Sharpen Simple
One knob, zero ceremony — and that's the point
- image
- IMAGE
Upscaled an image and it came out soft? Rendered something that looks like it was drawn with a wet brush? Sharpen Simple is the "just make it look a little crisper" node, and it has exactly one dial. From TrophiHunter's Photography Nodes pack, it's the least ambitious member of the sharpen family - and sometimes that's exactly what you want in a batch pipeline: something fast, predictable, and impossible to misconfigure into a mess.
How it works
The mechanism is a 3×3 high-pass kernel - a nine-value edge-detection filter that finds the transitions in the image - added back on top of the original, scaled by strength. That's the classic "unsharp" math minus the radius control: image + strength × (high-frequency detail). High-frequency detail means edges, so it makes lines and textures read sharper without re-sampling or hallucinating new pixels. It runs as a single grouped convolution, which is about as cheap as an image operation gets, and it handles whole batches without blinking.
Inputs
image- IMAGE in, IMAGE out.strength- 0 to 10, default 1. 0 is a no-op. Realistic working range is 1 to 3; past 4 you're chasing halos.
The output IMAGE feeds straight into Preview/Save. In the pack's example workflow it sits in the sharpen group alongside Sharpen Unsharp Mask and before the vignette - sharpen, then vignette, then save. That ordering matters, because sharpening after a vignette would sharpen the vignette's edge too.
Where it falls short
One knob means no control over the radius - the size of the edge detail you're boosting. On a soft, upscaled face, a 3×3 kernel catches micro-edges fine, but on a noisy image it also sharpens the noise, and pushed hard it puts halos on high-contrast edges. If you hit either wall, the same pack ships Sharpen Unsharp Mask, which is Photoshop's unsharp filter with a proper radius and strength pair. For everything else, this node is the one you reach for: it does the job and gets out of the way.
Installing it
Part of the Photography Nodes pack:
cd ComfyUI/custom_nodes
git clone https://github.com/TrophiHunter/ComfyUI_Photography_Nodes
# restart, then find it under TrophiHunter/Photography
Or ComfyUI Manager → search "Photography Nodes" by TrophiHunter. No models, no extra dependencies - torch and torchvision are already there.
Troubleshooting
The two failure modes are "I don't see a difference" and "everything has white edges." The first is usually a soft image at strength 1 - bump it to 2 and look at the eyes or hair. The second is strength too high; back off until the halos vanish. And if your image was upscaled with a model that already sharpens as part of its job, skip this node entirely - doubling up is how you get that overcooked, crunchy look.
Inputs (2)
| Name | Type | Default | Description |
|---|---|---|---|
| image | IMAGE | — | |
| strength | FLOAT | 1.000–10 | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| IMAGE | IMAGE | — |