Messages

This is quite amazing Danne. What method is it using? It's like Tlsophote, where he has a edge-detection algorithm and them blurs the areas or it's something more complex like superresolution?

Well, from what I know, g++ is a compiler, qmake is a makefile. The makefile links all the libraries and other code of the project and then sends to the compiler itself. The compiler will then... compile (transform to binary code). Inside the makefile you should add the compiler options, such as optimizations (-Ox in GCC, where "x" is a number from 1 to 3 or "s"), also archtechture specific parametters (x86, ARM, etc), and other stuff I really don't know.
Are you using some linux distro or Mac? Cygwin is very boring to deal with. If you're on Windows I'd suggest you see Chocolatey, a package manager that can install automatically many tools. On Mac, see Homebrew...

I think you have to use -std=c++11 on g++ options... at least that's what the error log is saying.

Other temps are quite off.

Any clue on why? What is necessary to bring it to mainline?

The WB branch seems stuck, right? Ilia even commited the Andy600 suggestion, but backed off...

I'd just been struggling using Switch and trying to correcting Log 4444 files in Premiere. I like Colorcast, but the rendering is slow.

This could be a solution: https://www.logarist.com/

Cool. Have you tried HDR with infrared? Would be interesting, I think. I'm doing experiments with HDRMerge and the new tone mapping from Rawtherapee 5.4 ("HDR Tonemapping", fattal method), and the results at ISO 100 is incredible, there's virtually no noise, even when pushing tones.

I don't work too much with Resolve, but you should check where the workflow is failing by testing.
- Does it looks OK inside Resolve? If not, then you have to configure the color management panel
- When you export to ProRes, does is look OK when playing with Quicktime? If not, then adjust the color/gamma output
- If all the above are OK, but when you open inside Premiere it is *not* OK, then you have to configure premiere color/gamma settings.

If even configuring everything nothing works, try to export in Rec.709 just to check if the issue is not the wide gamut. If Rec.709 is not OK too, check the display settings, maybe Resolve is automatically working with the systems Color Management settings, idk.

Thanks!

Does anyone know if Resolve and Premiere Pro has a tool like the Selective Color in Photoshop, where you can change specific color tones? I miss this feature when doing color grading in video...

You have to change the preview mode in Premiere Pro. In order to decrease the computational power required to playback in realtime, Premiere Pro lowers quality of the image. You can click in a button in the right corner. It's probably "1/4" you have to change to "full" or something like that...

Andy600 could answer better, but:

What I hope to have understood is that HDR refers to light, while gamut refers to color: two different things, but I'm interested into how they are related.

HDR refers to dynamic range, or, how many shades of gray you can obtain between black and white. Gamut, from what I know, is how many colors you can digitally represent.

Now, my camera (5D3) is set on AdobeRGB, and I shoot videos in RAW.

The colors space configured inside the camera will have no effect on Raw images (including MLV, of course). Raw don't have a color space assigned to it. It's up to the software you're using to process it.

I should have understood that HDR is a tripled-frame capturing to obtain high dynamic frames to both raise shadows and prevent highlights clipping at the same time (right?),

I don't think there's a feature like that in MLV recording... there is? I think you're confused with the h264 "HDR" recoding.

while DualISO should do the same thing but this time through a dual-frame capturing and reducing vertical definition (right?)

DualISO uses different sensibilities on each line on the sensor. So, the result is a image with two different exposures at the same time, in the same frame. Then, it gets interpolated to create a image with higher dynamic range.

while Auto ETTR should maintain the "distance" between shadows and highlights, raising the shadows and clipping a bit the highlights that are still recoverable thanks to Davinci "recover highlights" button (right?)

ETTR is "expose to the right". So, what this does (and you don't need magic lantern to do it), is get more light to be captured in the camera (generally 2 f-stops more than normal). With more light, more information. On post-processing, you can reduce the exposure and get a normal image, but with more information/less-noise in shadows.

Now, these three methods are different, but it seems to me they are oriented to a similar purpose: obtaining the widest dynamic range.

Yes.

What I hope to have understood is that both HDR RAW and non-HDR RAW videos are good for wide gamut workflow. (right?)

What do you mean by "non-HDR"? 10-bit and 12-bit raw?
But, yes, Raw information is not limited to any color space.

Now, can we consider DCI-P3 a wide gamut color space? I'm interested into this because I'm going to try coloring a video work for a theatrical premiere.

I don't have the answer, but the new "best practice" seems to be ACES.

Sounds like you havn´t been working with this tool tbh @50mm1200s.

I use it professionally every week for some time now. It works well.

More control will always be welcome imo. It also feels like a natural step to be able to do this before batch conversion.
My 50 cents...

I didn't say more controls are not welcome. I said the devs efforts should be put in more important things. That's why projects like rawtherapee have a roadmap... maybe that's a good idea for mlvapp? I don't think there's enough people working on this, but, still.

I'm ashamed to post some of my works now on this forum. You guys only post awesome high level shots, damn.

I don't want to sound like an idiot or something, but: I think there's a better way to put the efforts than colorwheels and gradient tools, don't you guys think? I can be wrong but I think most people don't do color grading on a software like MLVApp. It think the target public is using it mostly for simple conversion or log conversions, not for primary color grading...
For example, a color space management would be *very* useful. Lens correction through LCP profiles too (would fix the issue some other guy posted on the forum, with magenta fringe in highlights). A tone mapping operator like the "fattal method", to recover informations in highlights/shadows. So on...
Sorry if I seem rude, I can't write "gently" in english, it's not my mother language.

[So]

Damn. So good. I'll probably never do such a fine job like this.
I really like the grain (is it filmconvert?).

I tried to click your link but it doesn't take me anywhere.
I am going to check out ARRI's channel.

F*ck, didn't know the link was broken. Here:
https://support.google.com/youtube/answer/7126552

Why would one upload with a modified metadata and matroska container? First time I heard about it.

As described in the article linked, youtube needs a cue on if the file is HDR or not, so you have to write this on metadata. They have a tool on github and also a tutorial for premiere pro, you can just read on the article. The ARRI channel on youtube have now many HDR videos, if you want samples...
I you the file also needs to be in Rec.2020, so that's a problem if you're (like me) using MLVApp or other software that only supports Rec.709.

With fullres blending "off" and AMaZE preview the MLVApp gives me this:



You probably already know, but the stripes seems to be miss-aligned. There's ~6 horizontal stripes that are not in the correct colors (alternate between yellow and blue, probably one channel of YUV chrominance is off).

You mean the 10bit/12bit build?

No, I mean the build you made on this thread.

Delete the ML settings and try from a fresh start. On both the 7D and 500D I found that it works the first time but you get "earthquake" video on restart.

mlv_rec works fine (with crop_rec), but mlv_lite gives me the glitches. Indeed, the glitch is mostly when I start the recording and in the middle of the footage.

Hey, this is highly experimental stuff. Help figure it out.

Sure, as I said, if you need me to do any tests I can help. Everything seems to work normally (WB and black levels are ok), except for the glitchs:

Wow, very dynamic movements. I like the film-like color grading and the sound design is very good too.
Just a question: how are you uploading it to youtube? I've downloaded it using youtube-dl and the DASH manifest is odd. As a recommendation: youtube has now HDR support (10-12bit) and they suggest you upload in ProRes with modified metadata and matroska container (.mkv). Will take more time to upload, but I think this would fix some banding in your video...

I'm getting glitches
Setup is 50D crop_rec (3x), ML build from @dford, res 1920x1080 at 24fps. If you need any other info I can provide it...

Thought it would be a good idea to share this:

Welcome!

with highlights slider all down, a magenta little lined disc appears internally the sun.
What does it mean? A simple fringe artifact, or a damage of a portion of the sensor?

Probably the highlight reconstruction algorithm in ACR, don't worry. You can also try on rawtherapee, without reconstruction, so you can make sure about that.
tl;dr: if other photos are ok, don't worry.