Image fDOF Filter

The "f" is short for "fake" - per WAS's own description, this "applies a fake depth of field effect to an image." Real depth of field comes from a lens: background blurs because it's genuinely out of focus at a given aperture. A diffusion model doesn't have a lens or an aperture, so everything tends to come out roughly equally sharp (or equally soft) throughout the frame. This node is the workaround - a post-process that manufactures the look of shallow depth of field without any of the physical camera behavior behind it.

How it works

Depth-of-field simulation of this kind almost always works from a depth map: something that tells the filter which pixels are "near" and which are "far," so it knows how much to blur each region - sharp near the focal point, progressively blurrier toward the edges of depth. WAS ships its own MiDaS Depth Approximation node in the same pack, which is a natural pairing here, though I can't confirm from the README alone exactly how this filter sources its depth information - it may take a depth map as an input, or it may work from a simpler heuristic. Treat that mechanism as a reasonable inference rather than a confirmed fact, and check the node's actual inputs once it's on your canvas.

Why you'd use it

Portraits and product shots are the classic case - you want the subject crisp and the background to melt away, the way a fast prime lens would render it, but your generation came out flat and evenly detailed front to back. Running it through a fake-DOF pass after generation gets you closer to that photographic look without needing a specialized depth-aware sampling pipeline.

The inputs and outputs that matter

An image in (plausibly alongside a depth source), a depth-blurred image out. I'm not going to assert specific parameter names here since I can't verify them - the mechanism above is my best-grounded read, not a schema.

How to install it

Bundled with the full WAS Node Suite:

• ComfyUI Manager - search "WAS Node Suite", install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/WASasquatch/was-node-suite-comfyui, install requirements.txt (portable: python_embeded\python.exe -s -m pip install -r requirements.txt), restart.

If this node does depend on a depth-estimation step under the hood, be aware the pack's MiDaS-based depth nodes pull in their own model weights on first use - a heavier ask than a plain pixel filter, and worth expecting if the effect stalls the first time you run it.

Common issues & troubleshooting

Blur bleeding onto the wrong subject, or a visible halo around edges. This is the classic failure mode of any depth-based blur simulation: a noisy or imprecise depth map blurs parts of your subject that should stay sharp, or leaves a hard-edged ring where the blur transitions too abruptly. If the effect looks wrong, suspect the depth information behind it before assuming the filter itself is broken.

Too subtle or too aggressive to look convincing. Fake DOF is easy to overshoot into obviously-filtered territory, or undershoot into "nothing changed." There's no universal setting - it depends heavily on your source image's actual depth complexity.

Standard pack caveat: WAS Node Suite has been unmaintained since December 2023, and the recurring issue is the whole suite failing to import after a ComfyUI update. Reinstall requirements.txt against the right Python environment if the suite won't load.