dreevelleandClaude Opus 5
03ca33cb78
The floor sat at 0.25 MP, so 16:9's cheapest entry was 768x432 — and that one is K2-only (432 % 32 = 16), leaving 1024x576 as the smallest H3-legal 16:9 in the list. An audio-only H3 render wants 512x288, the n=1 lattice point at alignment 32, where the video canvas is pure overhead. Dropping PRESET_MIN_MP alone re-picks the geometric thinning and silently drops labels — 3:2 1152x768 and 1920x1280 among them — which breaks saved workflows, since the label string is what a graph stores. Raising PRESET_MAX_ENTRIES to 11 in the same step absorbs that: 79 -> 87 entries with every existing label still present and still pointing at the same size. The six entries beyond 512x288 / 288x512 are lattice points the old 9-entry cap had been skipping over. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01K1sGVWw5KU3GD7CrhdZR7M
425 lines
17 KiB
Python
425 lines
17 KiB
Python
"""ComfyUI-ResolutionTools.
|
||
|
||
Resolution maths for latent-space models, done exactly.
|
||
|
||
Three nodes:
|
||
|
||
``ResolutionPreset``
|
||
The everyday one. A single list of known-good resolutions, each exactly on
|
||
its aspect ratio and on the model's patch grid, tagged with the model
|
||
families it is legal for. Generated from the lattice at import time, so it
|
||
cannot drift from what the solver below produces.
|
||
|
||
``ResolutionSelectorMP``
|
||
The parametric one, for custom ratios and arbitrary budgets. Aspect ratio +
|
||
a *real* megapixel target (10^6 pixels, not 1024^2) -> width/height that
|
||
land on the model's patch grid **without drifting the aspect ratio**.
|
||
Rounding each axis independently -- what the built-in Resolution Selector
|
||
does -- turns 9:16 into 0.5649 instead of 0.5625. This solves on the
|
||
lattice of grid-aligned exact-ratio sizes instead.
|
||
|
||
``ResolutionAlignToGrid``
|
||
Snap an arbitrary width/height (e.g. 1920x1080) to the same grid
|
||
(-> 1920x1088), which is what stops the bottom-edge banding described below.
|
||
|
||
Why alignment matters
|
||
---------------------
|
||
A latent-space DiT reaches pixels through two reductions::
|
||
|
||
pixels --/ VAE spatial downscale --> latent --/ DiT patch size --> tokens
|
||
|
||
so both axes must be multiples of ``vae_downscale * patch_size``:
|
||
|
||
========================================== =========
|
||
Krea 2 / Flux / SD3 / Qwen-Image / Wan 16 (8 x 2)
|
||
MiniMax H3 32 (16 x 2)
|
||
SDXL / SD1.5 (UNet, 3 downsample stages) 64
|
||
========================================== =========
|
||
|
||
Miss it and ComfyUI's ``comfy.ldm.common_dit.pad_to_patch_size`` pads the latent
|
||
up to the patch grid with **circular** padding -- the pad row is a wrapped copy
|
||
of the top of the image. That fake row gets folded into the same patch token as
|
||
the real bottom row, which the model never saw in training, so the final latent
|
||
row decodes to a band of garbage at the bottom edge. One latent row = 8 px on
|
||
Krea 2, 16 px on H3. Padding is only ever appended, so the artifact is always
|
||
bottom/right, never top/left.
|
||
|
||
Self-contained on purpose (nothing imported from ``comfy_extras``) so a ComfyUI
|
||
core update can't silently change the maths.
|
||
"""
|
||
|
||
import math
|
||
from enum import Enum
|
||
|
||
from typing_extensions import override
|
||
|
||
from comfy_api.latest import ComfyExtension, IO
|
||
|
||
__version__ = "1.2.0"
|
||
|
||
|
||
# ---------------------------------------------------------------------------
|
||
# Aspect ratios. Values are (w, h) integer pairs, reduced before use.
|
||
# ---------------------------------------------------------------------------
|
||
class AspectRatio(str, Enum):
|
||
SQUARE = "1:1 (Square)"
|
||
PHOTO_V = "2:3 (Portrait Photo)"
|
||
PHOTO_H = "3:2 (Photo)"
|
||
STANDARD_V = "3:4 (Portrait Standard)"
|
||
STANDARD_H = "4:3 (Standard)"
|
||
SOCIAL_V = "4:5 (Portrait Social)"
|
||
SOCIAL_H = "5:4 (Social)"
|
||
WIDESCREEN_V = "9:16 (Portrait Widescreen)"
|
||
WIDESCREEN_H = "16:9 (Widescreen)"
|
||
ULTRAWIDE_V = "9:21 (Portrait Ultrawide)"
|
||
ULTRAWIDE_H = "21:9 (Ultrawide)"
|
||
CUSTOM = "Custom (use custom_w / custom_h)"
|
||
|
||
|
||
ASPECT_RATIOS: dict[AspectRatio, tuple[int, int]] = {
|
||
AspectRatio.SQUARE: (1, 1),
|
||
AspectRatio.PHOTO_V: (2, 3),
|
||
AspectRatio.PHOTO_H: (3, 2),
|
||
AspectRatio.STANDARD_V: (3, 4),
|
||
AspectRatio.STANDARD_H: (4, 3),
|
||
AspectRatio.SOCIAL_V: (4, 5),
|
||
AspectRatio.SOCIAL_H: (5, 4),
|
||
AspectRatio.WIDESCREEN_V: (9, 16),
|
||
AspectRatio.WIDESCREEN_H: (16, 9),
|
||
AspectRatio.ULTRAWIDE_V: (9, 21),
|
||
AspectRatio.ULTRAWIDE_H: (21, 9),
|
||
}
|
||
|
||
|
||
ALIGNMENT_TOOLTIP = (
|
||
"Pixel grid both axes must land on = VAE spatial downscale x DiT patch size. "
|
||
"16: Krea 2 / Flux / SD3 / Qwen-Image / Wan. "
|
||
"32: MiniMax H3. "
|
||
"64: SDXL / SD1.5, and universally safe. "
|
||
"Getting this wrong causes a band of artifacts along the bottom/right edge."
|
||
)
|
||
|
||
|
||
# ---------------------------------------------------------------------------
|
||
# Core maths
|
||
# ---------------------------------------------------------------------------
|
||
def lattice_step(w_ratio: int, h_ratio: int, alignment: int) -> int:
|
||
"""Smallest scale increment keeping an exact ratio on the alignment grid.
|
||
|
||
For a reduced ratio a:b, sizes are (a*s, b*s). ``a*s`` is a multiple of
|
||
``m`` only when ``s`` is a multiple of ``m // gcd(a, m)``; same for ``b``.
|
||
Both hold together on multiples of the lcm of those two, so the valid
|
||
resolutions are exactly ``(a*L*n, b*L*n)`` for integer n >= 1.
|
||
|
||
9:16 at m=16 -> L=16 -> 144n x 256n (n=8 gives 1152x2048)
|
||
16:9 at m=32 -> L=32 -> 512n x 288n (n=3 gives 1536x864)
|
||
"""
|
||
return math.lcm(alignment // math.gcd(w_ratio, alignment),
|
||
alignment // math.gcd(h_ratio, alignment))
|
||
|
||
|
||
def solve_exact(w_ratio: int, h_ratio: int, alignment: int, target_pixels: float) -> tuple[int, int]:
|
||
"""Grid-aligned size with the ratio held exactly, area nearest the target."""
|
||
step = lattice_step(w_ratio, h_ratio, alignment)
|
||
# area(n) = w_ratio*h_ratio*step^2 * n^2, so solve for n and round.
|
||
n = max(1, round(math.sqrt(target_pixels / (w_ratio * h_ratio * step * step))))
|
||
return w_ratio * step * n, h_ratio * step * n
|
||
|
||
|
||
def solve_nearest(w_ratio: int, h_ratio: int, alignment: int, target_pixels: float) -> tuple[int, int]:
|
||
"""Grid-aligned size with each axis rounded independently.
|
||
|
||
Hits the area target more closely than :func:`solve_exact` at the cost of a
|
||
small aspect-ratio drift. Useful when the exact lattice is coarse.
|
||
"""
|
||
scale = math.sqrt(target_pixels / (w_ratio * h_ratio))
|
||
width = max(alignment, round(w_ratio * scale / alignment) * alignment)
|
||
height = max(alignment, round(h_ratio * scale / alignment) * alignment)
|
||
return width, height
|
||
|
||
|
||
def align_value(value: int, alignment: int, mode: str) -> int:
|
||
"""Snap one dimension onto the grid, never returning less than ``alignment``."""
|
||
if mode == "up":
|
||
snapped = math.ceil(value / alignment) * alignment
|
||
elif mode == "down":
|
||
snapped = math.floor(value / alignment) * alignment
|
||
else:
|
||
snapped = round(value / alignment) * alignment
|
||
return max(alignment, snapped)
|
||
|
||
|
||
# ---------------------------------------------------------------------------
|
||
# Preset generation
|
||
#
|
||
# With exact_ratio on, the valid outputs are a discrete lattice, so a
|
||
# continuous megapixel input advertises precision that does not exist. These
|
||
# presets are the lattice, rendered readable -- derived at import time, never
|
||
# hand-maintained, so they cannot drift from what the solver produces.
|
||
# ---------------------------------------------------------------------------
|
||
# The floor sits just under 512x288 (0.147 MP), the n=1 lattice point for 16:9
|
||
# at alignment 32 -- the smallest exact 16:9 an H3 graph can ask for. Useful for
|
||
# audio-only H3 runs, where the video canvas is pure overhead.
|
||
PRESET_MIN_MP = 0.14
|
||
PRESET_MAX_MP = 4.3
|
||
PRESET_MIN_ENTRIES = 6 # coarsest grid that still offers this many sizes wins
|
||
PRESET_MAX_ENTRIES = 11 # thin geometrically beyond this, keeping both ends
|
||
|
||
# 16:9 and 9:16 lead: the common case for video and for delivery to a screen.
|
||
PRESET_RATIO_ORDER = [
|
||
AspectRatio.WIDESCREEN_H, AspectRatio.WIDESCREEN_V,
|
||
AspectRatio.SQUARE,
|
||
AspectRatio.PHOTO_H, AspectRatio.PHOTO_V,
|
||
AspectRatio.STANDARD_H, AspectRatio.STANDARD_V,
|
||
AspectRatio.SOCIAL_H, AspectRatio.SOCIAL_V,
|
||
AspectRatio.ULTRAWIDE_H, AspectRatio.ULTRAWIDE_V,
|
||
]
|
||
|
||
|
||
def compat_tag(width: int, height: int) -> str:
|
||
"""Which model families a resolution is legal for, from its divisibility."""
|
||
if width % 64 == 0 and height % 64 == 0:
|
||
return "K2+H3+SDXL"
|
||
if width % 32 == 0 and height % 32 == 0:
|
||
return "K2+H3"
|
||
return "K2"
|
||
|
||
|
||
def _lattice_points(w_ratio, h_ratio, alignment):
|
||
step = lattice_step(w_ratio, h_ratio, alignment)
|
||
out, n = [], 1
|
||
while True:
|
||
w, h = w_ratio * step * n, h_ratio * step * n
|
||
mp = w * h / 1_000_000
|
||
if mp > PRESET_MAX_MP:
|
||
return out
|
||
if mp >= PRESET_MIN_MP:
|
||
out.append((w, h, mp))
|
||
n += 1
|
||
|
||
|
||
def build_presets() -> dict[str, tuple[int, int]]:
|
||
"""Readable label -> (width, height), covering every aspect ratio.
|
||
|
||
Per ratio, take the *coarsest* grid that still yields PRESET_MIN_ENTRIES
|
||
sizes. Fine-lattice ratios (1:1 has L=16, so 16n x 16n) would otherwise
|
||
flood the list with a hundred near-identical entries; the coarse grid
|
||
lands them on canonical sizes instead (512, 768, 1024, ...). Coarse-lattice
|
||
ratios like 16:9 fall through to grid 16 and keep their full useful set.
|
||
"""
|
||
presets = {}
|
||
for ratio in PRESET_RATIO_ORDER:
|
||
w_ratio, h_ratio = ASPECT_RATIOS[ratio]
|
||
divisor = math.gcd(w_ratio, h_ratio)
|
||
a, b = w_ratio // divisor, h_ratio // divisor
|
||
|
||
points = []
|
||
for alignment in (256, 128, 64, 32, 16):
|
||
points = _lattice_points(a, b, alignment)
|
||
if len(points) >= PRESET_MIN_ENTRIES:
|
||
break
|
||
|
||
if len(points) > PRESET_MAX_ENTRIES:
|
||
span = len(points) - 1
|
||
keep = {round(i * span / (PRESET_MAX_ENTRIES - 1)) for i in range(PRESET_MAX_ENTRIES)}
|
||
points = [points[i] for i in sorted(keep)]
|
||
|
||
# label with the declared ratio (21:9), not the reduced pair (7:3)
|
||
name = ratio.value.split(" ")[0]
|
||
for w, h, mp in points:
|
||
presets[f"{name} · {w} × {h} · {mp:.2f} MP · {compat_tag(w, h)}"] = (w, h)
|
||
return presets
|
||
|
||
|
||
PRESETS = build_presets()
|
||
PRESET_LABELS = list(PRESETS)
|
||
DEFAULT_PRESET = next(k for k, v in PRESETS.items() if v == (2048, 1152))
|
||
|
||
|
||
# ---------------------------------------------------------------------------
|
||
# Nodes
|
||
# ---------------------------------------------------------------------------
|
||
class ResolutionSelectorMP(IO.ComfyNode):
|
||
"""Aspect ratio + real megapixels -> grid-aligned width/height."""
|
||
|
||
@classmethod
|
||
def define_schema(cls):
|
||
return IO.Schema(
|
||
node_id="ResolutionSelectorMP",
|
||
display_name="Resolution Selector (Real MP)",
|
||
category="utilities",
|
||
description=(
|
||
"Width/height from an aspect ratio and a real megapixel target "
|
||
"(1 MP = 1,000,000 pixels), snapped to the model's patch grid "
|
||
"with the aspect ratio held exactly."
|
||
),
|
||
inputs=[
|
||
IO.Combo.Input(
|
||
"aspect_ratio",
|
||
options=AspectRatio,
|
||
default=AspectRatio.WIDESCREEN_H,
|
||
tooltip="Aspect ratio of the output. Portrait entries are listed before their landscape counterparts.",
|
||
),
|
||
IO.Float.Input(
|
||
"megapixels",
|
||
default=1.0,
|
||
min=0.01,
|
||
max=64.0,
|
||
step=0.01,
|
||
tooltip=(
|
||
"Target area in real megapixels: 1.0 = 1,000,000 pixels. "
|
||
"Note this is NOT the built-in Resolution Selector's unit, which is 1024x1024 = 1.049 MP."
|
||
),
|
||
),
|
||
IO.Int.Input(
|
||
"alignment",
|
||
default=16,
|
||
min=1,
|
||
max=256,
|
||
step=8,
|
||
tooltip=ALIGNMENT_TOOLTIP,
|
||
),
|
||
IO.Boolean.Input(
|
||
"exact_ratio",
|
||
default=True,
|
||
tooltip=(
|
||
"On: hold the aspect ratio exactly by snapping to the lattice of valid sizes, "
|
||
"accepting a slightly different area. Off: round each axis independently, "
|
||
"hitting the area target more closely but drifting the ratio."
|
||
),
|
||
),
|
||
IO.Int.Input(
|
||
"custom_w",
|
||
default=16,
|
||
min=1,
|
||
max=4096,
|
||
tooltip="Width term of the ratio, used only when aspect_ratio is Custom.",
|
||
advanced=True,
|
||
),
|
||
IO.Int.Input(
|
||
"custom_h",
|
||
default=9,
|
||
min=1,
|
||
max=4096,
|
||
tooltip="Height term of the ratio, used only when aspect_ratio is Custom.",
|
||
advanced=True,
|
||
),
|
||
],
|
||
outputs=[
|
||
IO.Int.Output(display_name="width", tooltip="Width in pixels, a multiple of alignment."),
|
||
IO.Int.Output(display_name="height", tooltip="Height in pixels, a multiple of alignment."),
|
||
IO.Float.Output(display_name="megapixels", tooltip="Real megapixels actually produced."),
|
||
IO.String.Output(display_name="label", tooltip='Dimensions as "1152x2048", handy for filename prefixes.'),
|
||
],
|
||
)
|
||
|
||
@classmethod
|
||
def execute(cls, aspect_ratio: str, megapixels: float, alignment: int,
|
||
exact_ratio: bool, custom_w: int, custom_h: int) -> IO.NodeOutput:
|
||
if aspect_ratio == AspectRatio.CUSTOM:
|
||
w_ratio, h_ratio = custom_w, custom_h
|
||
else:
|
||
w_ratio, h_ratio = ASPECT_RATIOS[AspectRatio(aspect_ratio)]
|
||
|
||
# Reduce, so 16:9 and 32:18 give the same lattice.
|
||
divisor = math.gcd(w_ratio, h_ratio)
|
||
w_ratio, h_ratio = w_ratio // divisor, h_ratio // divisor
|
||
|
||
target_pixels = megapixels * 1_000_000
|
||
solve = solve_exact if exact_ratio else solve_nearest
|
||
width, height = solve(w_ratio, h_ratio, alignment, target_pixels)
|
||
|
||
return IO.NodeOutput(width, height, (width * height) / 1_000_000, f"{width}x{height}")
|
||
|
||
|
||
class ResolutionPreset(IO.ComfyNode):
|
||
"""Pick a known-good resolution from a list instead of computing one."""
|
||
|
||
@classmethod
|
||
def define_schema(cls):
|
||
return IO.Schema(
|
||
node_id="ResolutionPreset",
|
||
display_name="Resolution Preset",
|
||
category="utilities",
|
||
description=(
|
||
"Known-good resolutions, picked from a list. Every entry is exactly "
|
||
"on its aspect ratio and on the model's patch grid. The tag says which "
|
||
"models it is legal for: K2 = Krea 2 / Flux / SD3 / Qwen-Image / Wan "
|
||
"(grid 16), K2+H3 adds MiniMax H3 (grid 32), K2+H3+SDXL adds the UNet "
|
||
"models (grid 64)."
|
||
),
|
||
inputs=[
|
||
IO.Combo.Input(
|
||
"preset",
|
||
options=PRESET_LABELS,
|
||
default=DEFAULT_PRESET,
|
||
tooltip=(
|
||
"ratio · width × height · real megapixels · compatible models. "
|
||
"For a ratio or size not listed, use Resolution Selector (Real MP)."
|
||
),
|
||
),
|
||
],
|
||
outputs=[
|
||
IO.Int.Output(display_name="width"),
|
||
IO.Int.Output(display_name="height"),
|
||
IO.Float.Output(display_name="megapixels", tooltip="Real megapixels (10^6 pixels)."),
|
||
IO.String.Output(display_name="label", tooltip='Dimensions as "2048x1152".'),
|
||
],
|
||
)
|
||
|
||
@classmethod
|
||
def execute(cls, preset: str) -> IO.NodeOutput:
|
||
width, height = PRESETS[preset]
|
||
return IO.NodeOutput(width, height, (width * height) / 1_000_000, f"{width}x{height}")
|
||
|
||
|
||
class ResolutionAlignToGrid(IO.ComfyNode):
|
||
"""Snap an arbitrary width/height onto the model's patch grid."""
|
||
|
||
@classmethod
|
||
def define_schema(cls):
|
||
return IO.Schema(
|
||
node_id="ResolutionAlignToGrid",
|
||
display_name="Align Resolution to Grid",
|
||
category="utilities",
|
||
description=(
|
||
"Snap width/height to a multiple of the model's patch grid, e.g. "
|
||
"1920x1080 -> 1920x1088 for Krea 2. Unaligned dimensions get "
|
||
"circular-padded in the DiT and band along the bottom/right edge."
|
||
),
|
||
inputs=[
|
||
IO.Int.Input("width", default=1920, min=1, max=16384),
|
||
IO.Int.Input("height", default=1080, min=1, max=16384),
|
||
IO.Int.Input("alignment", default=16, min=1, max=256, step=8, tooltip=ALIGNMENT_TOOLTIP),
|
||
IO.Combo.Input(
|
||
"mode",
|
||
options=["round", "up", "down"],
|
||
default="round",
|
||
tooltip="round: nearest multiple. up: never shrink (may add letterbox). down: never grow (may crop framing).",
|
||
),
|
||
],
|
||
outputs=[
|
||
IO.Int.Output(display_name="width"),
|
||
IO.Int.Output(display_name="height"),
|
||
IO.Boolean.Output(display_name="changed", tooltip="True if either dimension was off the grid."),
|
||
IO.String.Output(display_name="label"),
|
||
],
|
||
)
|
||
|
||
@classmethod
|
||
def execute(cls, width: int, height: int, alignment: int, mode: str) -> IO.NodeOutput:
|
||
new_w = align_value(width, alignment, mode)
|
||
new_h = align_value(height, alignment, mode)
|
||
changed = (new_w != width) or (new_h != height)
|
||
return IO.NodeOutput(new_w, new_h, changed, f"{new_w}x{new_h}")
|
||
|
||
|
||
class ResolutionToolsExtension(ComfyExtension):
|
||
@override
|
||
async def get_node_list(self) -> list[type[IO.ComfyNode]]:
|
||
return [ResolutionPreset, ResolutionSelectorMP, ResolutionAlignToGrid]
|
||
|
||
|
||
async def comfy_entrypoint() -> ResolutionToolsExtension:
|
||
return ResolutionToolsExtension()
|