Flux Resolution Matcher MXD
Flux resolution presets, picked for you across three tiers
- image
- resolution
- vertical
Flux is more forgiving about resolution than SDXL - the KB's flux guide notes it "happily generates in the two-megapixel range," and power users push it well past 1024×1024. But "more forgiving" isn't "anything goes," and eyeballing a size every time is exactly the friction this pack exists to remove. Flux Resolution Matcher MXD looks at your input image and returns the closest Flux preset from a 15-entry list organized into three tiers - high (up to 2176×960), standard (the SDXL-era sizes), and low (as small as 320×320) - plus a vertical flag.
How it works
One input (image), two outputs:
resolution- the preset string, e.g.Widescreen (16:9) 1344x768vertical- boolean for portrait orientation
The matching logic mirrors the pack's SDXL matcher: find the aspect-ratio group closest to your image, then within that group pick the preset whose pixel area is closest to your image's. The interesting part is the tiering - the matcher only reasons over the standard tier when picking, then returns a preset from that tier. So a 2400×1350 image won't get matched up to the high tier; it'll land on the closest standard size. That's a deliberate, sensible choice: standard-tier sizes are the ones with the most community mileage (guidance ~3.5, Euler, 1024 base per the KB's flux baseline), and matching to them keeps you in well-trodden territory.
The wiring trick
Same as the SDXL version: the resolution enum is the exact same list the pack's Flux Empty Latent Image MXD uses as its dropdown, so resolution and vertical plug straight into that node's inputs. Drag in a reference image and the latent gets built at a Flux-friendly size that matches it, automatically. If you prefer plain old EmptyLatentImage, you'll need a converter - this enum only connects to nodes sharing the same list, which is the pack's own empty-latent node.
When you'd actually use it
Img2img, IP-Adapter reference workflows, or any time the source image's dimensions should drive the canvas. It's the "stop re-typing resolutions" node for the Flux half of the pack (the SDXL sibling handles the other half). Given Flux's ability to run at higher res, this is also handy when you know you want more detail: match to a standard tier, then push it in the empty-latent node or upscale after.
Install
ComfyUI Manager → search Maxed Out → install → restart, or:
cd ComfyUI/custom_nodes
git clone https://github.com/Maxed-Out-99/ComfyUI-MaxedOut
No requirements.txt, no model downloads. Under MXD/Latent. If you've been squinting at an image trying to decide between 1152×896 and 1216×832 - this node retires that decision for good.
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| image | IMAGE | — |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| resolution | — High Resolutions —,Square (1:1) 1408x1408,Standard (4:3) 1664x1216,Landscape (3:2) 1728x1152,Widescreen (16:9) 1920x1088,Ultrawide (21:9) 2176x960,— Standard Resolutions —,Square (1:1) 1024x1024,Standard (4:3) 1152x896,Landscape (3:2) 1216x832,Widescreen (16:9) 1344x768,Ultrawide (21:9) 1536x640,— Low Resolutions —,Square (1:1) 320x320,Standard (4:3) 448x320,Landscape (3:2) 384x256,Widescreen (16:9) 448x256,Ultrawide (21:9) 576x256 | — |
| vertical | BOOLEAN | — |