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

[2blackbar]

Same helios 58 -44-2 on both cameras with that aperture ring that can be very precise.
Look at the colors of the paint below yellow one, compare M and 5D2.
I tried to be as similar as i possibly can but i noticed that green bias just recently, i also tested older ml builds and its the same.
Maybe there is more green photodiodes in 5D2 sensor thats causing this.Or there just isnt any standard that canon follows in their cameras so colors wont be accurate and they dont have to be ? Wheres color police ? 
I will take regular RAW photo on 5D and compare that instead of MLV.
--
Just did that, its the same, same colors as in MLV file.
Anyway you can "solve" that in hsl panel in MLVApp, if you want to match 5D2 colors with real life.I just wanted to point out that its happening.
I wish someone would do that comparison with all canons side by side.

[Skinny]

@2blackbar I've noticed this behavior in regular cr2 photos, when switched from 550d to 5d2. But for my taste, 5d2 photos looks more natural, and 550d have some magenta cast to them..

[ilia3101]

Thank you for the raw files, I will look. I remember in rawtherapee 5D2 raw photos always looked like absolute garbage, but other cameras were ok, so there is something different about this camera (or Adobe's matrix that everyone uses).

BTW, with old lenses there may be colour and contrast changes. I definitely know that the Helios 44 contrast increases massively between f2.0 and f2.8. The best way to match the shots in camera would be with shutter and iso.

And I really appreciate the test, if you could do it in other conditions like daylight that would be amazing (I'd do it myself, but my 5D2 is broken for the millionth time)

[ilia3101]

Its worth noting that in order to get correct exposure on both i had to open up the lens on 5D2, which leads me to believe that ISO in canons doesnt follow any standards and isnt equal along their cameras or it just has less dynamic range.

Dxomark says ISO 100 on the mark II is actually equal to 73, and 98 on the EOSM.

[2blackbar]

it is daylight , sun  thru the window.
ithink coloors should be like real life and canon m is closer ,5d2 has color shift in raw, white balance wont help, needs hue shift

[ilia3101]

What are those paints? That difference is shocking and amazing... never seen that before. I want to see how other reds look.

I have heard before that the mark II has a strong infrared filter, so it is less sensitive to longer red wavelengths, and I guess those paints reflect mostly in the long reds. Could you see how the 5D reproduces some other red colours? (in comparison to eosm)

[ilia3101]

What are the used lenses? Comparing in LR makes it look nearly identically. M seems to have just a little more contrast because of less DR.

Can you show how it looks in LR?

I knew adobe must be doing something in addition to the matrices.

Also how does the WB picker in mlv app work? How many pixels does it sample?

[2blackbar]

These pics of actual acrylic paint box next to monitor with image from 5D2 show how different colors are, i dont mean contrast and brightness but hue of reds below yellow paint.
And that pic was taken with pixel 2 XL, it actually shows these reds quite accurate.I mean its how i see these colors in real life, they are definitely red , not orange like on 5D2.


Or maybe my camera is faulty? It would be helpful if someone who has 5D2 would test this as well on some paints .The more colours the better.
Strange thing is i dont see that color shift mentioned in any of the reviews but most of the reviews are just ads.
One more with 3 of them:

[masc]

Can you show how it looks in LR?

I knew adobe must be doing something in addition to the matrices.

Also how does the WB picker in mlv app work? How many pixels does it sample?

Here the LR5 results: (just WB corrected, all other sliders and options are default, no raw black level correction)

5D2:


M:


The WB picker in MLVApp has a range of 20x20 pixels, which are averaged to find the correct WB.

[DeafEyeJedi]

Fantastic comparisons @masc!

[2blackbar]

great, how i can get these colors on 5d2 in mlvapp?

[Danne]

The closest I can get is this in mlv app. Test this recipe https://bitbucket.org/Dannephoto/magic-lantern/downloads/5D2.marxml.

White balanced in acr:


White balanced in Mlv App:


Basically I do following:







I guess if we want to get closer we need to match the "adobe color" to the rec709. Maybe modifying rec709 itself could get us closer to adobe color? I also tested colors in MlRawViewer and they pretty much tells the same difference like with Mlv App.

[Milk and Coffee]

It is enabled by default, because most cameras need these corrections. Appart from that, this needs nearly no time in processing. As the name says (RAW correction) the RAW data is corrected and the corrected RAW is saved into cDNG. After having exported the cDNG, there is no easy way to correct it afterwards.

You can open all your files in one session, uncheck it for one clip, CTRL+C, CTRL+A, CTRL+V.
OR
You can open one file, uncheck RAW corrections, save the receipt. Load this receipt as default receipt. https://github.com/ilia3101/MLV-App/wiki#22-edit

Thanks all for the replies! Very very helpful!

So most cameras need this correction? Are they present, and I’m just not noticing them?

If the image looks good, does leaving “vertical  stripes” on, degrade an already good image?

[reddeercity]

why is everyone so hang up this 5d2 color thing ?
You shouldn't have to adjust black & white level at all . Just select the "Don't use Camera Matrix" & select "Highlight Reconstruction" .
On the 5d2 & 50D "camera matrix" does not do a good job , messes with colors .
I can match frames in 10,12&14bit  from A.E. CS6 to MLV App or match frames from MLV App to A.E. CS6 with out any problems as long as I select the "Don't use Camera Matrix" on 5D2.

One last thing - Is you monitor calibrated correctly ?
No one else seems to have a problem with 5d2 (including me) , you need to look deeper , did you re-set you camera , is there any w/b hue adjust in the camera custom setting  ?
I seen this happen to other people before .
 

[Danne]

You need the camera matrices if you want coherency with sensor and color calibration according to adobe. Without them it is all a "generic matrix roulette". I do admit Ilia3101 did a good job on this no matrix starting point but still.

What makes things difficult is when rec709 is supposed to look like adobe color when comparing. Instead compare with for instance resolve rec709:

In resolve, Apply Pre Tone Cuve set in Project settings


Tangent to rec709 in Mlv App


The provided clips from the 5D II do look greenish and colors differ from eos m as a starting point. When correcting white balance it do looks better. Not perfect but close when comparing in resolve. I didn´t touch white balance either.
 

EDIT:
Correcting white balance(Resolve)


Correcting white balance in Mlv App.



By the way. Here is no matrix selected:

[Luther]

Maybe modifying rec709 itself could get us closer to adobe color?

Using AP1 instead of Rec.709 seems to improve colors in my tests. AdobeRGB creates artifacts in blue hues.

On the 5d2 & 50D "camera matrix" does not do a good job , messes with colors .

I disagree with that. At least for 50D, the skin tones improve a lot with camera matrix.

[Danne]

Using AP1 instead of Rec.709 seems to improve colors in my tests. AdobeRGB creates artifacts in blue hues.

As a stand alone workflow AP1 probably retains more color and keeps them from breaking. But how to use this streamlined matching resolve? I would look at AP1 as an approach of keeping more color to work with in post but precision/matching is not there the same as rec709? At least when I try and reproduce AP1 in resolve comparing with Mlv App. Please prove me wrong on this.

Edit:
Personally I think no matrix selected or Alexa wide gamut would be interesting to explore and use here
No matrix


Alexa wide gamut


AP1, looks kind of wonky, needs color correction luts



And back to rec709 for comparing. Vivid colors for a happy world


Looks like Ilia3101 no matrix color science has a little deeper color compared to even alexa wide gamut. Not bad at all.

[reddeercity]

I disagree with that. At least for 50D, the skin tones improve a lot with camera matrix.

No can't agree , instead of using rec709 or AP1 use Adobe standard space and correct your output color space to 16-235 level instead of full range (0-255)
FYI: I like my image to be flat (log like) as I don't final grade with mlv app.

[masc]

No one else seems to have a problem with 5d2 (including me) , you need to look deeper , did you re-set you camera , is there any w/b hue adjust in the camera custom setting  ?
I seen this happen to other people before .

With my 5D2 I also never had a problem with that. I now tried to set w/b hue to a wrong setting (Canon menu), but the MLV clips look the same. Did you mean something else? Does this really affect RAW data saved to the card? Never heard about that before.

[Danne]

Edit: answer to previous posts. Masc posted while writing.

Please back up statements with images to compare. Simply too many ways to utilize mlv app.
Also important to now what we are looking for here. On one hand seeking color accuracy between apps. On the other hand sort of personal creative uses where more or less anything that looks kind of good or log might work on individual basis.

Edit: hue cannot be changed following into a raw mlv. Come on. Only metadata and apps interpreting differently.

[Luther]

But how to use this streamlined matching resolve?

Yeah, not sure. An alternative solution would be to use Log-C with Wide Gamut RGB. Then apply ARRI LUT on Resolve.

No can't agree , instead of using rec709 or AP1 use Adobe standard space and correct your output color space to 16-235 level instead of full range (0-255)

Why would you limit your range? Bt601 is not used anymore, there's no possible scenario (that I can think off) where your workflow would benefit from it.

FYI: I like my image to be flat (log like) as I don't final grade with mlv app.

If you don't grade in MLVApp and you don't need to do color match, then your goal should be to retain as much information as possible. Using Log-C with AP1 and exporting lossless would be the way to go. Or just convert to CDNG and work directly on Resolve, that would give much better/faster results.

[Danne]

Yeah, not sure. An alternative solution would be to use Log-C with Wide Gamut RGB. Then apply ARRI LUT on Resolve.

There's no general solution when it comes to color science. All needs testing. And then more testing. In theory, everything will work . Please prove above workflow working as expected. Would be nice.

[2blackbar]

I pushed these colors below yellow towards red in HSL section and saved it as default receipt for 5D2.White balancing is not affecting their hue enough in mlvapp tonemapped preset.
If it would be blue colors then i wouldnt bother but reds are on our skin, they are quite important.

[Danne]

Could you share example images?

[allemyr]

No can't agree , instead of using rec709 or AP1 use Adobe standard space and correct your output color space to 16-235 level instead of full range (0-255)

rec709 is 16-235, and hows that different from setting Adobe standard and 16-235 output?

My experience is that sRGB works best, when viewing videos from Youtube or Vimeo or other on displays. By works best i mean that it corresponds to what you see color like in your NLE and it a 100% match from NLE to online viewing. No color shifts after upload.