Author Topic: MLV App 1.14 - All in one MLV Video Post Processing App [Windows, Mac and Linux]  (Read 1038990 times)

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2007
FWIW, H.264 on EOS M (vanilla Canon firmware) is 3x3 with column binning and line skipping, unless you record at 720p50/60.
Are you sure? In my tests moiree looked more or less identical between 3x5 readout (1736*6?? (16:9) horizontally stretched by 1.67) in MLV and standard H264(1080p). For e.g. the 5D2 I agree with what you wrote. Standard H264(1080p) on EOSM looks way worse than e.g. on 7D, but sensor has the same resolution.
5D3.113 | EOSM.202

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2007
@Veerle: and where is your moiree? Can't see any... (pic=100% zoom in). And would be enough to get 1..5 frames of a scene to not load gigabytes over hours... you can easily shorten the clips with some simple clicks.
5D3.113 | EOSM.202

a1ex

  • Administrator
  • Hero Member
  • *****
  • Posts: 12564
Are you sure? In my tests moiree looked more or less identical between 3x5 readout (1736*6?? (16:9) horizontally stretched by 1.67) in MLV and standard H264(1080p). For e.g. the 5D2 I agree with what you wrote. Standard H264(1080p) on EOSM looks way worse than e.g. on 7D, but sensor has the same resolution.

Yes, I'm sure. EOS M uses 3x5 readout in standby, at 30 fps, regardless of settings in Canon menu (this is unusual about it). However, when recording H.264 at 1080p, it switches to 3x3, where it behaves like all other ML-enabled Canons other than 5D3 (i.e. column binning and line skipping).

Now, how bad the aliasing actually looks like, depends on both how strong the high-frequency details are (how sharp the focus actually is), and - to a lesser extent - on the image processing pipeline (which may be different on different camera models; currently it's a black box to me).

Feel free to do a controlled test and prove me wrong. You will need a static scene with lots of aliasing (e.g. a resolution chart or any kind of fine print), same lens, same aperture, same focus point (triple-check this one!), same exposure, same picture style settings... you know the story. You could also capture a LiveView frame in x5 and in 1080p without moving the camera (e.g. by tweaking this script); from that pair of images, I can identify the pixel binning pattern (but I don't expect any surprises here).

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2007
Yes I know the story. I hope I remember at the day when I don't know what to do else... :P I think better to trust you until then :D Thanks for clarifying, a1ex!
5D3.113 | EOSM.202

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7368
Both issues fixed. It was the cmos4 change that was causing it. Well, will be more cold pixels on higher iso but for now it could be taken care of in post:
cmos 4:


cmos 4 default:


Build posted here:
https://www.magiclantern.fm/forum/index.php?topic=9741.msg219657#msg219657

Thanks Veerle for reports. And thanks a1ex!

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2007
Thanks Danne. Is it possible to change that also in main build? All ML cams I have and get do the black sun since ages (5D2, 7D, EOSM, and saw it on others too). What has it to do exactly with cold pixels?
5D3.113 | EOSM.202

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7368
Cmos 4 is probably something else on digic IV. I can't say what will help for other cams but I just left it alone so it runs on default now. It's maybe used on digic iv so then maybe defaulting it will help your issue too.
Regarding pixels. I reduced amount of cold pixels tweaking cmos 4 but as you see it now turns out it would affect highlight handling.

Veerle

  • Member
  • ***
  • Posts: 113

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2007
Ah right. Focus pixel fix creates interpolation artifacts on this fine structure, no matter which of the 3 interpolation methods is used. So if someone is good in writing interpolation algorithms... we'll find a place for method 4 ;)
5D3.113 | EOSM.202

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2007
For everybody who knows how to compile MLVApp:
new feature! We can now mask the sharpener with a new masking slider. 0=no mask, 100=full mask. The mask detects edges, so only edges will be sharpened, but slight noise or fine structures won't be sharpend. Such a mask can look like this:


And there is another new feature: AVFoundation export can be resized now. This is not really fast, but from quality side this looks not bad.

If you're able to test, please let me know if it works for you, or if you get problems with it.
5D3.113 | EOSM.202

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7368
Ah right. Focus pixel fix creates interpolation artifacts on this fine structure, no matter which of the 3 interpolation methods is used. So if someone is good in writing interpolation algorithms... we'll find a place for method 4 ;)
Hm, this is problematic as it will probably affect overall fine texture in recordings?

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2007
If you like to interpolate a wrong pixel with the help of surrounding pixels, and these pixels are a fine structure, you always will have it hard. Best would be to find the function in camera which eliminates the focus pixels. No idea how Canon realizes this.
5D3.113 | EOSM.202

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7368
Yes, but how is chroma smooth working? Curious how they differ here.

Kharak

  • Hero Member
  • *****
  • Posts: 1022
How do cr2 files from cameras with focus pixels look? where the pixels have been removed in camera, are there any artifacts in fine details?
once you go raw you never go back

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2007
How do cr2 files from cameras with focus pixels look?
I never saw a cr2 with foucs pixels, so I think that does not exist.
5D3.113 | EOSM.202

Kharak

  • Hero Member
  • *****
  • Posts: 1022
You read it wrong.

“From cameras that have focus pixels” i am very aware that you dont see the focus pixels in the cr2. I was asking if there are any artifacts in cr2 image, there where the focus pixels have been removed.
once you go raw you never go back

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7368
I never saw a cr2 with foucs pixels, so I think that does not exist.
Check this @masc :) reply #572
https://www.magiclantern.fm/forum/index.php?topic=16054.msg207726#msg207726

Reg 8900 can reveal focus pixels in a cr2 file. Some vodooing is hiding those but seems blackboxed what. Maybe some behind the scene averaging?
Also seems different streams affects focus pixels showing or not or lesser pronounced.

theBilalFakhouri

  • Developer
  • Hero Member
  • *****
  • Posts: 1013
  • UHS-I
Hello @masc

Can we have Aspect Ratio option in exporting settings ? or as example I shot 1736x976 16:9 MLV video , in exporting settings I adjust resizing to 1920x816 to have in the end 2.35:1 Aspect Ratio but it's not working in that way it compress the height and give 16:9 video needs to be un-stretched to have the correct 16:9 same as 5x3 mode , instead of compressing the height or Binning the pixels vertically; it will be great if it works by cropping the top and bottom parts . .

So can we have it ?

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7368
Ffmpeg can do "padding" the way described.
https://superuser.com/questions/991371/ffmpeg-scale-and-pad

Maybe options to import into different ratios could be useful like in premiere. Preselect 2.39:1 for instance?

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2007
@theBilalFakhouri: if I understand right, you don't want to resize. You want to crop and then resize? Cropping is not realized in MLVApp, because in every NLE this is easier and better than we could ever do it (where to crop, automated moving crop, etc.). So we just rescale (and don't crop). There are presets under "Transformation", where you can auto resize 3x5, 1x3, 3x1,... and real anamorphic footage. If that is not enough, you can use export settings and stretch with that.
5D3.113 | EOSM.202

DavidP

  • New to the forum
  • *
  • Posts: 46
Hi guys, quick unscientific test. Turned out saving as bmfilm and don't use camera matrix matches the most close to my ursa. Only adjusted wb and exposure once I'd imported to resolve and used the 4.6k to extended video lut.


DavidP

  • New to the forum
  • *
  • Posts: 46
It's actually a lot softer than I had imagined... Bit of a shame. Ok for web viewing but jees, expected a more detailed image.

ilia3101

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 983
Impressive how well it matches... and yes looks a bit softer than I would expect, but not massively. Try a shot of something object using a non wide lens (50mm+) with some bokeh in the background, it will look sharp trust me, I always remember those kind of shots looking 100000000x times better than wide shots on my 5D2.

Also how about trying the "3.5K" / crop modes?

DavidP

  • New to the forum
  • *
  • Posts: 46
Yes im sure for tight shots it's fine. Actually though I really hoped to use the 5d for wides as it's full frame but I see that's not really an option. Maybe as a b cam for interviews, talking head shots... but unfortunately i think I may actually have to get a pocket cam to be the b cam as this isn't really sharp enough. P.s I only got 12 second record time with the crop mode. It was a bit more detailed but not worth it for the big crop in.


EDIT:

I added some sharpness to the 5d and you can see for yourselves... not too bad!

flostro

  • New to the forum
  • *
  • Posts: 38
  • EOS 5D Mark III / EOS 3 / A1
Any chance of MLV App supporting legacy .raw files? It would be great to convert old uncompressed .raw files to compressed MLV to save some space.