7 Commits
Author SHA1 Message Date
huchukato 5311be90e6 Increase max engine input from 1280 to 2048px 2026-08-14 07:56:02 +02:00
huchukato 0946dbcb7f Add dynamic scale support (2x/4x) for upscaler TensorRT node
- UpscalerTensorrt: new 'scale' input (2 or 4), replaces hardcoded *4
- get_final_resolutions: accepts upscale_factor param
- export_onnx.py: configured for 2xLexicaRRDBNet_Sharp (2x model)
- Allows using 2x upscaler models instead of only 4x
2026-08-13 08:19:33 +02:00
Yuvraj Seegolam 4db0c2f0f1 Update export_trt.py 2026-01-12 22:55:49 +04:00
Yuvraj Seegolam bb6c101909 Refactor export_trt to use trt_utilities module 2026-01-12 22:55:32 +04:00
Tomi K 6dc6af229d Check input model and use smaller dummy input if supported for less resource usage 2025-05-30 13:39:56 +03:00
yuvraj108c c412374900 fix export_onnx script 2025-03-04 16:14:41 +00:00
yuvraj108c ca44a17445 automatic model building + v2 2025-03-04 10:59:32 +00:00