Messages

By raw rec I meant mlv_lite! I'd still like to try out the h264 with high bitrate though, he quality might increase alot.

Ah, cool.

I remember being able to record continuously on 60fps with crop rec however at this moment It keeps hitting the red whenever I try it, only giving me about 15 seconds. Must be a bug.

Not a bug, the card and bus interface is too slow to record 1080p at 60fps. Maybe with the uhs module and more compression (10-bit) I might be possible to record continuously.

That's a different setup from what i usually have. I record in raw lossless 12bit most of the time on the raw video module (not the Mlv) because of space and things i've read, making it sound like a good mid-term.
48fps giving better results than the 60fps option? I use crop rec for the slow motion.
I also can't find the option to go up to CBR 3.0.
You're kinda disputing alot of what I've been doing Granted I'm still new to this

Well, forget what I said about CBR, I thought you was recording in normal (h.264), not raw.
CBR will only affect normal recording.
For slowmotion, 60fps will be better, but I don't think you can do 1080p at 60fps without frameskipping.
Also, you should be using MLV or mlv_lite instead of raw_rec.

I own a 5d M3 and several CF cards (Sandisk extreme 120mbs and 160mbs) should I use the CBR 3.0 for everything in general?

Yes, always. If you have these cards and MKIII, I highly suggest you use MLV rec instead of normal h.264. You probably know, but some people can record 48 FPS using 10-bit MLV. I can't recommend this enough for you, since the result will be dramatically better.
A simple workflow would be:
- Check if MLV_rec module is loaded
- Set to 1920x1080, 10-bit
- Record using ETTR and/or Dual-ISO
- Input the MLV files into MLVApp
- Check if the exposure and WB is correct. Choose "AlexaLog" in "Tonemapping" and export to Prores 444.
- If the output has noise, use NeatVideo or other denoiser
- Use Twixtor on AE and export to Lagarith or Cineform
- Color Grade as you wish (Logarist and Premiere Lumetri work well for me). For sharpen, I use Samurai (from Digital Anarchy).

Also, I recently downloaded the James Miller Log profiles and I still havent decided if I should go for the Log2 version or Log3 version.  Thanks again for all the help.

If you record using Raw (MLV), the picture style will not affect the footage, only if h.264. I haven't used these from James Miller, but some years ago I tested every single one of them. The VisionTech 2.0 (from Vision-Color.com) have good results and LogNeutral (linked above) is also very good. Today I only use LogNeutral in all my works.

In exposure override I don't have the option to increase the numbers I can only turn it on and off.

Yes, Exposure Override let you choose some shutter values you wouldn't have with default canon firmware. Just turn it on and then change the shutter speed. You can also use the "fine tunning" settings on the movie tab, for fine control over the shutter speed.

Then on fps override do I set it to 23.976 or 60?

Set to the maximum (probably 60 FPS), with the "High FPS" option (press the [Q] button).

Last question, is 1600 ISO too high for twixtor?

It depends on your camera. High-end cameras like MKIII will have no problem, even at 3200. On low-end cameras such as "Rebel" series (600D, 700D, etc) try not to go higher than 800 ISO and never higher than 1600, since on these cameras they apply digital gain, not analog (generates much more noise).
If you really need more exposure and your lens is already wide open, you can apply a denoising algorithm (like NeatVideo) and then use Twixtor.

I don't have any good demonstration, I generally record people talking, so there's nothing too dramatic like water splash's and so on. But, I assure you that the results are pretty good if you have enough light and a fast lens.

If you're using normal recording (not MLV) and have a fast card, use CBR 3.0x for higher bitrate. Also, use some flat picture style, for better dynamic range (see LogNeutral). On the AE export settings, use a lossless codec, such as Lagarith or Cineform, so you can do the color grading without losing even more information.

The shutter speed should be minimum of 2x the final fps you want. So, for example, if you want it to be 120fps, you'll need at least 1/240 shutter speed. I suggest you use the lens wide open (<f/2.0) or else you will have to up the ISO and Twixtor don't do well with noise.
If you're recording in sunlight, try to use Exposure Override in ML and set it to 1/8000.
Also, if you don't need audio, turn on the FPS override and choose "High FPS" preset.
On Twixtor, I use these settings (on AE - always work for me):
- Image Prep : Contrast/Edge Enhance
- Frame Interp : Motion Weighted Blend
- Warping : Foward

Ok thank you! It's S-log3 though, does it matter? I've been exporting with the alexa profile.

Yes, you have to output to S-Log3, not Alexa Log-C. These LUT's you're using are constructed based on S-LOG standard, if you output Log-C and put a LUT based on S-Log3 it'll end up with wrong colors and contrast, and might generate all sorts of issues, such as bandings and chroma artifacts. Even if you're working with the correct profile, these LUT's can generate wrong colors. That's the case of ImpulZ LUT's. I really like to output Log-C on MLVapp and then apply the Vision3 emulation from ImpulZ, but it always generate magenta midtones, for some reason.
If you want color precision, as I have said in many other posts, working with ACEScct on CinemaDNG files is probably the best non-destructive workflow today (although the color matrix on some processors like Resolve don't seem to be precise, as @Andy600 pointed in other posts too)...

Hey guys. I recently started working with James Miller's C-LOG profiles. Don't now if you guys are familiar with it but my question is what export profile should I pick in MLVapp? I got Cineon, Alexa's and the s-log. Anyone's been through this and reached a conclusion of what's best?

Below "Highlight reconstruction" box you'll have different spaces you can work with. MLVapp don't seem to support C-Log, but from what I checked, these LUT's you're using are based on SLOG3. This one is supported on MLVapp, so you just pick this one, adjust exposure/WB and export (preferably using ProRes XQ 4444 or DNxHR 444).

Yes they did ad editing several versions ago! I edit my projects in Resolve, since the XML workflow between Resolve and PremierePro seams to be broken in latest PP version.

Wow, that's nice. I'll look into it. Thanks.

Guys, I'm talking about video editing ("montage"), not color grading and image processing. Unless Resolve added video editing like Premiere Pro, it's not the solution I'm looking into...
Thanks for everyone, anyway.

I've reached my saturation point today with Adobe software. I work with their products for 7 years now and they gave me nothing but headches. Missed a deadline because of bugs without fix. Is anyone willing to point me alternatives (except FinalCut, I don't work on Mac)?
I always find facinating how open source software work better than a multi-million dolar company like adobe. Not only their software are full of bugs, but also their support for the community is null. They leeched ideas from community and, besides DNG format, they gave nothing back.
Except for some crash's on Rawtherapee, open source software have almost perfect reputation from me. I heard Natron and Blender could make a good workflow, but I need to have the animation functions found on Premiere Pro, such as smooth motion curves and motion blur on objects (vector or png's). I'm also tired of Windows instability. Adobe and Microsoft is nothing but trouble for people that need stability. The only thing holding me on windows is Adobe produts (mainly Premiere and Photoshop).

Nice.

Still having trouble compiling, anyone mind helping me? I'm running 16.04

Code Select Expand


$ make
g++ -std=c++11 -c -m64 -pipe -O2 -Wall -W -D_REENTRANT -fPIC -DQT_DEPRECATED_WARNINGS -DSTDOUT_SILENT -DQT_NO_DEBUG -DQT_MULTIMEDIA_LIB -DQT_WIDGETS_LIB -DQT_GUI_LIB -DQT_NETWORK_LIB -DQT_CORE_LIB -I. -isystem /usr/include/x86_64-linux-gnu/qt5 -isystem /usr/include/x86_64-linux-gnu/qt5/QtMultimedia -isystem /usr/include/x86_64-linux-gnu/qt5/QtWidgets -isystem /usr/include/x86_64-linux-gnu/qt5/QtGui -isystem /usr/include/x86_64-linux-gnu/qt5/QtNetwork -isystem /usr/include/x86_64-linux-gnu/qt5/QtCore -I. -I. -I/usr/lib/x86_64-linux-gnu/qt5/mkspecs/linux-g++-64 -o MainWindow.o MainWindow.cpp
MainWindow.cpp: In member function 'uint8_t MainWindow::drawZebras()':
MainWindow.cpp:2913:34: error: 'class QImage' has no member named 'pixelColor'
             QColor pixel = image.pixelColor( x, y );
                                  ^
MainWindow.cpp:2918:23: error: 'class QImage' has no member named 'setPixelColor'
                 image.setPixelColor( x, y, Qt::red );
                       ^
MainWindow.cpp:2925:23: error: 'class QImage' has no member named 'setPixelColor'
                 image.setPixelColor( x, y, Qt::blue );
                       ^
MainWindow.cpp: In constructor 'MainWindow::MainWindow(int&, char**, QWidget*)':
MainWindow.cpp:58:70: warning: ignoring return value of 'int chdir(const char*)', declared with attribute warn_unused_result [-Wunused-result]
     chdir( QCoreApplication::applicationDirPath().toLatin1().data() );
                                                                      ^
Makefile:905: recipe for target 'MainWindow.o' failed
make: *** [MainWindow.o] Error 1


It's probably some QT issue. Assuming MLVapp is using qt-5, try to install all QT base packages (don't know if ubuntu still has apt-get, but just typing "sudo apt-get install qt-5* qmake*" would do the job).

So, I got a Lexar Pro (1066x) 32GB, and it's able to record only 4 seconds on MLV at 50D's maximum res (without crop, I think it's 1556x10xx or something). I don't know why, but I was not able to record more than these 4 seconds, while my Komputerbay 64GB is able to record up to 2min in the same resolution.

I'm posting this here because I could try anything on this card, since I have no much use for it. If @a1ex or @dford want to fry this one in some crazy experiment, feel free to send me a module for 50D and I'll do it.

that's kind of inefficient since you can save yourself a full render of all the takes if you edit and grade in resolve, you can quickly change the grade of a single shot instead and make a single render, instead of two if you use two programs...

Read: "You can also use proxies with XML files."



--luiz r.

If you can provide some more details on the workflow you're thinking of, I'd be glad to try it out!

Convert MLV to CDNG. Import CDNG on Resolve. Make it use ACEScct, like in this video (not mine):
https://www.youtube.com/watch?v=B7DqtW7LB7g

You can work with the Rec.2020 ODT, for a bigger color space output (Youtube and Vimeo now supports Rec.2020 and 10-bit videos).
Then, export in ProRes XQ 444, 10 or 12-bit, Rec.2020 and edit on your software (Premiere, FCP, etc). You can also use proxies with XML files.
This is probably the best workflow for color grading you can get right now, Raw data with ACES color. Resolve has many issues, but this would work on other software that supports ACES and debayering.

i found it here: https://www.magiclantern.fm/forum/index.php?topic=20591.msg190613#msg190613

Meh. Thought it was some good workflow. There's too many transformations in this. Using ACEScct on CinemaDNG's would yield better results, or even just a direct transform from Log-C to rec.709, using something like Logarist...

looks great! can you share the link to @wangtrirat's workflow?

+1

6. You can use ctrl+c - ctrl+a - ctrl+v to apply changes to all clips.

Oh, that's cool. Thanks Kharak.

@50mm1200s

1 - Console windows is hidden for any OS, but unfortunately this does not work for Windows. If someone will came up with the solution we're gonna really appreciate this.

I don't know how mlvapp is calling ffmpeg, but the command "start" seems to work for Windows scripts...

2,9 - Avid codecs are implemented with those restrictions in ffmpeg. (I guess this is Avid spec restrictions)

:/

4 - On loading (parsing) the clip "MLV App" only caches some information derived from the MLV, like all absolutly needed block header metadata and video and audio frame offsets for fast and random frame access. The only case when actual video frames are cached if "Menu/Playback/Debayer for Preview/AMaZE Cached" is set (checked). In this case proggie tries to cache as many Amaze debayered frames as possible according to available memory.

Ok. Well, while I try to select a new clip it has a big latency (~10s) until I'm finally able to change settings for this file. It was a bit annoying, because I had about 40 mlv files and, even though I copied the first clip settings to all the other 39, I had to adjust the exposure manually...

7 - Simple WB color picker is on schedule, needs WhiteBalance branch to be settled.

Nice!

8 - OpenCL is not used right now. Only some experiments were done by me and Ilia.

Yeah, my impression is that the prog behind opencl I kinda complicated.


Thanks bouncyball!

So, I did a entire production with MLV and processed with MLVApp.
Here's some usability stuff I missed and would be really helpful if implemented:

1- Do not pop ffmpeg terminal (hide it). I myself work using workspaces with tiling windows. Everytime I export clips on MLVApp the ffmpeg terminal pops while I'm doing other stuff.
2- Present export options after choosing the folder for the selected clips or do a qeue panel. For example, DNxHR will not export without having the correct FPS, so I would find it better to have the export settings pop after the selected folder...
3- Fast kill (abort) export. Always I want to abort the rendering I have to wait like 1-2 minutes. Could be faster this, don't? Idk about windows, but Mac and Linux have the "kill -9" command that kill right away the process.
4- I think MLVapp is caching all MLV when loaded. Maybe it would be better to cache just the first frame, for fast loading, since (in my case) I did all the corrections using the first frame only.
5- The paused preview could be in AMaZe automatically, instead of bilinear. Save bilinear only for playback (?)
6- Copy parameters of the selected MLV to all files in session. I think it's a small change, but would save time from selecting all images manually.
7- WB Pipe for matching gray cards (I had to export a dng and check on rawtherapee to do this). I think @bouncyball already proposed this, but I'm confirming this would be very helpful (maybe when Ilia finish the new WB branch?)
8- OpenCL is in use? Could be a checkbox on the "About" panel, just to confirm it's working or not...
9- There's no resize options for DNxHR? I missed it

Very nice work. Alexa has these amazing colors, as always, the sensibility on high frequencies (violet/blue) is just fantastic, but 5D3 is not too far from it (at least on my display).

edit: Indeed, the crop mode is too much for professional works. I couldn't use it, at least...

I don´t copy anything to /bin... , so no security considerations

Oh, that's right. Just saw you're using home with a directory named /bin. Sorry.

BTW -- Any help on this would be greatly appreciated.

50D tests are needed? I'll test this.

I for my part do not want to install mlvapp system-wide. I want it locally in my home-folder ( /home/myusername/ ). So my way for building mlvapp is:

Code Select Expand

- download the static version from ffmpeg for linux from the ffmpeg-website
- copy ffmpeg to ~/bin
- cloning mlvapp from git
- making the buildfile with qmake-qt5
- compiling with make


Couldn't it use the package manager, instead of copying manually? It could have a script that gets the system name and then assign the correct expression to the package manager. If I remember correctly it could use:

Code Select Expand


$ uname -a | grep -E 'Debian|Arch|OpenSUSE|Ubuntu'


Depending on the output, it could automatically call the package manager. For example, Arch would use "pacman", debian will use "apt-get install". This would make it more reliable, since if you 'hardwrite' the link in the core, and the link on ffmpeg site changes, the code will not work. Also, there's some security considerations on copying something to /bin...

We could maybe use bash and then run with the new compatibility in Windows 10. It's dirty, I know, but the same script would run on linux and windows, so maintaining the script would be easier...
I don't have much experience with bash, besides managing servers, but I perhaps could help in some way.

Welcome Adam!