Budi HartonoandClaude Sonnet 5
17f6fba87b
Root-cause fix for the reported slowdown/wrong-resolution issue: Flux.2 uses 128 channels at 16x spatial downsample (comfy_extras/nodes_flux.py's EmptyFlux2LatentImage), not 16ch/8x like Flux.1/Krea/Qwen-Image. We hardcoded //8 for every model, so Flux.2 latents came out 2x too large per axis (4x the pixels) — explains both the wrong output resolution and the multi-x s/it gap versus the native node. Replaced the flat channel map with a per-model (channels, stride) spec, validated against every model in presets.py. By Axis previously had no model awareness at all — always assumed 4ch/8-stride regardless of target model, silently wrong for every 16/128-channel model. Gave it a model input (now last: primary_dim, reference, aspect_ratio, batch_size, model), wired to the same spec table. Both nodes now derive their model list/default from presets.py's own order at module level instead of duplicating it — whichever model is listed first in presets.py becomes the default everywhere, so a future architecture doesn't need code changes elsewhere to become the default. Also: rebuilt nodes.py/__init__.py/web extension from scratch (same node IDs, same behavior — verified via smoke test) as a clean base for the above; fixed two Flux.2 presets (1080 -> 1088) that weren't multiples of its 16-stride; added a 512x512 preset to every model for fast local iteration; documented the By Axis model selector in README. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
76 lines
3.1 KiB
Python
76 lines
3.1 KiB
Python
# presets.py
|
|
# Grouped most -> least relevant in 2026 usage (Krea/Flux.2/Qwen-Image lead current mindshare;
|
|
# SDXL/SD1.5 kept for their LoRA ecosystems but pushed to the bottom), then within a model
|
|
# square -> landscape -> portrait, widest ratio first. Dims kept multiples of 8 for
|
|
# latent-space compatibility.
|
|
|
|
PRESETS = [
|
|
# model, label, w, h
|
|
|
|
# --- Flux Krea (dev, photoreal-tuned, shares Flux.1 native grid) ---
|
|
("Krea", "1:1 Square", 512, 512),
|
|
("Krea", "1:1 Square", 1024, 1024),
|
|
("Krea", "3:2 Landscape", 1216, 832),
|
|
("Krea", "4:3 Landscape", 1152, 896),
|
|
("Krea", "16:9 Landscape", 1344, 768),
|
|
("Krea", "2:3 Portrait", 832, 1216),
|
|
("Krea", "3:4 Portrait", 896, 1152),
|
|
("Krea", "9:16 Portrait", 768, 1344),
|
|
|
|
# --- Flux.2 (dev/pro/flex, higher native ceiling, up to ~4MP) ---
|
|
("Flux.2", "1:1 Square", 512, 512),
|
|
("Flux.2", "1:1 Square", 2048, 2048),
|
|
("Flux.2", "3:2 Landscape", 1728, 1152),
|
|
("Flux.2", "4:3 Landscape", 1664, 1248),
|
|
("Flux.2", "16:9 Landscape", 1920, 1088),
|
|
("Flux.2", "21:9 Landscape", 2176, 960),
|
|
("Flux.2", "2:3 Portrait", 1152, 1728),
|
|
("Flux.2", "3:4 Portrait", 1248, 1664),
|
|
("Flux.2", "9:16 Portrait", 1088, 1920),
|
|
("Flux.2", "9:21 Portrait", 960, 2176),
|
|
|
|
# --- Qwen-Image (native ~1.7MP, wide native AR set) ---
|
|
("Qwen-Image", "1:1 Square", 512, 512),
|
|
("Qwen-Image", "1:1 Square", 1328, 1328),
|
|
("Qwen-Image", "3:2 Landscape", 1584, 1056),
|
|
("Qwen-Image", "4:3 Landscape", 1472, 1136),
|
|
("Qwen-Image", "16:9 Landscape", 1664, 928),
|
|
("Qwen-Image", "2:3 Portrait", 1056, 1584),
|
|
("Qwen-Image", "3:4 Portrait", 1136, 1472),
|
|
("Qwen-Image", "9:16 Portrait", 928, 1664),
|
|
|
|
# --- Flux.1 (dev/schnell/pro, native 1024, ~1MP) ---
|
|
("Flux.1", "1:1 Square", 512, 512),
|
|
("Flux.1", "1:1 Square", 1408, 1408),
|
|
("Flux.1", "3:2 Landscape", 1216, 832),
|
|
("Flux.1", "4:3 Landscape", 1664, 1216),
|
|
("Flux.1", "16:9 Landscape", 1920, 1088),
|
|
("Flux.1", "21:9 Landscape", 2176, 960),
|
|
("Flux.1", "2:3 Portrait", 832, 1216),
|
|
("Flux.1", "3:4 Portrait", 1216, 1664),
|
|
("Flux.1", "9:16 Portrait", 1088, 1920),
|
|
("Flux.1", "9:21 Portrait", 960, 2176),
|
|
|
|
# --- Stable Diffusion XL (native 1024, ~1MP) ---
|
|
("SDXL", "1:1 Square", 512, 512),
|
|
("SDXL", "1:1 Square", 1024, 1024),
|
|
("SDXL", "3:2 Landscape", 1152, 768),
|
|
("SDXL", "4:3 Landscape", 1152, 864),
|
|
("SDXL", "16:9 Landscape", 1360, 768),
|
|
("SDXL", "2:3 Portrait", 768, 1152),
|
|
("SDXL", "3:4 Portrait", 864, 1152),
|
|
("SDXL", "9:16 Portrait", 768, 1360),
|
|
|
|
# --- Stable Diffusion 1.5 (native 512, legacy) ---
|
|
("SD15", "1:1 Square", 512, 512),
|
|
("SD15", "3:2 Landscape", 768, 512),
|
|
("SD15", "4:3 Landscape", 768, 576),
|
|
("SD15", "16:9 Landscape", 912, 512),
|
|
("SD15", "2:3 Portrait", 512, 768),
|
|
("SD15", "3:4 Portrait", 576, 768),
|
|
("SD15", "9:16 Portrait", 512, 912),
|
|
|
|
# Ideogram 4.0 and ERNIE are hosted API models (no local LATENT/diffusion sampling
|
|
# in ComfyUI) — deliberately excluded so this node never implies false compatibility.
|
|
]
|