Messages

Never seen it in all my time with RAW/MLV. If it's only one frame, I think you can easily get away with it if you "repair" the frame by masking the black bar with something like the Wire Remover Tool in After Effects. Good luck, looks amazing!

1 - That's up to you, but I've done several projects for clients with ML RAW in the last year and it did not fail me once. A little bug here and there, so of course do your testing and take precautions (more than one take of anything, check with MLV_Play whenever possible)
2 - With a good card. 24p is relatively easy, stable 25 at 1080p might be a little harder to get.
3 - I've battle tested two Transcend 64gb 1000x: one year, countless writes-reads, never a problem. There has been Komputerbay users with and without problems, but my money is on Transcend. Lexar also has a good track record but I've never used their 1000x cards.

4 - Or would you rather pick the Canon C100 if constant and reliable video camera is needed?
Yes and no. If you want/need/fancy RAW, 5D3 (and if you ask me with 1.1.3 firmware) is the way to go. But, if you are doing client work with little or basic color correction and you cannot afford any tiny glitch whatsoever, I would suggest not using edge bleeding code, even if it is more stable that some manufacturers' factory software. Take into consideration also the different requirements of hard drive space between RAW and AVC/ProRes (C100).

RAW in 5D3 is relatively bug free in my experience and has many advantages, but it all depends on your risk assessment criteria and how comfortable you feel using one solution or another. They are tools, they all can fail, and I wouldn't really compare one with the other.

Which format are you exporting to? Check in the Export tab (Advanced one if using Resolve 11) which settings are you using. There's a parameter Video/Data level, below codec and resolution and framerate. You're probably using Automatic, you want to use Scaled Range (or with some codecs Data rate - instead of Video rate. I think it's the Data one but I always mess those two up, it's one of them for sure). Hope it helps.

But, when I open the DNG sequence, Davinci monitor shows a very saturated and high contrast image, even making some of the roll offs seem like a 6-bit image, and even a h.264 test we did had better colors.

I'm quite sure that's the Rec.709 curve. When you import your DNGs that's the default that Resolve applies to them (unless you change it in your project settings. In Resolve 10 (or 11, but that's a beta), if you change it in the Camera RAW tab to "BMCC Film" and you grade it, you will get nice gradients. When you first set it you'll get a completely flat image, that's what you grade. Even if you adjust the saturation and contrast to match the ACR image, you will get nicer colors and gradients that with 709.

However, I do not know if Premiere or Resolve will read that applied preset since cDNG is a different process than just the normal DNG (to my knowledge).

No, they won't. PP uses Speedgrade framework which doesn't read ACR and the same goes for Resolve (where you can apply a flat profile - BMPCC - and then a LUT of your choice to all your footage. Plus it encodes faster).

Forgot to mention that MLV to DNG Batch Converter v.1.9 generates DNG, not CDNG. So PP won't read them at all, for Resolve it's about the same between the two formats right now (reads both equally).

DNGs from RAWMagic work fine in both the old and new Premiere until you adjust the exposure using the new Lumetri controls. Once you drop the exposure, the pink highlights show up. Can you try this and see if you get the pink highlights? The footage must be overexposed for this to work.

Yes, I've tried tweaking the settings. With both 12bit and 16bit Cdng coming from raw2dng there are pink highlights, they "disappear" (burn) if I adjust the exposure slider to overexpose them (therefore losing detail and DR), in Resolve and ACR they are obviously fine. The flickering issue doesn't show up and the performance is oddly better with 16bit files than with 12bit ones in my system, which is weird to say the least.

My copy of PP 8.0 is doing weird stuff with CDNG files. Some frames randomly flicker and seem to be overexposed (not happening in PP 7.0), there's still pink cast on the highlights (disappears with chmee fix), the files still lag while editing (also when hitting playback sometimes), though image quality and processing seem a tad better (it was performance what I was expecting to get with this update).

I think I'm staying in CC 2013 for now. The flickering issue makes it unusable for me, maybe an update will help with this issue in the (near) future. I crossing my fingers for Resolve 11 (performance improvement, not MLV support).

PD: I tested several CDNG files from raw2cdng and it appears that the old files without 'partymode' are now properly displayed in PP and SpeedGrade. Maybe that's what Adobe was referring to.

Hey chmee, I wanted to offer you some testing results regarding the timecode issue of the WAVs auto-sync in Resolve, I've read that your software uses the datetime of the file to generate the timecode. I've tested this and I found why Resolve it's not syncing some of the files properly. I used two files M03-1753 and M03-1755.

The timecode generated for M03-1713 and its WAV was 20:15:59:00 to 20:16:28:12 (6 more for the WAV)
The timecode generated for M03-1715 was 20:15:28:00 - 20:16:05:18 (5 more for the WAV)

I don't think is a problem of timecode generation. Its consistent for video and audio, and they seem to match but, as you see, the timecodes for the two files are not unique and overlap, therefore, when Resolve syncs them it goes to the first file it can find that has some of that timecode (the first file when shorted by name that has some of that time interval, it defaults to that and ignores the reelname - because audio doesn't catch the reelname from the folder).
The timecode match the date of creation of the file on the hard drive, that means when the .MLV was copied to the hard drive, which obviously makes files overlap because the creation times are usually close.

I believe there can be a quick workaround for this (the proper fix would be emulating the BMCC WAVs) taking the timecode start from the name of the file i.e: "1713". As I mention this would be just a workaround because in the case of having files created at the "same time" in different days the timecode would overlap, also when using different cameras, and, Resolve would go crazy again. But this is solvable. When put in different folders of the Media Pool in Resolve (Day 1, Day 2 / Camera A, Camera B...) And selecting the clips in those folders, then clicking the auto-sync, the two files sync properly and only with the WAVs that have the same timecode in that folder. If timecodes did not overlap in one folder, we should be just fine. It's not perfect, but it works.

Well, in any case, I hope some of this testing offers some insight and helps in solving the problem.

Tech details:
- Camera: 5D Mark III
- raw2cdng 5.0beta6
- MLV / Nightly ML build of February 14th 2014 (I know, it's an old build, but I'm not aware of any changes on the timecode/name/file creation process that can influence the issue reported)

Hi there kareokie, I think I might be able to help.

The XML roundtrip is not an easy process (yet) with ML files but there are two workflow options I've tried with several projects and they have worked pretty well.
Option 1: A really fast computer, editing directly in Premiere CC.
- You don't create proxies, you edit directly in Premiere the CDNG files, the audio syncs automatically if you are using MLV (with some minor frame glitches you might need to manually adjust it). You can also sync dual audio with the "Multicam clip option". But:
a) You need a fast computer that can play the files without lagging (you can try and see what's your performance with a couple of clips)
b) You will have to divide the film in different short sequences. There's a bug in CC, in a timeline with about... 5 minutes of RAW (at least if you use Multicam clips with audio) it will start collapsing. And by that I mean a kind of buffering-lagging when you hit play in any clip. You can avoid it rendering the whole timeline and the audio but it's a crappy way to work. If you use this option, you don't need to create proxies and it syncs back and forth Resolve without any issues whatsoever, but Premiere can be picky with RAW sometimes.
c) If you use Multicam clips to sync the audio you need to collapse them (there's an option right-clicking the clip in the timeline in the submenu of Multicam) before exporting the XML or Resolve WON'T link them. And the audio will not go into Resolve, but that's a minor issue that shouldn't worry you.
d) When you finish you just have to export a XML or EDL and Resolve should import the files perfectly, all of them. With aligned cuts.

Option 2, the one you've tried: Proxies roundtrip
a) Open Resolve, go to your Project preferences (the little gear icon on bottom left), and look for the option to use the folders as reel name, without that it won't link the files when you import them back. If you don't know what I'm talking about, you can read this article (images in the "Abridged version" section).
b) Now you import the dngs into Resolve (they don't need to be CDNG if you are in Resolve 10, you can convert them to 14 bits DNG and you will save some disk space - but these won't work with Premiere)
c) You go to your render/export/deliver tab and select "FCP Roundtrip" preset, you can change after that the file format, but you need to select it because you do need the XML file that is generated.
d) You import the batch of proxies USING THE XML file generated in Resolve (you import the XML and the video files should come along, renamed to .DNG and with their reel names in their metadata). Each batch of clips you generate need to be imported this way, but it should not be problem if you import several XML files of different proxy-exports, which is what you wanted (I've done it, it works just fine). You don't need to use the timeline that comes from Resolve, you can create a new one as long as you use the clips that were imported with the XML.
e) You export your XML/EDL back to Premiere once you are finished and it should match the cuts. Thought, in my experience, this method can lost some clips or even force you to relink them in Resolve if you missed any of the steps.
IN ANY CASE:
1) I recommend you to do a test edit with the workflow you choose and do the roundtrip a couple of times to be sure of how it works and that it does what you want (try speed effects and several cuts with different clips to see if it's working)
2) ALWAYS export an Offline edit from Premiere (if you choose option 1 and edit the CDNG in Premiere the render will take a night or so for a feature, give it 12 hours, the RAW rendering is terribly slow). Import your edit in the Resolve project that you created to generate the proxies (you should generate all your proxies in the same Resolve project, just to be sure) and import your Offline edit (the final render from Premiere as the Offline clip - right click in the pool window in "Import"). Once you import your XML/EDL in conform, all the clips should link. Check with the offline to see that the cuts match properly (especially in the end, any sync issues are better detected over time).
3) Give the Offline render (really important to do this), the XML (if you give them several files, mark the one that is the final cut), the DNGs and a copy of your Resolve project to your colorist. Advise him about the reel name issue, he will need to be aware of this when importing the files or the XML might not import right.
4) If you have a postproduction house, or a colorist, you might want to ask them to create the proxies and XML and advise you about the workflow. They might have a specific way to handle the roundtrips with Premiere and you want to know these things before you start editing.

I hope this clarifies some of your doubts. If you have any questions I'm happy to try and help, and good luck with the project.

@WeekendWarrior It probably doesn't affect the flicker caused by using shadow/highlight recovery, only the grain effect of ACR. The recovery is done individually in each image to "optimize" the result, so that will probably remain unchanged with this update. If you use Whites/Blacks for the recovery instead of Shadows/Highlights there should be no flicker.

Hi there! I'm testing the nightly from the 15th of February in a 5D Mark III and I've had some issues with audio. I have the default MLV settings (2.35 aspect ratio) + MLV_Sound + Global Draw on while recording. The dngs are all fine, but some of the wavs stop in the middle of the recording and others are simply empty (it happens especially in the 3x crop recordings, but those are understandable). I've search around the forum to see if this was a known issue but I didn't find anything. Am I missing something?

Anyway, thanks for the great job. The new format and the audio integration are unbelievably amazing

PS: I tried the 23rd nightly but I had mayor issues with audio stopping and black frames in the middle of the recording (?), so I decided to try the version from the first post of this thread, which I've found to be much more stable and reliable.

maybe i can add an option "Files >4G supported" for those who know that they have exfat and a 5D3.
but this will cause for sure 10 other people asking why it doesnt work and that the option is faulty etc etc.
then there is again a lot of crying here. oh dear...

I understand your concerns, g3gg0, but I would really appreciate to have that option. For me it's much more convenient having one single .mlv file per shot. People using this module (or ML by extension) should know their cards file system and its limitations.

But that isn't important. We finally have synchronous audio and I cannot thank you enough. I've been working with the 'beep' and the slate method and it was quite painful. This makes things so much easier and quicker. For those doing their dailies in Resolve and struggling, there's an amazing tutorial from Abelcine (I know I should be posting this link in the postproduction section but it's much more likely to be seen here). Right now sync in Resolve can only be done manually, in order for it to do it automatically the video and audio need to have matching time codes. Anyway, manual sync it's a minor inconvenient for this awesome format. Thanks again!

Hi there Magic Lantern community! I'm here again sharing with you yet another video shot in beautiful Magic Lantern RAW. This time the location is London and we shot it for the ongoing Dorito's Super Bowl competition. This is the first RAW video I haven't processed with Camera RAW and I have to say Resolve 10 is, in my opinion, the best contender for RAW processing right now. I still have to iron out some issues related to doing the roundtrip with audio but it's way faster and requires far less hard drive space than my previous workflow (Lightroom (16bit TIFFs) - Resolve (proxies) - Premiere (editing) - Resolve (grading) - Premiere (finishing)).

Some technical details:
Camera: 5D Mark III
Cards (2x): Transcend 1000x 64gb
Build: Bot generated nightly build - 3rd November 2013
Lenses: 24-105 f4 L, 50 1.4, 85 1.8
ISO: Interior, shop - 200, exteriors 100-400, interior 1600
Audio: Zoom H4n (Manually synced with slate and the invaluable 'start beep' from the camera)
Software: Premiere Pro CS6 / Resolve 10 (Roundtrip with XML)

Feel free to ask any production/postproduction questions!

I'm loving ML RAW in my 5D Mark III. Simply put, the image coming out of it it's gorgeous and the grading possibilities are endless. This time I've recorded a short BTS video of the last shoot of a young Spanish photographer who does mostly analog photography, Javier Ruiz. The video was shot in 1920x1080p, with several cropped-sensor shots in that same resolution, and it has been graded to match the photographer's style.

Build: July 16th (cb7f14c) - 5D3
Lenses: Canon 24-105 f4 IS L & Canon 100 f2.8 IS L
Cards: 2x Transcend 64gb 1000x
The post workflow is the same of my previous video (http://www.magiclantern.fm/forum/index.php?topic=6718.0)

Any comments or questions are more than welcome 

I think it's something I would use if I were recording H.264. Even though I'd custom WB if I wanted to nail the white point I think the extreme saturation could be useful. I already do something like that using Portrait with the saturation all the way up as a WB-check custom picture profile.

Hi, David!

Thanks for your kind words. I love the Alexa look and I tried to replicate it with ML RAW, I'm glad you noticed it. I think the key for the interior shots was smoking the room, I used a small smoking machine with regular fog and I kept a light mist through most of the interior shooting.

For the interior shots I used: Canon 28 1.8 (although I can't recommend that one in terms of resolution or overall optical quality), Canon 50 1.4, Canon 85 1.8 and Canon 100 2.8 L IS (the last shots of him with the objects and the trunk, though there was almost no smoke on those ones). In the exterior shooting I mostly used my Canon 24-105 4 L IS with a Hoya HD polarizer. In the last shots, as it when darker I put away the pola and at the very end, I change the lens to my 50 1.4 and the 28 1.8.

I metered with the on camera histogram and with my laptop screen (IPS panel) which acted as a monitor through the usb connection and EOSUtility, as I mentioned in the first post. I relied on the histogram and I checked it every time I added the smoke to keep the overall contrast and luminance levels where I wanted them. I recommend you using a contrasty and saturated profile while recording RAW, it's easier to know where your mid-tones are and it helps me to distribute the dynamic range better.

I hope you find the information useful,

Regards!

I'm having the same black bar problem mentioned above with Lourenco's 16th July build (quite large, about 15% of the right side). It has happened in many of the cropped videos that I've recorded but not in every of them, some are clear and have the full 1920x1080 image. I don't remember moving the AF box around nor I change any settings during the recording. In the global draw preview it displayed correctly. Also I think I played at least one of the affected videos on camera and it covered the full screen. Could it be a raw2dng bug?

I'd be happy to upload reference frames and provide any info to help to solve the issue. As a side note I had "Small hacks" ON during the recording.

I should reiterate... STUNNING work dude 

Thanks a lot!!! I'm really happy you like it!

If you had shot any bigger than 2.2:1, you wouldn't be able to monitor through eosutility?
So if you wanted to shoot in 2.35:1 or 16x9 and you hooked it up to usb, the image would no longer show in EOS utility?

2:1 is feasible with that build but it buffered sometimes with my cards. We also wanted a wider aspect ratio without going all the way down to 2.35 and 2.20 was safer, so we went with that. 2.35 is absolutely reliable with EOSUtility and my cards.

What ops did you do in ACR or was it simply to get 16bit tiffs.

Import -> Whites -100 / Blacks +100 /// Slight noise reduction depending on the take -> Export

As I have previously stated in the forum, I was truly excited when the RAW hack appeared since it opened a new world of possibilities for my 5D Mark III. We shot this music video a couple weeks ago with the 25th May Build. I've found the hack to be incredibly reliable and I haven't lost a single file due to corruption or similar issues to the date.

We needed to monitor the video output so we shot in 2.20:1 which allowed me to feed the image through the USB to a Windows 7 laptop running EOSUtility while simultaneously recording continuous 24p. I also shot some 30p scenes. The slow motion shots in the end of the video were shot in 1920 (2.20:1 squeezed aspect ratio). The ISO employed through the shoot goes from 100 to 6400 (last 2 scenes, obviously denoised in postproduction).

The cards I used were two Transcend 64gb 1000x cards (not the fastest but they never let me down and they handle 1080p RAW quite well). I had to offload one card per shooting day (a total of two shooting days) but the cards read fast so it wasn't a problem.

The post workflow I chose was quite a nightmare but I decided to cope with it to get the ultimate image quality. I used RAWMagic to convert the .RAW files to CinemaDNG Files. At the time Driftwood hadn't published yet the Resolve workaround for the pink fringing so I imported all the frames (almost 83.000) to Lightroom and output them to 16bits TIFFs. Then I imported those into Resolve where the final grading was performed. Since the Lightroom processing took over a day (in a quite powerful desktop computer), I did quick Cineform proxies with RAWmanizer which I used for the offline edit. In the end, I didn't have to wait per se, I just used some more batch-processing Now, let's talk about the hideous roundtrip. I can't understand how Resolve can't match by itself files with the same name and the same number of frames. I mean, at least it could try. I had to force-conform each file in the project which took "just" 3 hours (1 hour per minute of final video edit approx.).

But, when the files are in Resolve and you see the RAW information... all the pain is gone. I LOVE the format, I couldn't be happier with the results and the image quality. I'll certainly be using it A LOT in the future. It's really hard to go back to H.264, all my shots seems out of focus now!!!

I wanted to share with you the video, I'd love to hear any comments about it! Of course, thanks again to all the developers and the wonderful ML Community. This video couldn't exist without you, guys.

Guys is the pink fringing the same as what I reported to chmee? Here's how I see it...

I think it is. It can seem a "fringe" around contrasty areas of the image. I've tried changing to BMC Film interpretation in the Camera Raw tab and it "solves" it. But I don't really like how it re-interpret it as BMC.

Thanks a lot for this tool chmee!!! I'd love to have batch enconding and renaming but it's already great to been able to do direct RAW->CDNG processing.

It's not part of the request by itself but I though I should update this topic in case someone reads it looking for a solution.

I've tried monitoring with a laptop and EOSUtility while raw-recording in previous builds (19th) but when I hit record the screen froze. I'm happy to report that with the build I'm currently using (28th - 7fdfc08) I'm able to monitor and raw-record simultaneously. The writing performances goes down but I can record 1920x960 2:1 continuously without any problems.

http://www.imageconverterplus.com/how-to-convert/raw_exr.html

anyone used this before?

I'm on a mac so I can't.

I've just tried it. It doesn't convert the .RAW, since it's not only one image, but it does convert the .dng to 24-bits TIFFs quite fast (no options to recover-change anything, though). Maybe it has some utility for fast preview or alternative proxy generation (I'm talking about the small jpeg option).

To my knowledge there's no video editor which is able to import DNG sequences in a usable way. Your best option for what you want is to convert the .RAW files to Cineform RAW ( http://www.magiclantern.fm/forum/index.php?topic=5479.0 ). You will have real-time playback and a lot of grading flexibility (not as much as with After Effects route, though).

I don't understand the need for pre-grading raw (preview while editing?) but I do something like that with my current RAW workflow.
I import the frames into lightroom (all of them). I recover the highlights all the way down and some of the shadows, I up a notch whites and blacks to make sure nothing is being clipped and I export them as TIFF sequences in 16bits that I can import into DaVinci Resolve. I batch process all the TIFF sequences to small 720p proxy files and I edit them in Premiere ( https://www.youtube.com/watch?v=-0S2MJT9wDY ). I do roundtrip to Resolve again and do the final grading. I still have some problems reconnecting the footage because of the file naming.

That way you are editing the TIFF-proxys which already have shadows and highlights "recovered" and should have a "proper" white balance for preview purposes. It a really slow workflow and requires tons of disk space but you can do it overnight. I've found it's the one which gives the best image quality to my eye (despite the flickering issue, which is not that noticeable in most sequences in my case http://www.magiclantern.fm/forum/index.php?topic=5710.0 ).

Another example for per-frame metadata: I wanted to add fine-tuning offsets for smooth panning with increments smaller than 8 pixels. But since I did not see any video sample with panning, I guess nobody uses it, so it's not worth the hassle.

I would love that. Digital panning is a great option when there's no physical space to put a slider (or for tight budgets). I don't use it more because I can't smooth start/stop and, also, because when I enter zoom mode the images pixelates and goes black and white, so I can't see on the screen where I'm focusing. But that only requires me to make focus marks and that's not a real problem.