Messages

[zoe-lossless-codec]

Thanks for the clarifications.
currently I'm trying with a video that contains 659 frames
I wait until the message " export is ready" appears. this advice appears after only a few seconds that the bar has reached 100%. the duration of the complete export process is about four minutes for entire video.
maybe a record video could explain better than my bad english:
in this video there is the whole export procedure. I exported only 100 frames of not edited video, to speed up the process.
https://youtu.be/b6iqG3AXDxI
Thx for help!

Hm... what is the size of the file? Can any other application play it (VLC for example)? Does it work better without aliasing filter?

why a use a very chunky AVI codec (1.5Gb/s) ?
IMO there's far better open source  AVI codec  out there ,  it's nearly impossible be to work with .

Japp... I think the same. Someone had a wish to have that in MLVApp and it was easy to add... this guy will have a reason to use it.

the codec is called  "zoe-lossless-codec" to download codec or source code go to https://github.com/edanvoye/zoe-lossless-codec , supports 32 & 64 bit windows

Hm... that's coded for Visual Studio. Unfortunately that is not very compatible to MinGW compiler (we use), if I understood right. I think we would have to "translate" some classes to Standard C/C++. Then it would be also cross platform. But that is much work!

Applying the raw processing just without the curve during export to ProRes or other video codec?

We could completely switch off processing... then you'll have a dark greenish picture and you nearly can't see anything (I made some performance tests in this way)... I don't think that is a good idea to export that.
I also would like to add BMD Film color space... but I don't find some useful specs about. I would need a formula from linear to BMD Film.

[sec]

Hi

Just testing:

v0.17 freezes up (App not responding / beachball) with 3-pass anti-alias when export to .mov reaches 100%  Resulting .mov file is unplayable.

v0.16 Was working fine on Export.

Mac 10.11.6

Hi!
Same here but on Windows 10 x64.
If I export in Appleprores 4444 with smooth aliasing 3 pass + unsharp mask. trying to import the file into adobe premiere cc 2018 as the error message "file has no audio or video streams" keeps appearing.
No problem at all with "smooth aliasing" option turned off.
Do you have any suggestions?

3-pass anti-alias was not existant in offical v0.16. Have you been waiting until MLVApp brought "Export finished"? I guess not. When process bar is at 100%, MLVApp rendering is done and ffmpeg filter starts its 3pass work - it is not possible to measure the progress here, thats why you see the 100% a very long time. For a 10 sec clip you'll see the 100% for 5-20 minutes until it is done, depending on your hardware. If you stop export before, file is not playable.
Other suggestion: your clip was smaller than 10 frames - as posted some posts above about 10frames is minimum to get a working export with 3-pass anti-alias.

I like to announce that v0.17 is out now:

-fixes for dual iso (processing and export)
-new highlight reconstruction for dual iso clips
-added 1D/3D LUT (.cube) support
-added sliders for RAW black and white level (& export corrected MLV / DNG)
-double click sliders for default value
-added white balance picker 🎉 (for grey and skin tone)
-added smoothing filter via ffmpeg export (cleans aliasing artifacts)
-added uncompressed AVI V210 export codec

Minimum OS:
-OSX 10.8.5 Mountain Lion
-Windows 7
-Ubuntu 14.04 LTS

https://github.com/ilia3101/MLV-App/releases/tag/QTv0.17alpha

Have you searched for a solution? There are sooo many solutions in this forum. A few:
- Convert MLV with MLVProducer to a format you like
- Convert MLV to CDNG with MLVFS and use Resolve or Lightroom or...
- Convert MLV to CDNG with mlv_dump and use Resolve or Lightroom or...
- Convert MLV with Switch to format you like
- Convert MLV with MLVApp to format you like
- and many more...

There's an issue with pink highlights that affected the ML build you were using but that can be corrected. MLV App has a RAW White Level slider but I couldn't get rid of the pink highlights.

Really? No such problem for me. Just lower White level (while RAW Correction is enabled) and enable Highlight Reconstruction. Pink highlights are gone...

One of the "interesting" things I noticed when running the "Smooth aliasing" scripts is that when using the 1 pass option it created 8 TIFF files and with the 3 pass option it saved only 4 files. Ran a 3 pass option on a much longer clip and it chopped off 6 frames so it looks like you'll loose a few frames when using the "Smooth aliasing" options.

Japp... but I can't tell you yet where the "lost" frames are and why they get lost. But had that issue as well. It seems each ffmpeg filter call cuts 2 frames.

@masc - So the reason we can't fix the bricks is because there is no motion? This is also an issue when trying to shoot a "super-resolution" shot. You need some movement or it won't work. I made a post about this a while back, Large image, low fps vs. Small image, high fps.

My filter idea works in a similar way as a motion blur, but without really bluring. So if there is no motion, nothing will happen.

here i have a short mlv of color artifacts of trees. can anyone check if it can also be removed
https://drive.google.com/file/d/1dQHbwIzPQaaJkDtUXLwpNPc6CMQyuULa/view?usp=sharing

Moving trees is perfect for the ffmpeg filter:
without

with (3 pass + unsharp mask)

Link to two examples: https://we.tl/QRiXYw8l8z
I realised I shot these back in October, time flies! I had just purchased this second-hand EosM and was exploring its potential. I believe I was recording in 3x5, but I would have to ask you guys to make sure I didn't just screw up!

Hope it helps

Whohooo... thanks, that is indeed extreme! And here we also see the biggest enemy of my ffmpeg based idea: a tripod. Without movement in the picture that does not work very good.

Why wait. Double click this and build your own updated version:
https://bitbucket.org/Dannephoto/mlv_app_compiler/downloads/mlv_app_compiler.dmg

https://bitbucket.org/Dannephoto/mlv_app_compiler/src

Correct! It is so easy! @Danne: I linked it now in our github readme

[EDIT]

Do you not get these results in your preview? I get the same result when exported as well.

No, no such artifacts also for me. What version of MLVApp did you use, and on which OS?


EDIT: STOP! I get the same result you have using v0.16. This version did not completely auto correct wrong black level of your file. The blacklevel of your file is over 5000... but it should be 2048. After correcting black level all is perfect (latest repos revision does it automatically).

Masc,
Thank you for checking out the files that I provided.  Right now, I am using the 016 version from the first page but work is still in progress.  As long as we are clear about all resolutions and center offsets, I will be more than happy to provide more test files as needed. In this way, it will be possible to refine the focus pixel maps at all resolutions.

It is always good to have newest stuff to test... thx again! v0.17 is in work for everybody and will be another step forward We just have to get some little things nice atm.

Thanks, it worked. I do however get weird artifacts on clips where the black level is wrong. When corrected with mlv_dump I don't get these.
It's like pink highlights/dots around edges etc.

Could you provide a MLV (shortend to some frames is enough for analysing) please?

These include 2520x1304; 3096x1320 and 4056x2280 resolutions.  Some of them include a few overexposed areas for you to check white levels that in my opinion, do not fully work yet.  I still notice a slight pink cast when I reduce the exposure or Highlights in MLVApp.  I have also included two 4K files shot in Dual ISO 100/800 which now again becomes very relevant, interesting and important for smooth timelapse videos.  Note the absence of noise in the dark areas of these Dual ISO shots.  This is really spectacular!

Dfort, please have a look at the shot where my daughter wipes out the table after breakfast.  In my opinion, there are still focus dots on that clip.

Thanks again for the clips!
Where do you see focus dots in the clip with your daughter? I can't find any... The pink highlights were gone in all shots, except for the last one. Note that I did not test with v0.16, I tested with the latest revision from the repos. For the last clip I changed dual iso highlight reconstruction in a way it works also for this clip now. For dual iso clips it is much more difficult than for standard clips. I know that my algorithm for dual iso is not perfect: if there is a bright pink in the picture which is no clipped highlight (e.g. a flower), it is filtered as well.

For the higher resolutions than 2.5k there is no focus pixel map yet there... so it has to be found again.

All unused focus pixel maps removed. Thx bouncyball & dfort - your work works great!

@dfort: thx, this works perfectly, also with MLVApp!
@bouncyball: can you also update the focus pixel calcuation for such resolutions for EOS 100D, please?

Thx @IDA_ML for the files. Highlight reconstruction works for this files. In the appartment file in the furniture window you can see a difference when toggling, in the city clip I only see that something happens in the histogram.

Thx @dfort for your explanation, but I don't get that to work. All focus pixels are still there. I tried exactly what you wrote, but MLVApp and MLVFS won't handle the focus pixels. MLVApp works with the same fpm files as MLVFS normally.

@IDA_ML: the focus pixels issue not really an issue - all is working - all we need are the focus pixel maps for these resolutions. These have to be found... if we/you have them, the MLVApp version you have now will handle it already.

@bouncyball: are we able with our algorithm for focus pixels to calculate the map somehow? Or do we need to know the pixels first?

[Edit]

@Danne: is there a way to convert the .map file to .fpm (FocusPixelMap)? If I look into these files it looks not so different, .map has just a 0 in the end of each line. The point: .fpm is supported by MLVApp. On windows just copy the fpm next to the exe, on OSX copy it to MLVApp.app/Contents/MacOS/ , then restart the app. The name of the map must be (if I remember right) camID_width_height.fpm (e.g. 80000331_1872x1059.fpm). Then focus pixel removal should work for this camera and this resolution in MLVApp, what should look way better than chroma smoothing and what should be way faster too.

Edit: could you please try out if that works? Just deleted the "0" in the file and renamed it...
https://www.dropbox.com/s/ff3hzq5h1f7nwii/0x80000346_2520_1304.fpm?dl=0

@masc since we're already talking about encoding, about our discussion some time ago on color management... no improvements? I see advanced color management can't be done with ffmpeg, but at least it supports bt2020 out of the box...

Hmm... it supports not really... it only supports adding the metadata, but that does not mean bt2020 is inside. You'll need a matrix conversion inside the ffmpeg command. But that won't "bring" more quality, because the input & encoding is still the same. There is this colormatrix=... command - this one works in 8bit - so you'll loose more than you'll get. You also can adjust colorspace via scale filter, this one works with higher bitdepth. Atm we convert and export to bt709 in this way. Otherwise you can export with AVFoundation, if you don't like bt709, or you must color correct in another software if thats not enough - you're right.

[Filmmakers]

Filmmakers
¿Any tip for someone who has a 5D MK2 (mlv 2.0) and wants to shoot a shortfilm in raw and then post-edit in Adobe Premiere? In terms of LUT's order..if it's more important to apply a LUT in MLVAPP or export (proRes 4444) and apply the LUT in Premiere?

¿Which LUT for general color-grading do you guys apply? MLVAPP or in a video editing software (such as Resolve, Final Cut, Premiere..). I have dozens of LUT's, but don't know when to apply them, or if there are some very common on this community.

I would add the LUT at the stage where you color correct / color grade. So if you do this in MLVApp, add it in MLVApp, if you do this in Premiere, do it in Premiere.

@50mm1200s: this is what mediainfo tells:

Code Select Expand

General
Complete name                            : /Users/masc/Desktop/M05-0608.avi
Format                                   : AVI
Format/Info                              : Audio Video Interleave
File size                                : 2.79 MiB
Duration                                 : 43 ms
Overall bit rate                         : 544 Mb/s
Writing application                      : Lavf57.71.100

Video
ID                                       : 0
Format                                   : YUV
Codec ID                                 : I420
Codec ID/Info                            : 8 bit Y plane followed by 8 bit 2x2 subsampled U and V planes.
Duration                                 : 40 ms
Bit rate                                 : 581 Mb/s
Width                                    : 1 856 pixels
Height                                   : 1 044 pixels
Display aspect ratio                     : 16:9
Frame rate                               : 25.000 FPS
Compression mode                         : Lossless
Bits/(Pixel*Frame)                       : 12.000
Stream size                              : 2.77 MiB (99%)

Keep adding power to this app and eventually there won't be a need to use anything else except a NLE!
With each new version, I find myself able to use more clips straight from mlvapp!

Thanks. That is our goal and what I do as well.

Masc,

I have the 100D and "Highlight reconstruction" does not seem to do anything on MLV files from that camera either.  Slightly overexopsed highlights stay pink all the time, no matter what I do.  I use the latest v.016 of MLVApp.  I hope, this issue can be fixed because Highlight reconstruction is a very important function.  Would you like me to send you some 100D footage?  If yes, please write down the command that I can use to shorten the clips here.  I cannot find it.

Sure, I take all I can get for testing, debugging and further development! THX! (I bet whitelevel in the file is wrong.) Shorten the clip via CutIn / CutOut and export as MLV.

Uncompressed AVI in v0.16 - 8bit 4:2:0
Uncompressed AVI by bouncyball - 10bit 4:2:2
ProRes ffmpeg - 10bit 4:2:2 or 4:4:4
ProRes AVFoundation - 12bit 4:2:2 or 4:4:4

@Enrico: 10bit from your sensor is different to 10bit processed video (in terms of values which are represented inside these 10bits).

I tried chroma smoothing with MLVApp applied to a 3096x1320@15fps file from the 100D to remove focus dots.  The 2x2 setting dampened the dots but did not remove them.  The 3x3 setting killed them totaly but I did not like the result.  The look is too clean, plastic and digital.  The effect is very much like noise reduction.  Is chroma smoothing supposed to work that way?  I think, focus pixel maps do a much better job.

Chroma smoothing and focus pixel removal are two very different things. Yes, chroma smoothing filters also focus pixels... but also a lot more (the algorithms does its job for each pixel in your picture). Focus pixel removal only removes focus pixels, so it is a lot faster and looks better for this task.

@Enrico: Uncompressed AVI is processed 8bit video. So you have to adapt whitebalance and exposure, maybe more to not loose too much information. Again: if you do that, it's no big deal. Filesize plays not the biggest role in terms of quality In cDNG the 14bit RAW from the sensor is untouched. Sure that is slower in Premiere... ProRes is somewhere between: 10-/12bit processed video (depending on chosen encoder).

With this little app you can look into the clips and look what exactly is coded: https://mediaarea.net/
AVI uncompressed is not AVI uncompressed. ffmpeg offers more than a hand full different uncompressed AVI... we've chosen the "rawvideo" one. But I still don't understand why you've chosen AVI instead of CDNG when working with Premiere. Do you color correct in MLVApp? Then it could be useful.

In this particular clip the highest value is 5632, whitelevel is 5586.
(5632-2047) << 2 + 2047 = 16384 ... so exactly 1 to high to work?!

Hehe, you're better at maths! (and know what you did). Yes, commit works as well!