This is a Tutorial to install and use the enhanced MOD for javplayer_109a, enhance the effect after removing mosaic.
This MOD was created by the Discord community kmkii#8701.
If you can run the original javplayer, the mod should be easy to run as well.
I state this mod is completely free. Just for learning and communication purposes only.Do not use it for any commercial purposes.
1.Add TopazVideoEnhanceAI enhancement function.
2.Support TecoGAN 2X 8X models and Secondary-enhancement model SE-2X,SE-4X (insufficient training).
3.Support Denosie Anti-flicker, but time-consuming and will reduce sharpness.
4.Support patch_tools for remove the original watermark.
5.Support safe mode picture quality enhancement.
6.Support esrgan model.
7.Support mPlus custom command mode.
8.Support 3000 series card (Requires cuda11, cudnn8.0.5 environment)
9.Python script extensions supported
This Mod base on TG-models, No matter SE-mode or TVE-mode, improved few.
Basically, the good will be better, the bad will be worse.
mod.ini #TG-PLUS mplus command line config file.
veai.ini #veai custom dictionary file.
(located in javplayer_109a/TG directory)
Function:scale:model-name.
a. tecogan esrgan tve. e.g., m1:tecogan:4:TG-AF1 , m2:esrgan:2:esrgan-2x_SHARP .
(The scale should be consistent with the model e.g., tecogan:2:SE-2X,TVE model can set to 2,3,4,6 e.g., tve:6:ahq)
b. denoise, no parameters. e.g., m2:denoise .
(Processing image noise, occupy cpu memory)
c. downscale. e.g.,downscale:2 .
(you can choose to reduce the ratio, speedup and Avoid OOM.)
d.res. e.g., res:1280 .
(images will be reduced first, Limit maximum output to 1280.)
1.use v1.9 parameters, to adapt the version of tve.
Default support for v1.5 1.7 1.8 1.9 veai cmd, use e.g. v1.9 in mod.ini to specify the tve version.
And generate veai.ini file in tg/ directory when tve is first called. for future updates of tve version,
use custom dictionary e.g. "ahq": "ahq-11"...,and delete v1.9 flag or modify it to v2.x.
! tve 1.7 1.8 does not support tiny images well and cannot be used for first step.(dont use for m1).
! tve 1.7 has many issues, not recommended.
2. esrgan model
! Use esrgan-ncnn-vulkan.exe to execute the standard model, You can use pth2ncnn.exe xxx.pth to convert the model.
put into TGMAIN\mplus\esrgan\models\ .
Note: If you are interested ,you can training standard model or secondary enhancement model.
3. mPlus custom command mode.
You can use mplus mode to custom execution.
! can be customized to order and model.
! model-name must be case sensitive.
! #The first line TG-PLUS is the enhanced mode flag, modify it to TG to turn off and ignore all command lines.
TG-PLUS
# This is a standard profile
# Note the TG-PLUS line cannot be removed.
M1:tecogan:4:TG-STD # Use tecogan TG-STD model
M2:denoise # denoise
M3:tecogan:4:SE-4X # Use SE-4X model for SE-enhance.
M4:denoise # denoise
M5:tve:4:ahq # tve enhanced, 4x, use ahq model.
res:1920 # Limit max resolution to 1920.
v1.9 # Use the command line of tve 1.9
TG-PLUS
# This is an example of functional demo that just show the custom command line usage.
# Note the TG-PLUS line cannot be removed.
m1:esrgan:2:esrgan-2x_SHARP # esrgan-2x_SHARP is a pth2ncnn.exe conversion model, model name case consistency.
m2:denoise # denoise
m3:tecogan::TG-AF1 # scale_factor is empty, :: scale is null use default 4, and TG-AF1 is used in se-mode.
m4:denoise
m5:denoise # m4,m5 If the noise is too serious, you can reduce the noise several times like this.
m6:esrgan:2: # esrgan model name is empty, model name is null,so 2x model will be used
m7:downscale:2 # Reduce the images in the input directory by 2 times.
m8:tve:2:thd # tve enhanced, 2x,use thd model, and affected by res:960 will downscale first
res:960 # Limit max output resolution to 960.
v1.9 # Use the command line of tve 1.9
- python script extensions are supported, will be useful. Extend your own functionality and implement your own ideas.
You can call python script. e.g., m1:mPlus\call-realsr.py. (checkin TGMAIN\mplus) .
the pass the arguments: -i inputdir -o ouputdir -m model -s sacle -se_mode se_mode
1.TG-Plus Packs. (support 3000 series)
https://thup.work/uploader/?mode=dl&id=24596&original=1
https://thup.work/uploader/?mode=dl&id=24597&original=1
2.runGan_veai_custom_patch
https://thup.work/uploader/?mode=dl&id=24710&original=1
Overwrite the source files in the tgmain directory, backup first.
3.download TG-MODEL
https://thup.work/uploader/?mode=dl&id=25181&original=1
4.Video Quality patch for 1.08 safe mode.
https://thup.work/uploader/?mode=dl&id=25758&original=1
5.Javplayer_watermark_patch 102 - 109a
https://thup.work/uploader/?mode=dl&id=22129
6.TG-PLUS_ConfigTool_0720 fix bugs
https://thup.work/uploader/?mode=dl&id=26488&original=1
7.Update CPU issues.
Because the cuda runtime support libraries could not be found in the system environment paths.
The solution is find these dlls from the system copy into the TGMAIN directory.
https://thup.work/uploader/?mode=dl&id=26420&original=1
cublas64_10.dll cudart64_101.dll cudnn64_7.dll cufft64_10.dll curand64_10.dll cusolver64_10.dll cusparse64_10.dll
1.install Topaz Video Enhance AI,copy "X:\Program Files\Topaz Labs LLC\Topaz Video Enhance AI"
to Javplyer_109a main folder. rename the folder to TVE.
2.backup you TG folder,rename it to TG.BAK .
3.download TG-Plus_3000.rar(2 packs) , unzip the file to javplayer_109a TG folder.
4.copy the backup TG-AF1 TG-AF2 . to TG/TG-MODEL .
5.download TG-MODEL.rar(SE-2X,SE-4X..), unzip the file to TG/TG-MODEL.
6.download Javplayer_watermark_patch,copy xxx\level0 to JavPlayer_Data.
7.download runGan_veai_custom_patch,Overwrite the source files in the tgmain directory, backup first.
8.put mod.ini into javplayer_109a TG folder.
9.download TG-Plus Configuration Tool,unzip to javplayer_109a main folder.
