The Gaming Restoration
Resident Evil 3.5 Trailer / Gameplay (Best Quality)
updated
Remastered in 4K using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I used ESRGAN with alsa64's denoising model and the Manga109 model to upscale the video 4 times its original resolution.
Original resolution: 640x448
[Resolution based on PCSX2’s internal buffer;
SDTV 480p Progressive Scan output]
ESRGAN Upscale: 2560x1792 --) 2560x1920
-----------------------------
THE COLORIZATION PROCESS INTRODUCES SOME
FLICKERING AND COLOR-INSTABILITY.
IF YOU EXPERIENCE DISCOMFORT WATCHING THIS VIDEO, OR
HAVE A HISTORY OF EPILEPSY OR SEIZURES, PLEASE PAUSE
THE VIDEO AND LOWER THE BRIGHTNESS OF YOUR SCREEN.
Fallout 4 - All Endings
Colorized using DeOldify [AI Machine Learning] [IMPRESSIVE RESULTS]
DeOldify is an AI Neural Network software that is capable of colorizing black and white images using thousands of pictures as references.
I took the original video files from the Fallout 4 PC disk, and used DeOldify with the Video training model to restore the color from the FMVs.
00:26 - Brotherhood/Railroad Ending (Male Character)
02:08 - Institute/Minutemen Ending (Male Character)
MOD DOWNLOAD LINK
----------------------------------
nexusmods.com/fallout4/mods/40364
Remastered in 1440p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining
most of their elements and details intact.
I took the original video file from the Mortal Kombat: Armageddon Wii disk, and used ESRGAN with alsa64's denoising model
and the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 736x480 --) 736x304
ESRGAN Upscale: 2944x1920 --) 2944x1216 --) 2560x1158
(Non-standard aspect ratio)
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Resident Evil - Code: Veronica GC disk, and used ESRGAN with alsa64's dedither model and the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 320x320
ESRGAN Upscale: 1280x1280 → 1440x1080
Physical RAM: 6GB
CPU: Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz
GPU: Nvidia GT 1030 2GB (GDDR5)
--------------------------------------------------------------
PSP (PPSSPP)
------------
PPSSPP v1.8.0-433-g37a97e741
Backend: Vulkan
Processing shader: FXAA Antialiasing
Rendering resolution: Auto (1:1) [1080p]
Hardware tessellation: Yes
Anisotropic filtering: 16x
Force real clock sync (slower, less lag): Yes [probably not needed]
Change emulated PSP's CPU clock (unstable): 1000MHz
60FPS Hack code:
(made by Kabuto_Kun and ported by VIRGIN KLM)
ULUS10160.ini
------------------------
_S ULUS-10160
_G Grand Theft Auto - Vice City Stories [USA]
_C1 60 FPS
_L 0x202070C8 0x00000000
_C0 30 FPS [Normal]
_L 0x202070C8 0x1000FFF7
--------------------------------------------------------------
PS2 PORT (PCSX2)
----------------
PCSX2 1.5.0-20190717090803 - compiled on Jul 17 2019
GSdX OpenGL (Hardware) -- Rendered at 3x Native Internal Resolution (1920x1536)
MTVU + Fxaa shader + 60fps hack
PCSX2 Config
------------
MTVU On
GSdx
OpenGL (Hardware)
Internal Resolution 3x Native
Automatic CRC Level Partial (OpenGL Recommended)
Blending Unit Accuracy High
Shader Configuration --) Enable FXAA
180%-300% EE cyclerate overclock for 60fps gameplay
60FPS Hack code:
(made by asasega)
4F32A11F.pnach
------------------------
gametitle=Grand Theft Auto: Vice City Stories SLUS-21590
//60 fps
patch=1,EE,20370314,word,00000000
Remastered in 4K using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Mortal Kombat: Deception PS2 disk, and used ESRGAN with alsa64's denoising model and the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 640x448
ESRGAN Upscale: 2560x1792 --) 2560x1920
Remastered in 4K using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
Since Tekken Tag Tournament 2 FMV files are encrypted, I had to record the opening cutscene using RPCS3.
I used ESRGAN with alsa64's denoising model and the Manga109 model to upscale the video 3 times its original resolution.
Original resolution: 1280x720
ESRGAN Upscale: 3840x2160
Colorized using DeOldify [AI Machine Learning] [IMPRESSIVE RESULTS]
DeOldify is an AI Neural Network software that is capable of colorizing black and white images using thousands of pictures as references.
I took the original video file from the Fallout 4 PC disk, and used DeOldify with the Video training model to restore the
color from the FMV.
This is my second attempt at colorizing this video, and the results are much more natural-looking and stable.
EPILEPSY WARNING
----------------
THE COLORIZATION PROCESS INTRODUCES SOME FLICKERING AND COLOR-INSTABILITY.
IF YOU EXPERIENCE DISCOMFORT WATCHING THIS VIDEO, OR
HAVE A HISTORY OF EPILEPSY OR SEIZURES, PLEASE PAUSE
THE VIDEO AND LOWER THE BRIGHTNESS OF YOUR SCREEN.
Using ESRGAN to fully reconstruct videos [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructinglow-resolution images, while retaining most of their elements and details intact.
This is an experiment I did to show what happens when an AI tries to reconstruct a video with almost no pixels or information in it.
Basically, I took some of the FMVs from the original Silent Hill 2 disk, and downscaled them by ~80%. Then I grabbed a training model that has dozens of references of human faces, and used ESRGAN to upscale the footage 8 times its original resolution.
The results are pretty abstract and disturbing, and they almost resemble the aesthetics of Francis Bacon (Masahiro Ito's favorite painter, whose works served as inspiration for the monsters designs)
Original resolution: 512x384 → 64x48
ESRGAN Upscale: 1024x768 → 1440x1080
(Re-upscaled using Lanczos and temporal sharpening)
Remastered in 8K using ESRGAN [AI Machine Learning] [UNBELIEVABLE RESULTS!]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original uncompressed trailer that CD Projekt released back in 2013, and used ESRGAN with the 4xBox model to upscale the video 4 times its original resolution.
After so many months of experimenting with different types of videos and training models, I can finally say with confidence that these are the best results so far.
The image is so sharp and detailed, and the details are so rich and vivid, that you can even appreciate the micro-details in every character. The wool of the clothes, the skin pores and imperfections
in the faces, the hair of Geralt and the lady... You can even notice the noise caused by the depth of field filter and the sampling in the render itself.
I invite everyone, specially the ones that don't have an actual 8K monitor, to download the video and zoom in the image to fully appreciate all these small details.
Huge thanks to buildist for this amazing training model.
Original resolution: 1920x800
ESRGAN Upscale: 7680x3200
Remastered in 4K using ESRGAN [AI Machine Learning] [IMPRESSIVE RESULTS]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Dirge of Cerberus: Final Fantasy VII PS2 disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 640x448
ESRGAN Upscale: 2560x1792 --) 2560x1920
Remastered in 4K using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from one of the Official U.S. PlayStation Magazine Demo disks, and used ESRGAN with alsa64's denoising model and the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 720x480
ESRGAN Upscale: 2880x1920
Remastered in 4K using ESRGAN [AI Machine Learning] [IMPRESSIVE RESULTS]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original uncompressed trailer that CD Projekt released back in 2013, and used ESRGAN with the Manga109 model to upscale the video 2 times its original resolution.
While the software is perfectly capable of upscaling 1080p images to 8K [7680x3200], the current models weren't trained for such high resolutions.
Once the image reaches a certain amount of pixels, the AI can't improve the details any further.
Original resolution: 1920x800
ESRGAN Upscale: 3840x1600
Logo made by crepet9000 (DeviantArt): deviantart.com/crepet9000/art/Cyberpunk-2077-Logo-764106006
Retroarch 1.7.7 (commit 57c373e1d7)
beetle-psx-libretro HW 0.9.44.1 (commit aae54be)
Vulkan Renderer
-Warping Textures fix [PGXP]
-Widescreen mode hack [Retroarch]
-Increased CD loading speed
-Native support for Mouse + Keyboard
-xBR Texture Filtering
OS: Linux 5.2.0-1-rc4-tkg-cfs
Physical RAM: 6GB
CPU: Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz
GPU: Nvidia GT 1030 2GB (GDDR5)
Config Tab
---------
Renderer: hardware
Software framebuffer: ON
Adaptive smoothing: ON
MSAA: 4x
MDEC YUV Chroma filter: ON
Internal GPU resolution: 8x
Texture filtering: xBR
Internal color depth: 32bpp
PGXP operation mode: memory + CPU
PGXP vertex cache: OFF
PGXP perspective correct texturing: ON
Widescreen mode hack: ON
CPU frequency scaling (overclock): 400% [PLAYABLE WITH LOWER VALUES]
Dithering pattern: OFF
CD Access Method: async
Increase CD loading speed: 14x
Japanese Audio [Optional English subtitles]
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the PSP disk of The 3rd Birthday, and used ESRGAN with alsa64's denoising model [JPG (20-40)] and the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 480x272
ESRGAN Upscale: 1920x1088 → 1920x1080
Remastered in 4K using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
El Tío Rata took the original video file from an Interactive Demo Disc. I used ESRGAN with twittman's antialiasing model, alsa64's denoising model [JPG (0-20%)], and the Manga109 model to upscale the video 4 times its original resolution.
Original resolution: 640x480
ESRGAN Upscale: 2560x1920
Remastered in 4K using ESRGAN [AI Machine Learning] [GREAT RESULTS]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the video file from the Kikizo Archives, and used ESRGAN with alsa64's denoising model and the Manga109 training model to upscale the video 4 times its original
resolution.
The audio was taken from IGN's footage, which featured the original audio from the trailer (instead of a recording from the E3 conference).
Original resolution: 640x480
ESRGAN Upscale: 2560x1920
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Tekken 3 PS1 disk, and used ESRGAN with alsa64's denoising model [JPG (0-20%)] andthe Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 256x224
ESRGAN Upscale: 1024x896 → 1440x1080
Remastered in 4K using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Biohazard 4 Secret DVD disk, and used ESRGAN with alsa64's denoising model [JPG (0-20%)] and the Manga109 model to upscale the video 4 times its original resolution.
In order to reduce the aliasing and motion blur, I tried to apply various filters (like Nnedi3, QTGMC and FFT3DFilter, etc). Obviously, the gameplay doesn't look like native 4K, but it does look noticeably sharper (it could even pass for a 720p PS3/X360 title).
Original resolution: 720x480
ESRGAN Upscale: 2880x1920
OS: Linux 5.2.0-1-rc2-tkg-cfs
Physical RAM: 6GB
CPU: Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz
GPU: Nvidia GT 1030 2GB (GDDR5)
NOT RECORDED IN REAL TIME
GSdX OpenGL (Hardware) -- Rendered at 3x Native Internal Resolution (1536x1344)
MTVU + Fxaa shader + 60fps hack + Maori-Jigglypuff patch
PCSX2 Config
------------
MTVU On
GSdx
OpenGL (Hardware)
Internal Resolution 3x Native
Automatic CRC Level Partial (OpenGL Recommended)
Blending Unit Accuracy High (Recommended high-end PC)
Shader Configuration --) Enable FXAA
180%-300% EE cyclerate overclock for 60fps gameplay
Maori-Jigglypuff patch
forums.pcsx2.net/Thread-Post-your-PCSX2-cheats-patches-here?pid=589325#pid589325
DIRECT LINK: drive.google.com/uc?export=download&id=1tjYL7NZwwLbGuuSUmzxxfRZtzMwzvuTi
GOOGLE DRIVE ONLY STREAMS AT 1080P, YOU HAVE TO DOWNLOAD THE ORIGINAL FILE IN ORDER TO WATCH IT IN 4K.
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Linkin Park - Breaking The Habit DVD, and used ESRGAN with alsa64's denoising model [JPG (0-20%)] and the Manga109 training model to upscale the video 4 times its original resolution.
The audio track was taken from hdtracks.com, and it was mixed at 96kHz with 24 bits per sample.
Original resolution: 720x480 (Deinterlaced using QTGMC)
ESRGAN Upscale: 2880x1920 --) 2560x1920
Remastered in 1080p using ESRGAN [AI Machine Learning]
2:30 Chapter 1: Ring the Church Bell
5:39 Chapter 2: Rescue Luis
13:16 Chapter 3: Retrieve the Sample
18:58 Chapter 4: Stop Leon's Assassination
22:51 Chapter 5: Obtain the Sample
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video files from the Resident Evil 4 PS2 disk, and used ESRGAN to upscale them 4 times their original resolution.
This time I used 3 different training models: alsa64's denoising model [JPG (0-20%)]; the Manga109 training model; and my own, which I made using the Dolphin Emulator (I took HD screenshots of several cutscenes from the main campaign, and used them to
train a new model).
Original resolution: 512x336 / 320x208
ESRGAN Upscale: 2048x1344 / 1280x832 --) 2560x1440
(Upscaled to 1440p with lanczos to preserve video quality on YouTube)
Next goal: Replacing the FMVs from the PC version with these ones.
Remastered in 4K using ESRGAN [AI Machine Learning] [IMPRESSIVE RESULTS]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Gran Turismo 4 PS2 disk, and
used ESRGAN with the Manga109 training model to upscale the videos 4 times its original resolution.
Big thanks to Gyan (from stackexchange.com) for helping me to transcode this video, and Mr T. (YouTube) for recommending me to use QTGMC.
Original resolution: 640x448
ESRGAN Upscale: 2560x1792 --) 2560x1920
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing
low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Galerians PS1 disk, and used ESRGAN with
the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 320x176
ESRGAN Upscale: 1280x704 → 1440x1080
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing
low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Beast Wars: Transformers PS1 disk, and used ESRGAN with
the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing
low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Age of Empires: Gold Edition PC disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Age of Empires uses the Indeo Video Interactive 4 codec for its FMVs, which (seems to) have
a low color depth and ordered dithering. As a result, the upscaled footage looks a bit foggy and grainy.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080
Remastered in 4K using ESRGAN [AI Machine Learning] [IMPRESSIVE RESULTS]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing
low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Gran Turismo Concept 2001 TOKYO PS2 disk, and
used ESRGAN with the Manga109 training model to upscale the videos 4 times its original
resolution.
Original resolution: 640x448
ESRGAN Upscale: 2560x1792 --) 2560x1920
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing
low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Duke Nukem: Time To Kill PS1 disk, and used ESRGAN with
the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080
OS: Linux 5.0.8-20-tkg-pds
Physical RAM: 6GB
CPU: Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz
GPU: Nvidia GT 1030 2GB
NOT RECORDED IN REAL TIME
GSdX OpenGL (Hardware) -- Rendered at 6x Native Internal Resolution (3840x3072)
MTVU + Fxaa shader + No deintarlacing hack + Widescreen hack (plus fixes)
PCSX2 Config
------------
MTVU On
GSdx
OpenGL (Hardware)
Internal Resolution 6x Native
Automatic CRC Level Partial (OpenGL Recommended)
Blending Unit Accuracy High (Recommended high-end PC)
Shader Configuration --) Enable FXAA
60FPS Hack code:
339A0B8C.pnach
------------------------
gametitle=Fatal Frame (SLUS-20388) 0x339A0B8C
//Disable interlacing
patch=1,EE,2023188C,word,00000000
**FILTERS & WIDESCREEN HACK CODES ARE ALREADY INCLUDED IN THE PCSX2 EMULATOR
Remastered in 4K using ESRGAN [AI Machine Learning]
2nd Attempt [IMPRESSIVE RESULTS]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Capcom Special DVD Tokyo Game Show 2003 disk, and used ESRGAN with the Manga109 training model to upscale the videos 4 times its original resolution…
This time the results were beautiful. The footage I found in the DVD was twice as big and detailed as the original PS2 opening video file, and there were barely any compression artifacts.
Original resolution: 720x480
ESRGAN Upscale: 2880x1920 --) 2560x1920
Remastered in 4K using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructinglow-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Silent Hill - Art of Silent Hill DVD, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 720x480
ESRGAN Upscale: 2880x1920 --) 2560x1920
Colorized using Colorful Image Colorization [AI Machine Learning]
Colorful Image Colorization is an AI Neural Network software that is capable of colorizing black and white images using thousands of pictures as references.
I took the original video file from the Fallout 4 PC disk, and used Colorful Image Colorization with the colorization_release_v2 training model to restore the color from the FMV.
NOTE: Keep in mind that colorizing black and white footage using reference images is a hit-and-miss process, and the AI software has to (literally) guess which colors go where. As a consequence, the colorized footage might look quite surreal and psychedelic.
crimson-ceremony.net/lostreleases/item.php?id=redemo_recvx-wr-jpn
Remastered in 1080p using ESRGAN & Colorful Image Colorization [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
Colorful Image Colorization is another AI Neural Network software that is capable of colorizing black and white images using thousands of pictures as references.
I took the original video file from the Tenchu: Stealth Assassins PS1 disk, used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution, and used Colorful Image Colorization with the colorization_release_v2 training model to restore the color from the FMV.
I also replaced the original music audio track with the remastered version found in the official Tenchu Soundtrack CD.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080
Original B/W Version: 2:22
Remastered in 1080p using ESRGAN & Colorful Image Colorization [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
Colorful Image Colorization is another AI Neural Network software that is capable of colorizing black and white images using thousands of pictures as references.
I took the original video file from the Tenchu: Stealth Assassins PS1 disk, used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution, and used Colorful Image Colorization with the colorization_release_v2 training model to restore the color from the FMV.
I also replaced the original music audio track with the remastered version found in the official Tenchu Soundtrack CD.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080
Original B/W Version: 6:11
OS: Linux 5.0.8-20-tkg-pds
Physical RAM: 6GB
CPU: Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz
GPU: Nvidia GT 1030 2GB
NOT RECORDED IN REAL TIME
GSdX OpenGL (Hardware) -- Rendered at 3x Native Internal Resolution (1920x1344)
MTVU + Fxaa shader + 60fps hack + Widescreen hack
PCSX2 Config
------------
MTVU On
GSdx
OpenGL (Hardware)
Internal Resolution 3x Native
Automatic CRC Level Partial (OpenGL Recommended)
Blending Unit Accuracy High (Recommended high-end PC)
Shader Configuration --) Enable FXAA
180%-300% EE cyclerate overclock for 60fps gameplay
60FPS Hack code:
0F6FC6CF.pnach
------------------------
gametitle=SOCOM II U.S. Navy SEALs SLUS-97275 0x0F6FC6CF
//60 fps
patch=1,EE,D03E1400,extended,0000001E
patch=1,EE,203E1400,extended,0000003C
**WIDESCREEN HACK CODES ARE ALREADY INCLUDED IN THE PCSX2 EMULATOR
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements
and details intact.
I took the original video file from the Silent Hill 2 PS2 disk (the original NTSC-U release), used ESRGAN with the Manga109 training model to upscale the video 4 times its
original resolution, and applied a few filters to restore the color from the first half of the FMV (the videotape has actual color, but it's heavily desaturated).
I also included the original black and white version.
The uncensored audio was taken from the original PS2 disk files.
Original resolution: 512x384
ESRGAN Upscale: 2048x1536 --) 2048x1152 (Unknown Aspect Ratio)
1:38 - Original B/W Version / Original Audio
Remastered in 1440p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructinglow-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Dog's Life PS2 disk, and used ESRGAN withthe Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 720x480
ESRGAN Upscale: 2880x1920 → 2560x1440
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Onimusha 3 PS2 disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Unfortunately, the video had a lot of noise and film grain, and I was unable to remove it without downgrading the quality of the visuals
(there are too many small details in the characters). The end results still look sharp and defined, but really dirty.
Original resolution: 512x352
ESRGAN Upscale: 2048x1408 --) 2048x1152 [5:3 Aspect Ratio]
ORIGINAL 5.1 JAPANESE AUDIO TRACK AVAILABLE
----------------------------------
I was able to extract the original Dolby Digital 5.1 audio track from the opening FMV. If you have a multichannel setup or a pair of surround headphones, I highly recommend you to try this out:
LINUX SCRIPT (uses mpv or vlc to play the video): drive.google.com/uc?export=download&id=1xC2XhEkwWxV1uHTd4yU-T416m4Zyj2L0
WINDOWS SCRIPT: Not made yet.
Original 5.1 Japanese Audio Track (Google Drive):
drive.google.com/uc?export=download&id=1EO0BPP8_2jLjf3wiU7m_4l5qVOcbSuAf
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Tekken PS1 disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Tekken 2 PS1 disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Tekken 2 PS1 disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080
OS: Linux 5.0.1-3-tkg-pds
Physical RAM: 6GB
CPU: Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz
GPU: Nvidia GT 1030 2GB
NOT RECORDED IN REAL TIME
GSdX OpenGL (Hardware) -- Rendered at 3x Native Internal Resolution (2382x1536)
MTVU + Fxaa shader + 60fps hack + Progressive scan hack (480p) + Widescreen hack
PCSX2 Config
------------
MTVU On
GSdx
OpenGL (Hardware)
Internal Resolution 3x Native
Automatic CRC Level Partial (OpenGL Recommended)
Blending Unit Accuracy High (Recommended high-end PC)
Shader Configuration --) Enable FXAA
180%-300% EE cyclerate overclock for 60fps gameplay
60FPS Hack code:
327053E8.pnach
------------------------
gametitle=Darkwatch SLUS-21042 0x327053E8
//60 fps
patch=1,EE,0071FA68,word,00000001
**WIDESCREEN + PROGRESSIVE SCAN CODES ARE ALREADY INCLUDED IN THE PCSX2 EMULATOR
OS: Linux 5.0.1-3-tkg-pds
Physical RAM: 6GB
CPU: Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz
GPU: Nvidia GT 1030 2GB
NOT RECORDED IN REAL TIME
GSdX OpenGL (Hardware) -- Rendered at 3x Native Internal Resolution (3072x3072)
MTVU + Fxaa shader + 60fps hack + Progressive scan hack (480p) + Widescreen hack
PCSX2 Config
------------
MTVU On
GSdx
OpenGL (Hardware)
Internal Resolution 3x Native
Automatic CRC Level Partial (OpenGL Recommended)
Blending Unit Accuracy High (Recommended high-end PC)
Shader Configuration --) Enable FXAA
130%-180% EE cyclerate overclock for 60fps gameplay
60FPS Hack code:
D6D704BB.pnach
------------------------
gametitle=Cold Winter SLUS-20845 0xD6D704BB
//60 fps
patch=1,EE,20DC2ECC,word,00000001
**WIDESCREEN + PROGRESSIVE SCAN CODES ARE ALREADY INCLUDED IN THE PCSX2 EMULATOR
Remastered in 4K using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Biohazard Outbreak PS2 disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Since Tekken 5 FMV files are encrypted, I had to record the opening cutscene using PCSX2.
I used AutoDeblock, FineDehalo, KNLMeansCL and LSFmod to remove some noise, sharp the edges and reduce the compression artifacts. Then I used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution… And the results are beautiful.
Original resolution: 640x448
[Resolution based on PCSX2’s internal buffer; SDTV 480p Progressive Scan output]
ESRGAN Upscale: 2560x1920
04:06 ESRGAN vs Original
PRESS THE CC BUTTON FOR LYRICS & SUBTITLES IN ENGLISH
ESRGAN: github.com/xinntao/ESRGAN
Manga109Attempt.pth: http://www.mediafire.com/file/w3jujtm752hvdj1/Manga109Attempt.pth.zip/file
fvsfunc (AutoDeblock): github.com/Irrational-Encoding-Wizardry/fvsfunc
FineDehalo: http://avisynth.nl/index.php/FineDehalo
KNLMeansCL: github.com/Khanattila/KNLMeansCL
LSFmod: forum.doom9.org/showthread.php?t=142706
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the PS2 port of Quake III (Quake III: Revolution), and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 512x512
ESRGAN Upscale: 2048x2048 → 2048x1152
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the original Quake II PC disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080
Remastered in 1440p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructing low-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Soul Calibur III PS2 disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 640x448
ESRGAN Upscale: 2560x1792
ORIGINAL 5.1 AUDIO TRACK AVAILABLE
----------------------------------
I was able to extract the original Dolby Digital 5.1 audio track from the opening FMV. If you have a multichannel setup or a pair of surround headphones, I highly recommend you to try this out:
LINUX SCRIPT (uses mpv or vlc to play the video): drive.google.com/uc?export=download&id=15M6Ym8xWeZT8_HYs1KGsU2hzmKmhsde0
WINDOWS SCRIPT: Not made yet.
Original 5.1 Audio Track (Google Drive):
drive.google.com/uc?export=download&id=1PI8BH6M7ywxepnamKJ-bcujdY-dTLJs6
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructinglow-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Parasite Eve PS1 disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080
Remastered in 1080p using ESRGAN [AI Machine Learning]
ESRGAN is an AI Neural Network software that is capable of upscaling and reconstructinglow-resolution images, while retaining most of their elements and details intact.
I took the original video file from the Parasite Eve 2 PS1 disk, and used ESRGAN with the Manga109 training model to upscale the video 4 times its original resolution.
Original resolution: 320x240
ESRGAN Upscale: 1280x960 → 1440x1080


