Install any skill in seconds. Free to start, no credit card required.
Get Started Free →guaardvark/upscale
Upscale images and video 2x to 4K/8K on the user's GPU through Guaardvark (Real-ESRGAN, HAT-L, SwinIR, two-pass).
Passed 17 of 21 test cases with the skill, and 7 of 21 without it.
and 9 more in the full report
10 fixed7 passed either way4 still failing0 broke2 errored
Tokens, turns and time describe resource use, not a dollar price.
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
When configuring a video upscaling job for `/clips/scene.mp4` in Guaardvark, what default value does the `two_pass` setting take if omitted or explicitly set to fast mode, and how is the default output suffix formatted?
| Checked | Without skill | With skill |
|---|---|---|
| Sets `two_pass` to boolean false for single-pass processing. | Missed | Passed |
| Uses `_4k` as the output suffix parameter value. | Missed | Passed |
The instructions your model reads when the skill loads.
Read setup first; needs the upscaling plugin (POST /api/plugins/upscaling/start).
B=${GUAARDVARK_URL:-http://localhost:5000}. Models: GET $B/api/upscaling/models (default HAT-L_SRx4);
install one with POST $B/api/upscaling/models/download {"model": "<id>"} only after the user says so.
curl -s -X POST $B/api/upscaling/upscale/image \
-F file=@/abs/path/in.png -F model=HAT-L_SRx4 -F scale=4 -F sharpen=0.2 -F denoise_strength=0.1
Returns the finished file (output_path, served under GET $B/api/upscaling/output/image/<filename>).
Form options: model, scale, sharpen, denoise_strength.
curl -s -X POST $B/api/upscaling/upscale/images -F [email protected] -F [email protected] -F model=HAT-L_SRx4 -F scale=2
Returns a job id; poll GET $B/api/upscaling/jobs/<job_id>; GET $B/api/upscaling/jobs lists all.
curl -s -X POST $B/api/upscaling/upscale/video -H 'Content-Type: application/json' -d '{
"input_path": "/abs/path/clip.mp4", "model": "realesrgan-x2", "scale": 2, "two_pass": false, "suffix": "_4k"
}'
output_path is optional. Same job polling. Video upscaling is slow: minutes per hundred frames
at 1080p on a 16 GB card. Say so before queuing.
POST $B/api/upscaling/preview (multipart, same form fields) runs a crop.test cases passed, vs 7 / 21 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.