MLV App 1.14 - All in one MLV Video Post Processing App [Windows, Mac and Linux]

[bouncyball]

Actually used imagej for getting coordinates into a pixel list:
... Code from raw2dng if I don´t fail to remember).

Yeah! That's right

[masc]

Don´t be sorry. Interesting bug. Any idea what it might be? How is the vidstab setting applied? I only ran the script with mov files in Switch but always working:

Code: [Select]

ffmpeg -i input.mov -vf vidstabdetect=shakiness=10:accuracy=10:result=transform_vectors.trf -f null -
ffmpeg -i input.mov -c:a copy -c:v prores_ks -profile:v 2 -pix_fmt yuv422p10 -color_primaries bt709 -color_trc bt709 -colorspace bt709 -vf vidstabtransform=input=transform_vectors.trf:smoothing=10,unsharp=5:5:0.8:3:3:0.4 output.mov
Could you point me to code? Maybe the the trf file is erased too early or something? Only guessing...

@Danne: have you tested this with a ProRes mov or a RGB-avi file? Did that work? I think, even if MLVApp does export a ProRes with vid.stab enabled, there is no staibilizer applied, while exporting h264 it works fine. I remember I had similar issus in analysis pass: it only works with h264, so I implemented the pipe to pipe thing. Maybe here we have the same problem, but pipe to pipe means quality loss (not good).

@Bouncyball: Picker is no problem... functionality is there, we just need another icon -> easy exercise. But we need some buttons for adding/deleting pixels, for loading/saving a pixelmap. We should think about another widget. 

[Danne]

Will check a file for vidstab later.

[Danne]

Check wav(audio) routines. When enabling Export Audio files comes out. Without it files are blank(checked prore422 and prores444).

[masc]

Check wav(audio) routines. When enabling Export Audio files comes out. Without it files are blank(checked prore422 and prores444).

Thanks Danne. But exported now without audio... empty file, ffmpeg crashed and pulled MLVApp down. Somehow vid.stab doesn't like RAW to ProRes...

[Danne]

Yes, but enable audio and then export. Working? At least your 5D2 file was working.
https://www.magiclantern.fm/forum/index.php?topic=20025.msg216212#msg216212

EDIT: Tested some other file. WIth audio enabled it seems to work but don´t think vidstab is applied.
Also seems to crash on the second file. Well. Problematic.

[bouncyball]

Picker is no problem... functionality is there, we just need another icon -> easy exercise. But we need some buttons for adding/deleting pixels, for loading/saving a pixelmap. We should think about another widget. 

Cool! I'm gonna open the new issue on github then

[masc]

For all guys who know how to compile MLVApp:
I had success in adding the chromatic abberation correction from RawTherapee into MLVApp. This means, an older version of it. The older version has a code interface very close to what we need, while the current version is so compilacted, that I had no success in just compiling. What you get are two new sliders to correct red and blue CA's. Moving the red one to around 0.7-1.0 fixed most CA's in my test clips. Especially for all EOS M guys with Viltrox Speedbooster, this is a great addition, because the booster produces so many CA's. Here an example - not perfect, but better than without.




Would be nice if someone else could test this...

Same for our vid.stab problem. Would be nice if someone could test and report what works and what doesn't.

[70MM13]

while you are on the subject of rawtherapee, may i make a suggestion?

the dynamic range compression function is really superb, and it's in the older version (at least, older as in from a year ago? is that old enough?) but there's only one issue:  it doesn't maintain consistency in changing lighting conditions.  I don't know if that is something you can fix, but I guess the first question is will it compile?

just a thought... it is a really great tool for stills and would be incredible if you can make it work for video!

[masc]

RawTherapee... you never can say for sure if any of those modules are compaible with our software. In general it isn't at all out of the box. The first time I found some CA correction code from RT was in february, now I got the code adapted to be able to compile and use it. I was very happy got this to work, without knowing how it works and what it does. There is no documentation (or I haven't found it).
In the current official MLVApp version DR compression is done behind profile combobox. I don't expect it even to be a little similar.

[Luther]

I had success in adding the chromatic abberation correction from RawTherapee into MLVApp.

Nice! Do you think it would be possible load profiles automatically using lensfun (based on metadata)? Also, this lensfun software seems really cool, might be of some help in MLVApp.

In the current official MLVApp version DR compression is done behind profile combobox. I don't expect it even to be a little similar.

The "DR Compression" in RT is actually a tonemapping operator. Specifically the Fattal method:
http://www.cs.huji.ac.il/~danix/hdr/hdrc.pdf

It is a bit different than what MLVApp does in profiles. The way I see it is: where the profile is used to "comply" with a standard (e.g, Log-C) in a linear way, the tonemapping tries to compress or lift the illumination dynamically.

I'll see if I can test the vid.stab and the CA next week. Also would like to test the ACES branch

[andy kh]

highlight recovery/reconstruction dont work in 10bits lossless and 12bit lossless atless in 70D. it works only in 14bit lossless

10bit lossless




12bit lossless




14bit lossless

[Danne]

Please upload mlv samples.

[andy kh]

10bit lossless https://ufile.io/io5nnqba

12bit lossless  https://ufile.io/nsiobnhu

14bit lossless https://ufile.io/i0c31svn

[Danne]

THanks. Heavily overexposed highlights, wouldn´t expect any info left in those areas. Do you mean the problem only occurs on the 70D?

[andy kh]

THanks. Heavily overexposed highlights, wouldn´t expect any info left in those areas. Do you mean the problem only occurs on the 70D?

this was made just for testing purpose and show you. i dont get the pink highlight  after pulling all the way down the exposure if its shot in 14bit. problem occurs only in 10bit and 12 bit lossless getting pink highlight

[andy kh]

TDo you mean the problem only occurs on the 70D?

its occurs on 70D but not sure about other camera. i always get pink highlight when i pull down the exposer or highlight if the footage is in 10bit or 12bit lossless

[Danne]

Try following
1 - Select Highlight Reconstruction then pull highlights slider all the way down. All is pink.
2 - Go to White Level slider under RAW Correction tab and pull it down until you find white level sweet spot(I set it to 3493). Once there you can pull down exposure etc how much you want.

10bit:

[andy kh]

yeah it works. i never use to touch the raw white level as it works without touching it in 14bit lossless
thanks danne

[Danne]

It's a bit odd. I have the same problems. Maybe white level should be set according to this clipping point instead in cam. Maybe we loose details in other places if so, not sure what to do

[masc]

Do you think it would be possible load profiles automatically using lensfun (based on metadata)? Also, this lensfun software seems really cool, might be of some help in MLVApp.

Yes, lensfun is really nice - but very huge. And I did not found out, how to use it. But what I found out: have a look here, you can see that it won't correct CA's for most lenses. And I am not sure if it would when using a Speedbooster on a EOS M (which produces a lot of CA's in another way than the lens would without).

The "DR Compression" in RT is actually a tonemapping operator. Specifically the Fattal method:
http://www.cs.huji.ac.il/~danix/hdr/hdrc.pdf
It is a bit different than what MLVApp does in profiles. The way I see it is: where the profile is used to "comply" with a standard (e.g, Log-C) in a linear way, the tonemapping tries to compress or lift the illumination dynamically.
I'll see if I can test the vid.stab and the CA next week. Also would like to test the ACES branch

Fattal is in theory not far away from Reinhard Tonemapping. And this you can choose in Profiles comboxbox. In ACES branch we separated gamuts and tonemapping. But in current master branch it is still mixed.

@andy kh: this is a known issue with all ML cameras. It happens not always, but if it happens you just should lower RAW white level. We also wrote this in the manual https://github.com/ilia3101/MLV-App/wiki#highlight-reconstruction

Maybe white level should be set according to this clipping point instead in cam.

I already tried this, but without success. No idea what happens with less than 14bits in camera. The identified value was mostly way to high and the pink was still there. Looked as if the clipping point was different for each pixel and each frame

[andy kh]

in mlvproducer it seems to work the same for all the bits
anyway thanks guys atleast we can adjust the raw white balance and recover

[masc]

That's right, MLVProducer does a perfect job in removing pink highlights. Seems to be a different method. Unfortunately I haven't understood how it is done - nearly no comments in code

[ilia3101]

It is a bit different than what MLVApp does in profiles. The way I see it is: where the profile is used to "comply" with a standard (e.g, Log-C) in a linear way, the tonemapping tries to compress or lift the illumination dynamically.

""Complying" with a standard" isn't the intention of profiles. The original intent of the profiles was to give different looks. Back in the day, MLV APp only had standard and tonemapped, to give different looks. These profiles were never supposed to control anything like output colour spaces/logs or stuff like that. It was always planned to be separate but that's not how it turned out. Anyway, the BetterProcessing branch has separate controls for these things.

Also fattal tonemapping looks really creepy.


@masc Really incredible work with the chromatic aberrations!

[masc]

Thanks Ilia.

For getting something very close to Fattal Tonemapping, do the following:
- Highlights: -100
- Shadows: 100
- Contrast: -100
- Clarity: 100