Executable link fails because /SUBSYSTEM:WINDOWS is defined when add_executable() has WIN32 argument.

Suggested fix:

• Remove WIN32 from add_executable(${PROJECT_NAME} WIN32 ${sources}) call
• Remove set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} /SUBSYSTEM:CONSOLE") line because this flag is controlled by the presence of WIN32 argument as stated above

Thanks for this open-source example. Unfortunately, while reading the README file, I had to really work hard to determine what this tool would even do. Only after searching the AMD website, I came to the realization of what "FidelityFX Super Resolution (FSR)" and "Contrast Adaptive Sharpening (CAS)" would even imply.

While I don't mean to dismiss the importance of AMD's efforts with regard to open-source technologies, I think some improvements to the manner in which these are presented to third parties would be welcome. Perhaps the README could be enhanced to clearly demonstrate and sell the utility of this tool? Even links to the AMD website or relevant Youtube videos might be use?

It do not generate output even when pass with no warning or errors... it pretend that it save destination but nothing... 1.0.2 it was okay

tried in both modes: FSR1 and CAS

it also do not provide option to upscale more then 2x... where it say 4x is recommended :-) 1.0.2 didn't have issue with 4x upscale, now tried it with 8x even 20x and works like charm

it start fail with 25x (FidelityFX_CLI.cpp(657): FAILED( dev->CreateTexture2D(&dstTextureDesc, nullptr, &dstTexture) ))

so it is very very big downgrade about scalability :-(

Treating input/output images as sRGB space, processing as sRGB.

Loading "312981607_5590160864395316_8625555753732413662_n.png"... Image size: 720 x 515 scaled to 2880 x 2060. WARNING: SCALING RATIO HIGHER THAN 2X IN EITHER DIRECTION IS UNSUPPORTED! WARNING: Scaling factor is greater than recommended 4. Saving "312981607_5590160864395316_8625555753732413662_n-FSR1.png"...

but even with no upscale it do not generate output C:\FidelityFX-CLI-v1.0.3>

./FidelityFX_CLI.exe -Mode FSR1 -Scale 100% 100% -QualityMode UltraQuality 312981607_5590160864395316_8625555753732413662_n.png 312981607_5590160864395316_8625555753732413662_n-FSR1.png Treating input/output images as sRGB space, processing as sRGB. Loading "312981607_5590160864395316_8625555753732413662_n.png"... Image size: 720 x 515 scaled to 936 x 670. Saving "312981607_5590160864395316_8625555753732413662_n-FSR1.png"...

pls note that with 100% scale it change scaling as you can see from Image size, very crazy behavior :-( tried with no QualityMode it do not change Image Size...

never save anything :-(

Hello.

I like to use tool without storing temporary files for video transcoding. So, piped IO is required.

For example vmaf tool on Windows can be used like this:

D:\vmaf\2.3.0\vmaf.exe --reference \\.\pipe\Process_Pipe_20900_33_1121943890000000 --distorted \\.\pipe\Process_Pipe_20900_34_1121943890000000 --width 3840 --height 2160 --pixel_format 420 --bitdepth 8 --json --output R:\Temp\g3o8le0l.tmp\outFile

Or something like image2pipe from ffmpeg (see ffmpeg faq):

cat *.jpg | ffmpeg -f image2pipe -c:v mjpeg -i - output.mpg

Thanks!

It's possible to use a loop to upscale a folder full of images, but this is pretty slow for what it could be. Could you make it able to work on folders, probably by not restarting itself between frames?

Helpful for: Upscaling a video's worth of frames Upscaling a lot of images that don't relate to each other. This would be drastically sped up in either case.

Thanks in advance regardless of decision.

when trying to run the following command:

FidelityFX_CLI.exe image.png out.png

i get this error message:

ERROR: C:\GitLab-Runner\builds\d617e10a\0\gpuopen\GPUOpen-Effects\fidelityfx-cli\FidelityFX_CLI.cpp(606): FAILED(CoCreateInstance(CLSID_WICImagingFactory, NULL, CLSCTX_INPROC_SERVER, IID_PPV_ARGS(&m_WICImagingFactory)) )

i'm using the latest binaries currently available (FidelityFX-CLI-v1.0.3.zip), latest Visual C++ 2022 runtimes, a known good PNG image (640x480 res), and the latest NVIDIA drivers available for Windows 7.

##GUI Assistance

Prefix: The prefix or word before the number sequence starts.

Starting Number: number of first image after prefix

Ending Number: number of last image after prefix

Output Prefix: the prefix all the output images will have

Input Extension: aka image type Supported Formats: BMP, PNG, ICO, JPG, TIF, GIF

For example, to upscale two images which are render0.png and render1.png and get output as ups0.png and ups1.png these are the settings required:

Prefix: render Starting Number: 0 Ending Number: 1 Output Prefix: ups Input Extension: .png

Protip: Use absolute or relative directories in both prefix fields to get input and store output in different folder.

Does not create folders automatically. Only use existing folders.

Hi Dear @rys :

it seems can use VS2019 to build this project, but I do not find the .sln file, it is only have a CMakeList.txt, so Could you please help me: How to build in windows?

thank you very much~