_MAX_REFS = 16
|
|
|
|
|
|
def _legacy_edit_inputs():
|
|
"""gpt-image-1 / gpt-image-1.5: base sizes, transparent allowed, input_fidelity."""
|
|
return [
|
|
IO.Combo.Input("size", options=params_mod.SIZES_BASE, default="auto",
|
|
tooltip="Output size (base sizes supported by gpt-image-1.x)."),
|
|
IO.Combo.Input("background", options=params_mod.BACKGROUNDS, default="auto",
|
|
tooltip="Background handling. 'transparent' needs png or webp output."),
|
|
IO.Combo.Input("input_fidelity", options=params_mod.INPUT_FIDELITIES, default="high",
|
|
tooltip="How closely to preserve the reference image: 'high' "
|
|
"(default) or 'low'."),
|
|
]
|
|
|
|
|
|
class OpenAIImageEdit(IO.ComfyNode):
|
|
@classmethod
|
|
def define_schema(cls):
|
|
return IO.Schema(
|
|
node_id="ImagentOpenAIImageEdit",
|
|
display_name="🤖 Imagent: OpenAI Image Edit",
|
|
category="Imagent",
|
|
description="BYOK image edit / inpaint / multi-reference compose via OpenAI "
|
|
"gpt-image models (direct API, no proxy).",
|
|
inputs=[
|
|
IO.String.Input("prompt", multiline=True, default="",
|
|
placeholder="Describe the edit to apply…",
|
|
tooltip="Describe the edit to apply to the reference image(s)."),
|
|
IO.DynamicCombo.Input(
|
|
"model",
|
|
tooltip="OpenAI gpt-image model. Switching this changes the options below.",
|
|
options=[
|
|
IO.DynamicCombo.Option("gpt-image-2", _gpt_image_2_inputs()),
|
|
IO.DynamicCombo.Option("gpt-image-1.5", _legacy_edit_inputs()),
|
|
IO.DynamicCombo.Option("gpt-image-1", _legacy_edit_inputs()),
|
|
],
|
|
),
|
|
IO.Autogrow.Input(
|
|
"images",
|
|
template=IO.Autogrow.TemplateNames(
|
|
IO.Image.Input("image"),
|
|
names=[f"image_{i}" for i in range(1, _MAX_REFS + 1)],
|
|
min=1),
|
|
tooltip=f"Reference image(s) to edit; grows up to {_MAX_REFS}. A mask "
|
|
"(if used) applies to the first image and requires exactly one image."),
|
|
IO.Combo.Input("quality", options=params_mod.QUALITIES, default="auto",
|
|
tooltip="Rendering quality. 'auto' lets the model decide; "
|
|
"higher quality costs more."),
|
|
IO.Combo.Input("output_format", options=params_mod.FORMATS, default="png",
|
|
tooltip="Image file format returned by the API."),
|
|
IO.Combo.Input("moderation", options=params_mod.MODERATIONS, default="auto",
|
|
tooltip="Content moderation strictness. 'auto' (default) or "
|
|
"'low'. Sent only when not 'auto'."),
|
|
IO.Int.Input("n", default=1, min=params_mod.N_MIN, max=params_mod.N_MAX, step=1,
|
|
tooltip="How many images to generate (1-8)."),
|
|
IO.Mask.Input("mask", optional=True,
|
|
tooltip="Inpaint mask: white marks the region to edit. Requires a "
|
|
"single reference image."),
|
|
],
|
|
outputs=[IO.Image.Output(display_name="image")],
|
|
)
|
|
|
|
@classmethod
|
|
def execute(cls, prompt, model, images, quality, output_format, moderation, n,
|
|
mask=None) -> IO.NodeOutput:
|
|
size, custom_width, custom_height = unpack_size(model)
|
|
ref_tensors = [t for t in (images or {}).values() if t is not None]
|
|
img, _info = build.run_edit(
|
|
prompt=prompt, model=model["model"], size=size, quality=quality,
|
|
background=model["background"], output_format=output_format, n=n,
|
|
images=ref_tensors, mask=mask,
|
|
input_fidelity=model.get("input_fidelity", "high"), moderation=moderation,
|
|
custom_width=custom_width, custom_height=custom_height)
|
|
return IO.NodeOutput(img)
|