OpenSourceJoker/video2x
3
0
Fork 0
mirror of https://github.com/k4yt3x/video2x.git synced 2025-02-26 17:39:07 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

chore(models): add the Real-ESRGAN general models (#1319)
Some checks failed
Build / ubuntu (push) Has been cancelled
Details
Build / windows (push) Has been cancelled
Details
Build / container (push) Has been cancelled
Details
Build / appimage (push) Has been cancelled
Details

Browse Source 
Signed-off-by: k4yt3x <i@k4yt3x.com>
This commit is contained in:
k4yt3x 2025-02-02 00:00:00 +00:00
parent 2c5a059d39
commit afa37f9e58
No known key found for this signature in database
10 changed files with 195 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
CHANGELOG.md
View File

@ -5,12 +5,18 @@ All notable changes to this project will be documented in this file.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
## [Unreleased]
### Added
- Real-ESRGAN models `realesr-general-x4v3` and `realesr-general-wdn-x4v3` (#1319).
## [6.4.0] - 2025-01-24 ## [6.4.0] - 2025-01-24
### Added ### Added
- Multi-versioning to critical functions to enhance performance in generic architecture builds. - Multi-versioning to critical functions to enhance performance in generic architecture builds.
- RIFE models v4.25 and v4.26 (#1304). - RIFE models `v4.25` and `v4.26` (#1304).
- Support for processing videos without PTS information (#1278). - Support for processing videos without PTS information (#1278).
- The feature to copy input streams' metadata to the output streams (#1282). - The feature to copy input streams' metadata to the output streams (#1282).
- (Video2X Qt6) German translation (#1279). - (Video2X Qt6) German translation (#1279).

2
include/libvideo2x/filter_realesrgan.h
View File

@ -18,6 +18,7 @@ class FilterRealesrgan : public Filter {
int gpuid = 0, int gpuid = 0,
bool tta_mode = false, bool tta_mode = false,
int scaling_factor = 4, int scaling_factor = 4,
int noise_level = 0,
const fsutils::StringType model_name = STR("realesr-animevideov3") const fsutils::StringType model_name = STR("realesr-animevideov3")
); );
@ -47,6 +48,7 @@ class FilterRealesrgan : public Filter {
int gpuid_; int gpuid_;
bool tta_mode_; bool tta_mode_;
int scaling_factor_; int scaling_factor_;
int noise_level_;
const fsutils::StringType model_name_; const fsutils::StringType model_name_;
AVRational in_time_base_; AVRational in_time_base_;
AVRational out_time_base_; AVRational out_time_base_;

BIN
models/realesrgan/realesr-generalv3-wdn-x4.bin vendored Normal file
View File

Binary file not shown.

75
models/realesrgan/realesr-generalv3-wdn-x4.param vendored Normal file
View File

@ -0,0 +1,75 @@
7767517
73 74
Input data 0 1 data
Split splitncnn_input0 1 2 data data_splitncnn_0 data_splitncnn_1
Convolution /Conv 1 1 data_splitncnn_1 /Conv_output_0 0=64 1=3 4=1 5=1 6=1728
PReLU /PRelu 1 1 /Conv_output_0 /PRelu_output_0 0=64
Convolution /Conv_1 1 1 /PRelu_output_0 /Conv_1_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_1 1 1 /Conv_1_output_0 /PRelu_1_output_0 0=64
Convolution /Conv_2 1 1 /PRelu_1_output_0 /Conv_2_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_2 1 1 /Conv_2_output_0 /PRelu_2_output_0 0=64
Convolution /Conv_3 1 1 /PRelu_2_output_0 /Conv_3_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_3 1 1 /Conv_3_output_0 /PRelu_3_output_0 0=64
Convolution /Conv_4 1 1 /PRelu_3_output_0 /Conv_4_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_4 1 1 /Conv_4_output_0 /PRelu_4_output_0 0=64
Convolution /Conv_5 1 1 /PRelu_4_output_0 /Conv_5_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_5 1 1 /Conv_5_output_0 /PRelu_5_output_0 0=64
Convolution /Conv_6 1 1 /PRelu_5_output_0 /Conv_6_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_6 1 1 /Conv_6_output_0 /PRelu_6_output_0 0=64
Convolution /Conv_7 1 1 /PRelu_6_output_0 /Conv_7_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_7 1 1 /Conv_7_output_0 /PRelu_7_output_0 0=64
Convolution /Conv_8 1 1 /PRelu_7_output_0 /Conv_8_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_8 1 1 /Conv_8_output_0 /PRelu_8_output_0 0=64
Convolution /Conv_9 1 1 /PRelu_8_output_0 /Conv_9_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_9 1 1 /Conv_9_output_0 /PRelu_9_output_0 0=64
Convolution /Conv_10 1 1 /PRelu_9_output_0 /Conv_10_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_10 1 1 /Conv_10_output_0 /PRelu_10_output_0 0=64
Convolution /Conv_11 1 1 /PRelu_10_output_0 /Conv_11_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_11 1 1 /Conv_11_output_0 /PRelu_11_output_0 0=64
Convolution /Conv_12 1 1 /PRelu_11_output_0 /Conv_12_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_12 1 1 /Conv_12_output_0 /PRelu_12_output_0 0=64
Convolution /Conv_13 1 1 /PRelu_12_output_0 /Conv_13_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_13 1 1 /Conv_13_output_0 /PRelu_13_output_0 0=64
Convolution /Conv_14 1 1 /PRelu_13_output_0 /Conv_14_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_14 1 1 /Conv_14_output_0 /PRelu_14_output_0 0=64
Convolution /Conv_15 1 1 /PRelu_14_output_0 /Conv_15_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_15 1 1 /Conv_15_output_0 /PRelu_15_output_0 0=64
Convolution /Conv_16 1 1 /PRelu_15_output_0 /Conv_16_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_16 1 1 /Conv_16_output_0 /PRelu_16_output_0 0=64
Convolution /Conv_17 1 1 /PRelu_16_output_0 /Conv_17_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_17 1 1 /Conv_17_output_0 /PRelu_17_output_0 0=64
Convolution /Conv_18 1 1 /PRelu_17_output_0 /Conv_18_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_18 1 1 /Conv_18_output_0 /PRelu_18_output_0 0=64
Convolution /Conv_19 1 1 /PRelu_18_output_0 /Conv_19_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_19 1 1 /Conv_19_output_0 /PRelu_19_output_0 0=64
Convolution /Conv_20 1 1 /PRelu_19_output_0 /Conv_20_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_20 1 1 /Conv_20_output_0 /PRelu_20_output_0 0=64
Convolution /Conv_21 1 1 /PRelu_20_output_0 /Conv_21_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_21 1 1 /Conv_21_output_0 /PRelu_21_output_0 0=64
Convolution /Conv_22 1 1 /PRelu_21_output_0 /Conv_22_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_22 1 1 /Conv_22_output_0 /PRelu_22_output_0 0=64
Convolution /Conv_23 1 1 /PRelu_22_output_0 /Conv_23_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_23 1 1 /Conv_23_output_0 /PRelu_23_output_0 0=64
Convolution /Conv_24 1 1 /PRelu_23_output_0 /Conv_24_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_24 1 1 /Conv_24_output_0 /PRelu_24_output_0 0=64
Convolution /Conv_25 1 1 /PRelu_24_output_0 /Conv_25_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_25 1 1 /Conv_25_output_0 /PRelu_25_output_0 0=64
Convolution /Conv_26 1 1 /PRelu_25_output_0 /Conv_26_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_26 1 1 /Conv_26_output_0 /PRelu_26_output_0 0=64
Convolution /Conv_27 1 1 /PRelu_26_output_0 /Conv_27_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_27 1 1 /Conv_27_output_0 /PRelu_27_output_0 0=64
Convolution /Conv_28 1 1 /PRelu_27_output_0 /Conv_28_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_28 1 1 /Conv_28_output_0 /PRelu_28_output_0 0=64
Convolution /Conv_29 1 1 /PRelu_28_output_0 /Conv_29_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_29 1 1 /Conv_29_output_0 /PRelu_29_output_0 0=64
Convolution /Conv_30 1 1 /PRelu_29_output_0 /Conv_30_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_30 1 1 /Conv_30_output_0 /PRelu_30_output_0 0=64
Convolution /Conv_31 1 1 /PRelu_30_output_0 /Conv_31_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_31 1 1 /Conv_31_output_0 /PRelu_31_output_0 0=64
Convolution /Conv_32 1 1 /PRelu_31_output_0 /Conv_32_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_32 1 1 /Conv_32_output_0 /PRelu_32_output_0 0=64
Convolution /Conv_33 1 1 /PRelu_32_output_0 /Conv_33_output_0 0=48 1=3 4=1 5=1 6=27648
PixelShuffle /DepthToSpace 1 1 /Conv_33_output_0 /DepthToSpace_output_0 0=4
Interp /Resize 1 1 data_splitncnn_0 /Resize_output_0 0=1 1=4.000000e+00 2=4.000000e+00
BinaryOp /Add 2 1 /DepthToSpace_output_0 /Resize_output_0 /Add_output_0
Clip /Clip 1 1 /Add_output_0 output 0=0.000000e+00 1=1.000000e+00

BIN
models/realesrgan/realesr-generalv3-x4.bin vendored Normal file
View File

Binary file not shown.

75
models/realesrgan/realesr-generalv3-x4.param vendored Normal file
View File

@ -0,0 +1,75 @@
7767517
73 74
Input data 0 1 data
Split splitncnn_input0 1 2 data data_splitncnn_0 data_splitncnn_1
Convolution /Conv 1 1 data_splitncnn_1 /Conv_output_0 0=64 1=3 4=1 5=1 6=1728
PReLU /PRelu 1 1 /Conv_output_0 /PRelu_output_0 0=64
Convolution /Conv_1 1 1 /PRelu_output_0 /Conv_1_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_1 1 1 /Conv_1_output_0 /PRelu_1_output_0 0=64
Convolution /Conv_2 1 1 /PRelu_1_output_0 /Conv_2_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_2 1 1 /Conv_2_output_0 /PRelu_2_output_0 0=64
Convolution /Conv_3 1 1 /PRelu_2_output_0 /Conv_3_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_3 1 1 /Conv_3_output_0 /PRelu_3_output_0 0=64
Convolution /Conv_4 1 1 /PRelu_3_output_0 /Conv_4_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_4 1 1 /Conv_4_output_0 /PRelu_4_output_0 0=64
Convolution /Conv_5 1 1 /PRelu_4_output_0 /Conv_5_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_5 1 1 /Conv_5_output_0 /PRelu_5_output_0 0=64
Convolution /Conv_6 1 1 /PRelu_5_output_0 /Conv_6_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_6 1 1 /Conv_6_output_0 /PRelu_6_output_0 0=64
Convolution /Conv_7 1 1 /PRelu_6_output_0 /Conv_7_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_7 1 1 /Conv_7_output_0 /PRelu_7_output_0 0=64
Convolution /Conv_8 1 1 /PRelu_7_output_0 /Conv_8_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_8 1 1 /Conv_8_output_0 /PRelu_8_output_0 0=64
Convolution /Conv_9 1 1 /PRelu_8_output_0 /Conv_9_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_9 1 1 /Conv_9_output_0 /PRelu_9_output_0 0=64
Convolution /Conv_10 1 1 /PRelu_9_output_0 /Conv_10_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_10 1 1 /Conv_10_output_0 /PRelu_10_output_0 0=64
Convolution /Conv_11 1 1 /PRelu_10_output_0 /Conv_11_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_11 1 1 /Conv_11_output_0 /PRelu_11_output_0 0=64
Convolution /Conv_12 1 1 /PRelu_11_output_0 /Conv_12_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_12 1 1 /Conv_12_output_0 /PRelu_12_output_0 0=64
Convolution /Conv_13 1 1 /PRelu_12_output_0 /Conv_13_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_13 1 1 /Conv_13_output_0 /PRelu_13_output_0 0=64
Convolution /Conv_14 1 1 /PRelu_13_output_0 /Conv_14_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_14 1 1 /Conv_14_output_0 /PRelu_14_output_0 0=64
Convolution /Conv_15 1 1 /PRelu_14_output_0 /Conv_15_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_15 1 1 /Conv_15_output_0 /PRelu_15_output_0 0=64
Convolution /Conv_16 1 1 /PRelu_15_output_0 /Conv_16_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_16 1 1 /Conv_16_output_0 /PRelu_16_output_0 0=64
Convolution /Conv_17 1 1 /PRelu_16_output_0 /Conv_17_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_17 1 1 /Conv_17_output_0 /PRelu_17_output_0 0=64
Convolution /Conv_18 1 1 /PRelu_17_output_0 /Conv_18_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_18 1 1 /Conv_18_output_0 /PRelu_18_output_0 0=64
Convolution /Conv_19 1 1 /PRelu_18_output_0 /Conv_19_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_19 1 1 /Conv_19_output_0 /PRelu_19_output_0 0=64
Convolution /Conv_20 1 1 /PRelu_19_output_0 /Conv_20_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_20 1 1 /Conv_20_output_0 /PRelu_20_output_0 0=64
Convolution /Conv_21 1 1 /PRelu_20_output_0 /Conv_21_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_21 1 1 /Conv_21_output_0 /PRelu_21_output_0 0=64
Convolution /Conv_22 1 1 /PRelu_21_output_0 /Conv_22_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_22 1 1 /Conv_22_output_0 /PRelu_22_output_0 0=64
Convolution /Conv_23 1 1 /PRelu_22_output_0 /Conv_23_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_23 1 1 /Conv_23_output_0 /PRelu_23_output_0 0=64
Convolution /Conv_24 1 1 /PRelu_23_output_0 /Conv_24_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_24 1 1 /Conv_24_output_0 /PRelu_24_output_0 0=64
Convolution /Conv_25 1 1 /PRelu_24_output_0 /Conv_25_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_25 1 1 /Conv_25_output_0 /PRelu_25_output_0 0=64
Convolution /Conv_26 1 1 /PRelu_25_output_0 /Conv_26_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_26 1 1 /Conv_26_output_0 /PRelu_26_output_0 0=64
Convolution /Conv_27 1 1 /PRelu_26_output_0 /Conv_27_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_27 1 1 /Conv_27_output_0 /PRelu_27_output_0 0=64
Convolution /Conv_28 1 1 /PRelu_27_output_0 /Conv_28_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_28 1 1 /Conv_28_output_0 /PRelu_28_output_0 0=64
Convolution /Conv_29 1 1 /PRelu_28_output_0 /Conv_29_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_29 1 1 /Conv_29_output_0 /PRelu_29_output_0 0=64
Convolution /Conv_30 1 1 /PRelu_29_output_0 /Conv_30_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_30 1 1 /Conv_30_output_0 /PRelu_30_output_0 0=64
Convolution /Conv_31 1 1 /PRelu_30_output_0 /Conv_31_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_31 1 1 /Conv_31_output_0 /PRelu_31_output_0 0=64
Convolution /Conv_32 1 1 /PRelu_31_output_0 /Conv_32_output_0 0=64 1=3 4=1 5=1 6=36864
PReLU /PRelu_32 1 1 /Conv_32_output_0 /PRelu_32_output_0 0=64
Convolution /Conv_33 1 1 /PRelu_32_output_0 /Conv_33_output_0 0=48 1=3 4=1 5=1 6=27648
PixelShuffle /DepthToSpace 1 1 /Conv_33_output_0 /DepthToSpace_output_0 0=4
Interp /Resize 1 1 data_splitncnn_0 /Resize_output_0 0=1 1=4.000000e+00 2=4.000000e+00
BinaryOp /Add 2 1 /DepthToSpace_output_0 /Resize_output_0 /Add_output_0
Clip /Clip 1 1 /Add_output_0 output 0=0.000000e+00 1=1.000000e+00

16
src/filter_realesrgan.cpp
View File

@ -17,12 +17,14 @@ FilterRealesrgan::FilterRealesrgan(
int gpuid, int gpuid,
bool tta_mode, bool tta_mode,
int scaling_factor, int scaling_factor,
int noise_level,
const fsutils::StringType model_name const fsutils::StringType model_name
) )
: realesrgan_(nullptr), : realesrgan_(nullptr),
gpuid_(gpuid), gpuid_(gpuid),
tta_mode_(tta_mode), tta_mode_(tta_mode),
scaling_factor_(scaling_factor), scaling_factor_(scaling_factor),
noise_level_(noise_level),
model_name_(std::move(model_name)) {} model_name_(std::move(model_name)) {}
FilterRealesrgan::~FilterRealesrgan() { FilterRealesrgan::~FilterRealesrgan() {
@ -35,10 +37,16 @@ int FilterRealesrgan::init(AVCodecContext* dec_ctx, AVCodecContext* enc_ctx, AVB
std::filesystem::path model_param_path; std::filesystem::path model_param_path;
std::filesystem::path model_bin_path; std::filesystem::path model_bin_path;
fsutils::StringType param_file_name = fsutils::StringType param_file_name = model_name_;
model_name_ + STR("-x") + fsutils::to_string_type(scaling_factor_) + STR(".param"); fsutils::StringType bin_file_name = model_name_;
fsutils::StringType bin_file_name =
model_name_ + STR("-x") + fsutils::to_string_type(scaling_factor_) + STR(".bin"); if (model_name_ == STR("realesr-generalv3") && noise_level_ > 0) {
param_file_name += STR("-wdn");
bin_file_name += STR("-wdn");
}
param_file_name += STR("-x") + fsutils::to_string_type(scaling_factor_) + STR(".param");
bin_file_name += STR("-x") + fsutils::to_string_type(scaling_factor_) + STR(".bin");
// Find the model paths by model name if provided // Find the model paths by model name if provided
model_param_path = std::filesystem::path(STR("models")) / STR("realesrgan") / param_file_name; model_param_path = std::filesystem::path(STR("models")) / STR("realesrgan") / param_file_name;

1
src/processor_factory.cpp
View File

@ -91,6 +91,7 @@ void ProcessorFactory::init_default_processors(ProcessorFactory& factory) {
static_cast<int>(vk_device_index), static_cast<int>(vk_device_index),
config.tta_mode, config.tta_mode,
proc_cfg.scaling_factor, proc_cfg.scaling_factor,
proc_cfg.noise_level,
config.model_name config.model_name
); );
} }

35
tools/video2x/src/argparse.cpp
View File

@ -156,23 +156,23 @@ int parse_args(
"anime4k-v4-b, anime4k-v4-b+b, anime4k-v4-c, anime4k-v4-c+a, anime4k-v4.1-gan)") "anime4k-v4-b, anime4k-v4-b+b, anime4k-v4-c, anime4k-v4-c+a, anime4k-v4.1-gan)")
; ;
po::options_description realesrgan_opts("RealESRGAN options"); po::options_description realesrgan_opts("Real-ESRGAN options");
realesrgan_opts.add_options() realesrgan_opts.add_options()
("realesrgan-model", PO_STR_VALUE<video2x::fsutils::StringType>() ("realesrgan-model", PO_STR_VALUE<video2x::fsutils::StringType>()
->default_value(STR("realesr-animevideov3"), "realesr-animevideov3") ->default_value(STR("realesr-animevideov3"), "realesr-animevideov3")
->notifier(validate_realesrgan_model_name), ->notifier(validate_realesrgan_model_name),
"Name of the RealESRGAN model to use (realesr-animevideov3, realesrgan-plus-anime, " "Name of the Real-ESRGAN model to use (realesr-animevideov3, "
"realesrgan-plus)") "realesrgan-plus-anime, realesrgan-plus, realesr-generalv3)")
; ;
po::options_description realcugan_opts("RealCUGAN options"); po::options_description realcugan_opts("Real-CUGAN options");
realcugan_opts.add_options() realcugan_opts.add_options()
("realcugan-model", PO_STR_VALUE<video2x::fsutils::StringType>() ("realcugan-model", PO_STR_VALUE<video2x::fsutils::StringType>()
->default_value(STR("models-se"), "models-se") ->default_value(STR("models-se"), "models-se")
->notifier(validate_realcugan_model_name), ->notifier(validate_realcugan_model_name),
"Name of the RealCUGAN model to use (models-nose, models-pro, models-se)") "Name of the Real-CUGAN model to use (models-nose, models-pro, models-se)")
("realcugan-threads", po::value<int>()->default_value(1), ("realcugan-threads", po::value<int>()->default_value(1),
"Number of threads to use for RealCUGAN") "Number of threads to use for Real-CUGAN")
("realcugan-syncgap", po::value<int>()->default_value(3), ("realcugan-syncgap", po::value<int>()->default_value(3),
"Sync gap mode; 0:no sync, 1: accurate sync: 2 = rough sync, 3: very rough sync") "Sync gap mode; 0:no sync, 1: accurate sync: 2 = rough sync, 3: very rough sync")
; ;
@ -212,7 +212,7 @@ int parse_args(
<< " video2x -i input.mp4 -o output.mp4 -w 3840 -h 2160 \\" << std::endl << " video2x -i input.mp4 -o output.mp4 -w 3840 -h 2160 \\" << std::endl
<< " -p libplacebo --libplacebo-shader anime4k-v4-a+a" << std::endl << " -p libplacebo --libplacebo-shader anime4k-v4-a+a" << std::endl
<< std::endl << std::endl
<< " Upscale a film by 4x using RealESRGAN with custom encoder options:" << " Upscale a film by 4x using Real-ESRGAN with custom encoder options:"
<< std::endl << std::endl
<< " video2x -i input.mkv -o output.mkv -s 4 \\" << std::endl << " video2x -i input.mkv -o output.mkv -s 4 \\" << std::endl
<< " -p realesrgan --realesrgan-model realesrgan-plus \\" << std::endl << " -p realesrgan --realesrgan-model realesrgan-plus \\" << std::endl
@ -385,16 +385,19 @@ int parse_args(
} }
case video2x::processors::ProcessorType::RealESRGAN: { case video2x::processors::ProcessorType::RealESRGAN: {
if (!vm.count("realesrgan-model")) { if (!vm.count("realesrgan-model")) {
video2x::logger()->critical("RealESRGAN model name must be set for RealESRGAN." video2x::logger()->critical("The model name must be set for Real-ESRGAN.");
);
return -1; return -1;
} }
if (proc_cfg.scaling_factor < 2 || proc_cfg.scaling_factor > 4) { if (proc_cfg.scaling_factor < 2 || proc_cfg.scaling_factor > 4) {
video2x::logger()->critical( video2x::logger()->critical(
"Scaling factor must be set to 2, 3, or 4 for RealESRGAN." "Scaling factor must be set to 2, 3, or 4 for Real-ESRGAN."
); );
return -1; return -1;
} }
if (proc_cfg.noise_level < 0 || proc_cfg.noise_level > 1) {
video2x::logger()->critical("Noise level must be 0 or 1 for Real-ESRGAN.");
return -1;
}
proc_cfg.processor_type = video2x::processors::ProcessorType::RealESRGAN; proc_cfg.processor_type = video2x::processors::ProcessorType::RealESRGAN;
video2x::processors::RealESRGANConfig realesrgan_config; video2x::processors::RealESRGANConfig realesrgan_config;
@ -406,31 +409,31 @@ int parse_args(
} }
case video2x::processors::ProcessorType::RealCUGAN: { case video2x::processors::ProcessorType::RealCUGAN: {
if (!vm.count("realcugan-model")) { if (!vm.count("realcugan-model")) {
video2x::logger()->critical("RealCUGAN model name must be set for RealCUGAN."); video2x::logger()->critical("The model name must be set for Real-CUGAN.");
return -1; return -1;
} }
if (vm.count("realcugan-threads") && vm["realcugan-threads"].as<int>() < 1) { if (vm.count("realcugan-threads") && vm["realcugan-threads"].as<int>() < 1) {
video2x::logger()->critical( video2x::logger()->critical(
"Number of threads must be at least 1 for RealCUGAN." "Number of threads must be at least 1 for Real-CUGAN."
); );
return -1; return -1;
} }
if (vm.count("realcugan-syncgap") && (vm["realcugan-syncgap"].as<int>() < 0 || if (vm.count("realcugan-syncgap") && (vm["realcugan-syncgap"].as<int>() < 0 ||
vm["realcugan-syncgap"].as<int>() > 3)) { vm["realcugan-syncgap"].as<int>() > 3)) {
video2x::logger()->critical( video2x::logger()->critical(
"Sync gap mode must be set to 0, 1, 2, or 3 for RealCUGAN." "Sync gap mode must be set to 0, 1, 2, or 3 for Real-CUGAN."
); );
return -1; return -1;
} }
if (proc_cfg.scaling_factor < 2 || proc_cfg.scaling_factor > 4) { if (proc_cfg.scaling_factor < 2 || proc_cfg.scaling_factor > 4) {
video2x::logger()->critical( video2x::logger()->critical(
"Scaling factor must be set to 2, 3, or 4 for RealCUGAN." "Scaling factor must be set to 2, 3, or 4 for Real-CUGAN."
); );
return -1; return -1;
} }
if (proc_cfg.noise_level < -1 || proc_cfg.noise_level > 3) { if (proc_cfg.noise_level < -1 || proc_cfg.noise_level > 3) {
video2x::logger()->critical( video2x::logger()->critical(
"Noise level must be set to -1, 0, 1, 2, or 3 for RealCUGAN." "Noise level must be set to -1, 0, 1, 2, or 3 for Real-CUGAN."
); );
return -1; return -1;
} }
@ -447,7 +450,7 @@ int parse_args(
} }
case video2x::processors::ProcessorType::RIFE: { case video2x::processors::ProcessorType::RIFE: {
if (!vm.count("rife-model")) { if (!vm.count("rife-model")) {
video2x::logger()->critical("RIFE model name must be set for RIFE."); video2x::logger()->critical("The model name must be set for RIFE.");
return -1; return -1;
} }
if (proc_cfg.frm_rate_mul < 2) { if (proc_cfg.frm_rate_mul < 2) {

5
tools/video2x/src/validators.cpp
View File

@ -24,7 +24,10 @@ void validate_anime4k_shader_name(const video2x::fsutils::StringType& shader_nam
void validate_realesrgan_model_name(const video2x::fsutils::StringType& model_name) { void validate_realesrgan_model_name(const video2x::fsutils::StringType& model_name) {
static const std::unordered_set<video2x::fsutils::StringType> valid_realesrgan_models = { static const std::unordered_set<video2x::fsutils::StringType> valid_realesrgan_models = {
STR("realesrgan-plus"), STR("realesrgan-plus-anime"), STR("realesr-animevideov3") STR("realesrgan-plus"),
STR("realesrgan-plus-anime"),
STR("realesr-animevideov3"),
STR("realesr-generalv3"),
}; };
if (valid_realesrgan_models.count(model_name) == 0) { if (valid_realesrgan_models.count(model_name) == 0) {
throw po::validation_error( throw po::validation_error(