SadTalker GFPGAN: enhancer ON vs OFF

Same input, restoration OFF and ON

Only enhancer changes in the main pair: none versus gfpgan. Both use size 512, crop, still OFF, batch size 2, expression scale 1.0 and pose style 0. The synchronized player shows GFPGAN OFF on the left, ON on the right and uses one audio track.

Inspect eyes, skin, mouth and face edges

In the sampled close-ups, GFPGAN ON gives eyes and facial hair more defined edges and produces smoother-looking skin than OFF. The mouth and teeth also look different, but motion and expression differ between these independent renders. The sharper presentation is visible here; exact identity recovery and improved temporal stability were not measured.

Look across the whole clip, especially around teeth, eye shape, facial hair and the jaw boundary. These are visual observations from one input, not scored identity, lip-sync or temporal-consistency measurements.

Measured render time and sampled VRAM

We rendered four combinations once each to cover both questions with minimal GPU time. There were no extra warm-ups or repeated trials. The node starts a new worker for every render; model-loading time is included. Values are individual measurements, not medians or a universal speed ranking.

Swipe the table sideways to view time, resolution and VRAM.

At base size 512, OFF took 55.41 s and ON took 119.39 s: ON added 63.99 s, or 2.15× the total elapsed time in this pair. The sampled whole-GPU peak rose from 10,047 MiB to 10,207 MiB. GFPGAN doubled final dimensions to 1024 × 1024; it did not change the base render setting.

Server execution_start → execution_success timing includes input loading, worker/model startup, rendering and export; queue wait and provisioning are excluded. Fresh filenames invalidate input caching, and successful outputs were checked for video/audio streams and hash integrity.

Reproduce the comparison in ComfyUI

Start with our ComfyUI setup guide. This workflow extends its LoadImage / LoadAudio template with two SadTalker nodes sharing the same inputs.

1. Get the comparison workflow and drag it into ComfyUI.
2. Upload the portrait and the narration. Both nodes share them.
3. Keep size = 512 on both. Set the left enhancer to none and the right to gfpgan. Keep crop, still OFF (displayed as animated), batch 2, expression 1.0 and pose 0.
4. Click Run. Generated MP4s and logs are saved under ComfyUI/output/sadtalker/<run-id>/. Repeating an unchanged workflow can return cached results.

for file in 512-off-api.json 512-on-api.json; do
  python -c 'import json,sys; print(json.dumps({"prompt":json.load(open(sys.argv[1]))}))' "$file" > prompt.json
  curl -X POST http://127.0.0.1:8188/prompt -H 'Content-Type: application/json' --data-binary @prompt.json
done

When does GFPGAN help, and when can it hurt?

Limits and failed runs

4 of 4 configurations completed successfully. No render errors occurred in these four runs. This is one portrait, one narration and one rented GPU. There are no repeat-trial error bars, scored lip-sync or identity metrics, or out-of-memory boundary tests.

GFPGAN OFF and ON are separate SadTalker runs. The template has no seed control, so part of a frame-level difference may come from expression, blinking or head motion. This is an end-to-end ComfyUI toggle comparison; it does not isolate restoration by processing the exact same generated frame twice.

If a render fails or memory runs out, inspect its worker log and reduce batch size to 1. That setting was not tested here. Changing batch size, preprocessing, still mode, hardware or worker CPU threads makes the timings less directly comparable.

Implementation references: the tested enhancement/upscaling code, GFPGAN’s official restoration project, and the tested ComfyUI node.