Image Vignette

A vignette is the cheapest "this was shot on a lens" trick in the book, and WAS Image Vignette does it properly - or in reverse. A small darkening at the corners pulls the eye to the middle of a shot; a negative amount lifts the corners and takes an existing vignette back out. It's a one-stop filter with controls a photographer would recognise, and it runs on a whole batch frame-for-frame with identical falloff.

The core is amount, -1 to 1: 0.5 puts corners at half brightness, 1.0 makes them black, 0.0 changes nothing, and -0.5 lifts them instead. The most useful trick here is the negative direction - if your footage or a render already has a heavy lens falloff and you want it gone, dialing a negative amount of roughly matching strength cancels it out rather than trying to mask it.

Then three controls shape the falloff. size sets how far out the falloff finishes - 0.75 keeps the middle clear and goes full-dark only at the corners, 1.0 darkens only the very corners, 0.3 reads as a tight spotlight. feather spreads the transition: 0.0 is a hard-edged circle, 0.5 a gradual one, 1.0 falls off from the centre out. The two of them together are how you get everything from "barely there, photo-like" to "dramatic film still."

shape answers the question of what "the corners" means: to the frame stretches the falloff into an oval matching the aspect ratio - what a lens actually does on a 16:9 sensor - while circular keeps it round, which on a wide frame darkens the left and right edges before the top and bottom. For anything meant to look lens-shot, to the frame is the honest choice.

The sleeper feature is that the falloff centre moves. centre_x and centre_y default to 0.5 (dead middle) but range from -1 to 2, so you can park the brightest point on a subject that isn't in the middle of the frame. A portrait subject off to one side shouldn't be sitting in the darkest part of your own vignette - nudge the centre onto them and the falloff does its eye-guiding job without crushing the person you care about.

Output is images, each frame back at the size it went in at with the same falloff applied. It slots anywhere along a post-processing chain - right before a save, after a composite, or as the finishing pass on a batch you're exporting. As post-processing filters go this one is small, deterministic and instantly tunable, which is precisely why it keeps showing up in finishing chains. Just don't overshoot it: heavy vignettes are the look that reads as "amateur filter" the moment anyone else sees it. Small amounts, negative amounts for correction, and you're done.

Install

Part of WAS Node Suite v3. ComfyUI Manager, search "WAS Node Suite", or:

cd ComfyUI/custom_nodes
git clone https://github.com/WASasquatch/was-node-suite-comfyui

Restart after. Needs ComfyUI 0.14.0+ and Python 3.10+. Nothing pip-installs, no model files. (The v2 pack's heavy install is a thing of the past in v3.)

Common issues

• Corners black when you wanted a hint. amount 0.5 is already strong. Start at 0.15–0.3 for the subtle lens look.
• The subject is the darkest thing in the shot. Move centre_x/centre_y onto them instead of leaving the falloff centred.
• Trying to remove a vignette from footage. Use a negative amount around the strength of the one present - and keep size/feather similar to whatever created it.