Nodes/WAS Node Suite v3/Image Style Filter
ComfyUI Node Runs on cloud

Image Style Filter

Instagram-style presets for ComfyUI, via Pilgram

By WASasquatch·Created 3 years ago·Updated 5 days ago· 1,844
Image Style Filter
  • image
  • IMAGE
style
strength1.00
contact_sheettrue
use_gputrue

This one's straightforward: it applies Instagram-style filter presets to an image - the README literally says it uses Pilgram, a real Python library that reimplements Instagram's classic CSS filters (1977, Aden, Brooklyn, Clarendon, and the rest of that familiar set) programmatically. No AI, no model weights, just the same color-grading math Instagram used to ship.

Why you'd use it

It's a fast, deterministic way to grade a batch of images toward a consistent look without hand-tuning curves yourself. If you're producing social-style content and want that recognizable filtered aesthetic, or you just want a quick, repeatable mood shift (warmer, cooler, higher contrast, faded) applied uniformly across a batch, this does it in one node with none of the fiddliness of manually stacking color adjustment nodes. It's not going to compete with a proper film-emulation LUT workflow for precision, but for "make this look like it came out of a phone camera filter," it's exactly the tool.

What it needs

The image, and which named filter preset to apply. Pilgram ships a fixed set of named presets rather than continuous sliders - you're picking a look, not dialing one in from scratch - which is honestly part of the appeal: it's fast because there's nothing to tune.

Installing it

Through ComfyUI Manager: search "WAS Node Suite", install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/WASasquatch/was-node-suite-comfyui
cd was-node-suite-comfyui
pip install -r requirements.txt

This is one of the few nodes in the suite with a named, specific dependency worth calling out: it needs the pilgram package. The suite is supposed to install it automatically on first run, but if the node throws an import error the moment you try to use it (as opposed to the whole suite failing to load at startup), try pip install pilgram directly inside your ComfyUI environment - portable users run python_embeded\python.exe -s -m pip install pilgram.

Troubleshooting

Most issues with this node trace back to that pilgram dependency specifically failing its auto-install rather than anything about the filter logic - the presets themselves are simple, well-tested CSS-filter math that isn't going to behave inconsistently once pilgram is actually present.

The broader WAS Node Suite caveat still applies underneath that: the pack hasn't been actively maintained since December 2023, so a ComfyUI core update can occasionally break the suite's import at startup over an unrelated pinned dependency (BLIP support is the usual repeat offender). If Image Style Filter is missing entirely rather than erroring when used, check the console at startup first - that's a suite-wide import failure, not this node specifically.

CategoryWAS Suite/Image/Filter

Inputs (5)

NameTypeDefaultDescription
imageIMAGEThe image to treat. A batch is handled one image at a time.
styleCOMBOWhich of the 37 looks to apply. Tick `contact_sheet` to draw every one of them over this picture and click a tile to pick it. `inkwell` and `moon` are black and white, `kelvin` and `golden hour` the warmest, `moody blue` the coolest, `bleach bypass` the hardest, and `fairy tale` adds glitter instead of a grade.
strengthFLOAT1.000–1How far towards the look the result sits. 1.0 is the whole look, 0.5 half of it, 0.0 the image unchanged. Ignored by `fairy tale`.
contact_sheetBOOLEANtrue`true` draws every style over this picture on the node, so one run shows all 37 and a cell can be clicked to pick it. `false` draws the picture beside the graded result with the difference between them, which is quicker on a large frame.
use_gpuBOOLEANtrue`true` grades on the graphics card, which is around six times quicker on a 3840 by 2160 frame and around twice as quick at 512 by 512. `false` keeps it on the processor. A card that refuses the work falls back to the processor either way.

Outputs (1)

NameTypeDescription
IMAGEIMAGEThe styled images, one for each that went in and the same size as the source. Every image in a batch gets the same style, and under `fairy tale` each one gets its own glitter.