Messages

Hi,
I think MLVProducer is discarding timecode track of MLV files. Every file I convert starts from 00:00:00:00 timecode. Timecode is essential track to match back originals.

Using 2214 Alpha INTEL version.

https://vimeo.com/137039292

Prores 4444 will be better than 422

I can see major of cross dissolve you have done isn't working properly. I see the last dissolve frame just popping out rather than smoothly go away. Was it intentional?

Render audio separately from Premiere and import it manually in resolve. It would be better if you merge everything to one Video track and delete all other audio and video tracks and export XML. I don't understand why would u import audio into resolve. Wouldn't it would be better to do only visual related stuff in resolve > Render to DNXHD then import the Video master file to After effects and import separately the audio and sync it back (off course with beginning beep and end beep sound to be in precise sync) and then prepare Digital Masters? (DCP, bluray, DVD, Youtube?)

So you want to use Premiere as finishing tool.. What i do is bring all the final cut from premiere to resolve via XML and do finishing in Resolve. If there any AE files, i render it out and place in resolve so that i get precise color. For DCP i render as j2k, for any other Bluray/DVD/Youtube i render it out DNXHD 220 mb and then do authoring in either premiere or AE.

Why do you send from Resolve to Premiere after grading? For finishing?

Hi chmee,
There's no alternate to your tool. What about commandline version of your software? So that it would be robust? Or Linux based tool? So that i can dedicate a server for the job and not appoint a person to watch crash and reload the job?

And about filename i would propose REELID and Filename should be the same.

Regards

Mohan

Why would anyone need date/time in place of shotname which has to be in sequential? Isn't it better be in metadata section? Even if its recorded on filename whats use anybody can make it out with?

And also a big no no to MLRAWVIEWER software's prores mov files. As it doesn't output timecode. Maybe useful for people who does Editing and Finishing in one system or software.

[Mohan]

Hi,
I did my entire movie with 5D Mark 3 RAW. Taught of sharing my experience with you guys up to post.

CAMERA: 5D Mark3
Lens: CP2
Build: magiclantern-Nightly.2014Oct07.5D3113.zip

------------------
PRODUCTION
------------------
I had an onsite DIT who was doing footage dumping and conversion onsite.
Problem #1: The MLV filename were not sequential. My DIT was in doubt if camera assistant was erasing the clip which it wasn't the case. Maybe some bug?
Problem #2: Using HDMI Monitor big No No. Recording was constantly stopping.
Solution #2: Use AV Cable (SD Resolution). No dropouts.
Request #3: In-Camera format to Ex-fat will be big plus. Avoiding Laptop when DIT is away.

------------------------
POST PRODUCTION
------------------------
Problem #1: For faster and direct conversion to offline (proxy) files I used Adobe media encoder (with support of Drastic). This method is faster and i don't have to go through DNG's extraction and then conversion to offline. But Adobe media encoder workflow gave additional job in my post grading session. As media encoder don't write any timecode into the proxy Resolve nor Scratch doesn't Read my XML very well.
  Solution #1: Extract DNG's using RAW2CDNG. (Don't throw 100+ files on it. Crashes often.)Use Davinci Resolve Lite to do proxy conversion to DNXHD 720p 60. Use REEL Name, Source timecode burn-in.

Problem #2: Related to scratch reading directly MLV files. Scratch uses Drastic inside which does very poor job of reading. I get lot of artifacts. Poor Demosaicing algorithm.
  Solution #2: Extract DNG's using RAW2CDNG.

And don't use MLRAWViewer to convert DNGs as it doesn't include REELID and CinemaDNG format conversion. RAW2CDNG 16bit gives 3+mb individual DNG whereas MLRAWViewer give 2 mb approx. (donno what bit depth it is exporting at)

I Hope somebody may find this useful.

Regards

Mohan
DOP, Colorist.

Can you implement other types of encoding formats like dnxhd or photo-jpeg codec formats? photo-jpeg format will be useful for proxy editing workflow. And also a timecode and filename burn-in (overlay) in the final exported mov file?

Hi folks,

I haven't seen any definitive guide on the internet on how to use an anamorphic lens on a Canon 5D Mark-whatever, and I'm wondering if anyone can help me out with resources so I can make a guide here.  Basically I'm interested in leveraging the entire sensor for 2.35:1 and squeezing in post, and I just can't find anything terribly useful - doubting I'm the only one, too.

I'd love to make this a repo... please help!

Tell me which anamorphic lens you are planning to use? If it is PL mount based Kowa anamorphic lens then you have to mod your 5d to allow PL lenses (http://cinemods.com/)

And no. You are not only the one... !   

mm, I don't think so

here's a frame with ETTR. Does not look much better

http://l.bitcasa.com/9K6i7-TX

your expose isn't correct. Spot meter with your camera and +1 stop you will get good exposure in this particular shot. IMO ETTR does expose by taking everything in the frame. And hence your problem. Or get a light meter and expose properly. You won't get any noise in the shadow.

I feel like this is the place I should ask, but it doesn't relate directly with the discussion on the above posts...

I'm using 70's and 90's Anamorphic LOMO Zooms on a 50D, great looking thing, and shooting 4:3 (1344x1008px), but this area includes the black bars on top and below the frame. I can take them out with "ClearOverlays", but then I don't control my framing on a moving shot, for example.

Any ideas, clues or suggestions on how to remove these bars when shooting raw video? (same goes for the 5D3)
Thanks a lot!

I'm desqueezing in post, and the loss in image quality is almost unnoticeable.

Taping on monitor would be quick and easy way.

Resolve doesn't decode DNGs properly. Don't do any workaround with it. I use photivo to generate 16bit tiff with lowering the contrast and using Adobe ICC profile (not linear) so i could get log style image. From there you can go anywhere > Smoke > Resolve > Blender > Scratch.  This is the cleanest workflow i have come so far.

Anything under 1920 wide by 1280 high (on a 5D3) can be shot normally.  If you need high resolution, then you must shot in crop mode.
Currently the closet you can get to 1440x1080 is 1472x1104, which is still 4:3 aspect and can either be cropped or resampled in post for the same result.



I've only brushed over this thread, but there's a difference between aspect ratio, and shooting anamorphic.  Some might be happy with the 2.35:1 aspect, others might want to do it with anamorphic lenses.

Thank you Audionut.
It would help a lot if we get 1920x1440 / 1472x1104 for shooting with anamorphic. If i get 1920x1440 i would later scale it to 2048x858 for cine out complaining to DCI specification. In mean time is there any build so i could shoot with 1472x1104?

And this is my anamorphic lens test video. Was in hurry and didn't have tripod while testing.
Workflow:
Shot RAW, (by mistake 1920x818)
Extracted with raw2cdng 16bit linear
imported dng to photivo
exported to 16bit tiff
imported and graded with scratch

I don't think that this attitude will help you with getting devs to help you. Might be better for you to keep using professional stuff like REDs with master primes anyway if you want to look professional on a professional set.

And you think that putting a projector lens in front of an lens would yield good results? I don't think anybody would have projected those footage on atleast 36+ feet screen. If i want to match to red/ alexa cam the least i can do is stay away from these projector lens and shoot straight with the lenses which are on the red/alexa body.

I don't want to start a big discussion here, but the way data is encoded in the image, stretching the image horizontally produces a less artifacted image than stretching it vertically so going from 1920 to say 2500 will still look quite acceptable, especially coming from RAW.

https://vimeo.com/66574661

Realistically, most people who want to shoot anamorphic on a DSLR are going to do it with a projector lens. True Kowa anamorphics (the type you're talking about and which can run in the tens of thousands of dollars) can't even mount on an EF DSLR!

In this particular context, shooting using anamorphic lenses is very much about the flares, the oval bokeh and squeezing a higher resolution out of the sensor without compromising the IQ too much. Otherwise why not shoot in 1920X817 and save yourself some CF card space and hassle? If you want "cinematic" feel out of your glass, you'd probably be much better off investing in some vintage Zeiss or Leica with mounts which can be converted to EF without too much fuss.

I am using modified PL mount Mark3 body, and i know kowa doesn't mount directly. As an cinematographer i have used Master primes, ultra primes for my various project with Red cam, and i don't want these flat lenses.

As far projector lens is considered where most people shooting anamorphic, i have been there and not satisfied with end result and i don't want to do amateur job on professional set.

I don't want to start a big discussion here

And now, if you don't want you to start discussion then stay away from my thread.

What he just said!   

The idea of recording an an inferior resolution than what the camera is capable of giving you, just to that you can stretch the footage back into what the camera could have recorded in the first place seems bizarre.

The way to record anamorphic on the 5D3 is to record at 3:2 resolution (1920X1280) and then stretch that out in post to roughly 3008X1280 for a 2.35:1 ratio. That assumes that your anamorphic lens is a 1.5X stretch and not a 2X stretch which will give you a much wider image that you'll need to crop in post.

Anyway, the point of using an anamorphic lens is to increase resolution, not decrease it. Yeah, I know, you also get the nifty horizontal flares and oval bokeh...but seriously, resolution should be the most important.

Stretching doesn't yield good result. And the point of using anamorphic lens is not to increase resolution and not about flares or oval bokeh, its all about giving film feel to it. I am talking about true Anamorphic lens (kowa 40, 50, 75 ,100) and not about any projector lens, which produces sharpest image.

Wouldn't that yield 1440x612, not 1920x818? Shouldn't you capture at the final horizontal resolution you want so you don't have to upscale and lose detail? So you should be capturing at something like 1920x1440 and compress it vertically in post instead of stretching it horizontally. But I believe the max recording resolution of 1:1 crop on the 5D3 is 1920x1280. My point is record as close to the final resolution as you can to minimize the need to stretch your footage.

I agree. If it does 1920x1440 or 2048x1536 res then it would be great. Then ML will be creating Mini Arri Alexa.. Will any developers hop in to comment?

While shooting anamorphic to get exact 2.35 ratio we need the resolution be recorded at 1440x1080. So that it would fit to 1920x818 cine 2.35 ratio. So is it possible to implement the custom resolution / 4:3 ratio?

Since the DNG format was created by Adobe I thought that Adobe software should handle DNGs very well. And from my experience ACR has a great debayering algorithm. So I'm curious: Which software do you recommend?

photivo
Total control over a raw file.

If you aren't going to release your movie in theaters then Adobe is enough.

Yes, it works really well with ML RAW

There is some nasty pink cast in your footage. Use Raw2cdng software and output to 16 bit linear. And don't use Adobe AE, or any of the Grading suite to directly import dng. These software lack control over the DNG and their debayer algorithm is bad.