@fifonik Ok, so far the Calculate button is greyed out. Ok fixed, forgot to add the source file, I'm definitely getting too old for this :D.

.

Small other detail, although you mention .NET Framework requirements in your post here, its not in the "readme.txt" requirements, needs to be added there.

Checking the .NET framework v4 can be made easier using this batch file. Just copy and paste the following into a batch file eg call it GetNETv4.bat ...

@echo off
cls
reg query "HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\NET Framework Setup\NDP\v4\full" /v version
pause

> Calculate button is greyed out.

You need to add reference file as well (notice the red input line).

> Checking the .NET framework v4 can be made easier

Program checks installed .NET version and add warning in program log. It should suggest to install it if needed. However, I do not have Win 7 to check if this works so I've added this text as pre-cautions. I've added information about .NET requirements in the readme.txt as well. Will be in next "release". Thanks.

@fifonik

So far, very nice. The screenshot below is your saved log file with the same file I tested in my batch file

I use a colon delimiter so as to easily import into a spreadsheet. I normally do SSIM, PSNR, VMAF so I rearranged the line to match yours. You can see that the SSIM and PSNR values are identical but the VMAF values are different.

I cannot comment about different values of VMAF. You should have ffmpeg commands in log. Feel free to run them manually and/or compary with your commands you use.

As per colon-separator. There are a number of different flavours of csv files. I'm using that is known as "tab delimited". It is also can be imported in Excel if you choose this type in dropdown. If I remember correctly, it even automatically suggest it (cannot check as I do not have Excel at my home PC). I prefer this one as it is also much easier to read.

@fifonik I checked my test files, items, 19, 20 and 21 against using your ffmetrics, again, maybe worth to mention that the VMAF are different, all three of my results are a little smaller.

I really like the progress feedback, frames processed per SSIM etc, cool!

This is my output from Batch file, formatted for SS import. Colon delimited.

@fifonik Program checks installed .NET version and add warning in program log. It should suggest to install it if needed.

Excellent, perhaps add that also to your readme, save users like me doing it the hard way.

Indeed, the TAB delimited is easier to read.

@fifonik I assume that its only available for FHD, not 4K VMAF?

> I assume that its only available for FHD, not 4K VMAF?

Oh, you are right. The included model is for FHD. I will think about 4K.

@fifonik "I cannot comment about different values of VMAF. You should have ffmpeg commands in log. Feel free to run them manually and/or compary with your commands you use."

Having the commands in the log is excellent. There must be some difference between your version and the one I'm using, yours really, thanks.

I am trying to figure it out, maybe you could take a look at these two lines below. They are obviously giving two slightly different results for VMAF.

The first one is yours, Source and test file, %1 changed already to match mine. I'm outputting to a different file.

ffmpeg -i %1 -i Source.mp4 -lavfi "[0:v]setpts=PTS-STARTPTS[main];[1:v]setpts=PTS-STARTPTS[ref];[main][ref]libvmaf='log_fmt=json:eof_action=endall:model_path=D\:/RQM/00 ffmetrics/vmaf_v0.6.1.pkl:log_path=C\:/Users/John/AppData/Local/Temp/FFMetrics/%1_75641112-c517-4da5-83e0-5c023cf7b7d6.log'" -f null -

ffmpeg  -i %1 -i Source.mp4 -lavfi "[0:v]setpts=PTS-STARTPTS[main];[1:v]setpts=PTS-STARTPTS[ref];[main][ref]libvmaf=model_path=vmaf_v0.6.1.pkl:pool=harmonic_mean:eof_action=endall"  2>"ffmpeg VMAF.txt" -f null -

Basically, what do I need to change in the second syntax to give similar results to yours?

@fifonik Maybe its the "Harmonic Mean"?, I'll check that by leaving it out, hold that thought, I''ll be back.

The difference is: pool=harmonic_mean. You should also double check if the model files are the same.

P.S. The command will be configurable in the FFMetrics soon.

@fifonik Just tested, its that, got same result by not using Harmonic Mean, excellent.

Your program loads with the last files used, which is very useful.

If I want to cancel I can of course "X" close the program, but a user might simply want to stop instead because they forgot to add another item. So can a "Cancel" option be added?

can a "Cancel" option be added?

Sure, it is on my todo list.

@fifonik Everyone seems to have one of those lately. Your app is limited to 10 at a time, maybe this is to do with graphics and colour options?

The functionality of being able to process a lot more files may be something to consider even if it meant disabling some of the format, i.e. an either or capability.

Yes, the number of processed files are limited by graphs. Unfortunately, the plot library I use is very flexible and easy to use, but not very fast.

@fifonik Anyway, its all very well done, a great addition to wwaggs. No reason any more to be in the dark re: codec quality.

@JN- Thanks

Updated.

Changes in version 0.7.0b:

    New:    Calculation can be cancelled
    New:    UI: VMAF model can be selected
    New:    UI: Only small part of video can be processed (pre-defined set on options: 5, 10, 30, 60 seconds, whole file)
    New:    New command line option "-log-level-debug" added (can be useful for investigating issues)
    New:    Error (if any) is shown in tooltip
    Change: UI: slightly re-designed
    Change: UNC paths should be supported now
    Change: VMAF model files now should be in "vmaf-model" sub-folder
    Change: Files limit increased from 10 to 12
    Change: Target changed back to .NET 4.7.2 (included in Windows 10 1803+, older versions of Windows will require .NET 4.7.2 installation)
    Change: Internal changes and optimizations
    Bugfix: Race conditions while working with external process in background thread should be fixed

@fifonik Excellent. I left my drag and drop batch file at the same download link as the tables, in profile, must do some more testing tomorrow. I think its finished.

@fifonik Been busy, only got to download and test now. Great having 4k addition.

Question: RE: VMAF ... should the user be allowed to use the two different sets of PKL files on say FHD and 4k, interchangebly, or should the 4K PKL drop-down be restricted to only 4K files? Also should the non-4K PKL drop-down be restricted to only non-4K files? The return values ARE different.

I assumed that the two sets of PKL files were to be used only for their respective file resolutions?

The option to limit testing to as low as 5s is great for testing.

A valid point about VMAF models. I also thought about this.

However, having in mind that this is only FFMpeg GUI designed for one specific task, it is up to end user to specify what model to use.

If I decide to change this one day, I will need to do the following:

- Create UI/config for model <=> file properties mapping. There are already quite a few VMAF models available and I know nothing about them. At this stage I do not even know if this only depend on resolution or not.

- Add FFProbe as another dependence as I cannot get video file's properties with FFMpeg only.

As a result, I'm not even considering to add this on my ToDo list at this moment. Sorry.

@fifonik In case I wasn't clear ... "At this stage I do not even know if this only depend on resolution or not."

I assume that the different models are specific to the resolution, i.e. that the 4K model should only be used with 4K reference and test files. This assumption of course may be wrong. If it's right then what I was suggesting (maybe not clearly enough) was to restrict/lock the drop down selection to match the resolution type in use.

Example don't allow the user to select a 4K model with reference/test files that are not 4K resolution and visa versa.

On a more serious topic, I really have to go shopping for a quality TO DO list, everyone appears to have one nowadays. Nick Hope, WWAAG and now you.

If you know of a good quality product do let me know. :D

Anyway, we're swimming in RQ metrics, great stuff indeed.