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

[masc]

in their site pytorch.org I saw that you don't need cuda if you use libtorch for c++

They don't say, CUDA is not needed. They say, CUDA is not supported. (on mac)
But on the VRT project page you can read the requirments needed: Python 3.8, PyTorch >= 1.9.1, CUDA 11.1 
So I don't expect it to work here on my computers.

[ML700D]

pass.. 😂

[iaburn]

About the pink horizontal lines on dual iso, I've been experimenting and it seems like the white level is set too high on the high ISO part of the image. If I set it to half of the value (white_bright *= 32 instead of 64 on dualiso.c - diso_get_full20bit), in most cases it fixes it completely or at least it gives a constant pink highlighted area, like on the example below.
I'll experiment a bit more, but maybe someone who actually understands the math behind all this can think of a reliable solution that works on all cases and it doesn't affect the general quality. Maybe a slider to adjust white balance individually for the bright and dark parts of the image when dual ISO is selected would help?

[ML700D]

my dual iso test..
I use iso 50 tweak -1 EV (canon iso 100)

no adjustment


after adjustment (w/o noise reduction), max detail I can get from my lens capability.

[ML700D]

About the pink horizontal lines on dual iso, I've been experimenting and it seems like the white level is set too high on the high ISO part of the image. If I set it to half of the value (white_bright *= 32 instead of 64 on dualiso.c - diso_get_full20bit), in most cases it fixes it completely or at least it gives a constant pink highlighted area, like on the example below.
I'll experiment a bit more, but maybe someone who actually understands the math behind all this can think of a reliable solution that works on all cases and it doesn't affect the general quality. Maybe a slider to adjust white balance individually for the bright and dark parts of the image when dual ISO is selected would help?

for pink highlight I check the highlight reconstruction to make it back to white

[iaburn]

for pink highlight I check the highlight reconstruction to make it back to white

Hi, what I described is a particular situation that you might have not experienced. Highlight reconstruction has no effect in this case.

[ML700D]

oh.. I see
maybe masc can figure it out.. 💡

[iaburn]

After more experimentation, it looks like modifying the safety_margins on the white_detect function has the same effect. Top margin is set to 1500, but a value of around 4000 fixes the stripes (at the expense of a bit less DR, like it says on the comments)

Code: [Select]

int safety_margins[2] = {100, 1500}; /* use a higher safety margin for the higher ISO */
It should be possible to adjust this value automatically I suppose, but it could also be customizable from the Dual ISO section

[Danne]

As you dive into dualiso swamp you'll probably notice more edgecases causing issues, especially with mlvfs based fork used in mlv app. A few refinements were achieved by bouncyball but the more robust cr2hdr code from a1ex original code should be revisited against edgecases causing trouble otherwise.

Searching this very thread reveals some of the issues. Lucklily I think I keep later crhdr code in this branch. Now it "only" needs porting 
https://bitbucket.org/Dannephoto/magic-lantern_jip_hop_git/src/master/

[iaburn]

Lucklily I think I keep later crhdr code in this branch. Now it "only" needs porting 
https://bitbucket.org/Dannephoto/magic-lantern_jip_hop_git/src/master/

Thanks for the code! It looks very similar, I'll give it a try next weekend

[Danne]

That sounds fantastic.
As I have no access ro computer atm I am almost certain this code is the updated one. If possible you could check against this branch too, it's either one with latest cr2hdr code.
https://bitbucket.org/Dannephoto/magic-lantern_dannephoto_git/downloads/

If you find the problematic test files somewhere in this thread it should be good to check against that. Sorry I could not be of more assistance here but I am positive others could find this.

[iaburn]

That sounds fantastic.
As I have no access ro computer atm I am almost certain this code is the updated one. If possible you could check against this branch too, it's either one with latest cr2hdr code.
https://bitbucket.org/Dannephoto/magic-lantern_dannephoto_git/downloads/

If you find the problematic test files somewhere in this thread it should be good to check against that. Sorry I could not be of more assistance here but I am positive others could find this.

That code already helps a lot, now I just need time to understand XD

I found this post with a problematic video:

Hello.
Canon 650d. Shooting in the morning with dual iso 100/400 fusion with the sun. After incorporating dual iso in mlvapp -  appears a strange spot in the sun:

Photo with dual iso - in order
Sources:
https://mega.nz/folder/HlxkhZaa#xB_W3KI5mMn9mwOHuhiB8g
Is there any way to fix this? Such a problem on many source videos
(Highlight Reconstruction and White Level Change - do not help)

And it looks good after adjusting the tolerance:
https://mega.nz/file/8glgmZ6Q#2C1oQBeyokxAx-RVwt3oA5OEGMw5afdoyt6smdy8snY

[Danne]

Yes, those white level issues are easy to fix. There's a lot more to be found when running other scenarios.
Let's see what I can find.

[Danne]

If you want you can start reading from post number #4235 and if possible download the test file. Long story short. Here's one file which will benefit from having code updated.

[iaburn]

If you want you can start reading from post number #4235 and if possible download the test file. Long story short. Here's one file which will benefit from having code updated.

Now I see the problem... Fun fact, I never understood what the Preview button under Dual ISO was supposed to do, as the video looked all broken with pink/green tint, until now!
On that problematic video the preview looks quite good and it's super fast

Very interesting topic and quite complex for someone that just started reading about how sensors work, but it would be awesome as a final degree project for a student! I'll give it a try though

[ML700D]

btw the problem is in the camera or mlvapp actually?

nice finding.. so dual iso will be better next.. 😃

[masc]

btw the problem is in the camera or mlvapp actually?

You forgot the user. If you setup low-iso/high-iso wrong, preview also doesn't work. But the problem can be one of the 3, or a mixture of all. DualIso is a difficult topic.

For the sun problem: it is not just a DualIso problem - google for "black sun": you'll find this problem with many RAW cameras, even ARRI and Blackmagic have (or had) this. If you like to implement and if have a nice idea: it would be a great new "RAW Correction".

[ML700D]

You forgot the user. If you setup low-iso/high-iso wrong, preview also doesn't work. But the problem can be one of the 3, or a mixture of all. DualIso is a difficult topic.

For the sun problem: it is not just a DualIso problem - google for "black sun": you'll find this problem with many RAW cameras, even ARRI and Blackmagic have (or had) this. If you like to implement and if have a nice idea: it would be a great new "RAW Correction".

ah.. yes I forgot 

it means that MLV RAW is not alone in the world 😁

[masc]

I fixed audio export on AVFoundation export with Qt6.4: it seems there are more bugs in Qt lib, because ffmpeg executable could not be found in app image (in Qt5 it works and file definitively was/is existing). I removed Qt calls and replaced with standard C call. Was this working (without my change) with Qt6.5beta?

[Danne]

What are you downloading prior to compiling? Qt6.5beta from where? Llvm from where? I am trying a lazy homebrew install and I almost get it compiling but rpath issues and two macdeploy files are missing. Not sure homebrew is working 100%.

[masc]

I used Qt5.15 and Qt6.4 from official Qt-Online-Installer and llvm@15 from brew, on BigSur Intel.
On M1 I am still on Qt5.15 selfcompiled and llvm@15 from brew.

On the Intel machine I see the differences between Qt5 and Qt6: e.g. file access. But maybe it is a problem on this computer, or with Qt6.4, or...

[Danne]

It seems to be issues with homebrew qt so will download and test macports. If not working I'll do it your way.
My qt5.15 is broken since your lates commits so I went ahead and downloaded ventura to get it working cleanly with latest os and so on. Will test some more this evening . Very nice to see this progress.
Also nice to see somebody taking some serious steps to update dualiso code  @iaburn👍.

[Danne]

I used Qt5.15 and Qt6.4 from official Qt-Online-Installer and llvm@15 from brew, on BigSur Intel.
On M1 I am still on Qt5.15 selfcompiled and llvm@15 from brew.

On the Intel machine I see the differences between Qt5 and Qt6: e.g. file access. But maybe it is a problem on this computer, or with Qt6.4, or...

Could you link to what you are using to install Qt6? A lot of alternatives and I bet all of them takes at least an hour to test..
https://www.qt.io/download

EDIT: Might have found something here:
https://www.qt.io/offline-installers

Also found this:
https://wiki.qt.io/Building_Qt_6_from_Git

[masc]

Could you link to what you are using to install Qt6? A lot of alternatives and I bet all of them takes at least an hour to test..
https://www.qt.io/download

Exactly this link. Then scroll to "Go opensource", Then "Download the Qt Online Installler". Then you'll need a free Qt account. Then install:
- Qt 6.4.2 macOS, with QtMultimedia
- QtCreator

I think, if you find an Offline Installer from official Qt page, it will do the same. (Years ago I used this - no idea if it changed.)

[Danne]

Yes. Got a version going. Thanks for testing. Seemingly working but should also be tested on a vanilla M1. Anyway. Good jump into the future here.