Author Topic: Uncompressed 14-bit RAW video testing - 5D Mark III  (Read 586291 times)

aaristim

  • New to the forum
  • *
  • Posts: 13
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #425 on: February 14, 2014, 05:39:57 AM »
I had the same issue. I realized that my SD card was locked, and basically was reverting my settings every time I turned the camera off. I think that I had 'Global Draw Disable' on Allow (or maybe it was turned off) saved as a setting. In any case, I switched it (also, I think that I switched Global Draw to 'Live View Only') and all was well.

I just checked my settings and my SD card is unlocked and GD in Live View. Still having that bug

ayshih

  • Contributor
  • Senior
  • *****
  • Posts: 266
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #426 on: February 14, 2014, 05:44:53 AM »
On the Feb 8 nightly on the 5D3, I have GD on, but I set it to turn off during recording. Well, GD will be on, then I hit record, but once recording stops GD doesn't come back on until I turn the camera off and then back on.

Bug or am I doing something wrong?

Can you (and others) try the latest nightly build (Feb 14), which includes this commit?  When mlv_rec turns GD off during recording, it should now come back after the recording stops.
Canon EOS 50D | 17–40mm f/4L & 70–300mm f/4.5–5.6 DO IS | Lexar 1066x

aaristim

  • New to the forum
  • *
  • Posts: 13
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #427 on: February 14, 2014, 06:28:00 AM »
Can you (and others) try the latest nightly build (Feb 14), which includes this commit?  When mlv_rec turns GD off during recording, it should now come back after the recording stops.

Thanks! Working now :)
However, i noticed the recording time at 1920 x 500 60fps has decreased from about 6s to 5s. Any idea about what is causing that?

eyeland

  • Member
  • ***
  • Posts: 162
  • 5D Mark Spree
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #428 on: February 20, 2014, 04:53:19 PM »
I'm using the Dec 27th build and sometimes i have to remove and return the battery for the camera to boot (simply switching it on and off doesn't help) does anybody else have this issue?
I currently started experiencing the same or a similar issue when shooting MLV. My camera freezes with Err 70..
So many things change all the time that causal determinations becomes rather tricky.
I believe I am on the dec 12 build. Lurencos last build from the autum 2013 worked well for me, but I changed to a newer build to test MLV. Seeing as Lourenco seems to have stopped updating the builds, has anyone taken over that task or do you guys "just"  play around with nightlies now?
Another thing, do we discuss MLV here as well or should that be reserved for the dedicated MLV thread?
Daybreak broke me loose and brought me back...

Midphase

  • Hero Member
  • *****
  • Posts: 712
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #429 on: February 20, 2014, 06:37:32 PM »
Can you (and others) try the latest nightly build (Feb 14), which includes this commit?  When mlv_rec turns GD off during recording, it should now come back after the recording stops.

I am testing the latest nightly right now. I'll report if there are any problems.


dude

  • Member
  • ***
  • Posts: 205
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #430 on: February 24, 2014, 04:12:03 PM »
Tried the Feb 24th.,
when recording raw, there are no grey bars which show the recorded area. Or do i miss something?

kawasakiguy37

  • New to the forum
  • *
  • Posts: 8
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #431 on: April 17, 2014, 11:12:36 PM »
Is it normal to drop frames for the first 5-10 seconds, and then have no problem with continuous 1920 x 1080 at 23.976? Cant for the life of me get it to work perfectly from the get go.....I've read this entire thread plus many others

anandkamal

  • Freshman
  • **
  • Posts: 81
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #432 on: April 20, 2014, 11:25:31 AM »
hi, is 5d3 1.2.3 no more a bleeding edge? can newbies try installing the ML on 1.2.3 5d3? from what i gone through the threads, it seems like .raw works pretty well than .mlv which currently is slightly unstable. Sorry i got to go through the entire thread to find the latest update.

rainless

  • Member
  • ***
  • Posts: 152
  • Don't Leave Without Pictures
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #433 on: April 20, 2014, 09:41:44 PM »
Yeah... That's kinda why I created this thread: http://www.magiclantern.fm/forum/index.php?topic=11243.0

Anyway you should always just TRY whatever the latest version is.  If you don't like the way 1.2.3 runs you can always downgrade back to 1.1.3.

Both of them work just fine for a lot of people.
The Gear - Canon 5D Mark II, Yongnuo 565EX flash, PhotoSel 3mx3m backdrop stand with 3mx3m muslin backdrops. Elinchrom D-Lite 4 it studio lights, some big-ass 110cm reflector. Unlimited German Models

mir

  • New to the forum
  • *
  • Posts: 2
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #434 on: May 30, 2014, 06:25:11 PM »
Is it normal to drop frames for the first 5-10 seconds, and then have no problem with continuous 1920 x 1080 at 23.976? Cant for the life of me get it to work perfectly from the get go.....I've read this entire thread plus many others

I have the exact same problem. I've read the raw recording guide and everything is set exactly as described. I even ran a card benchmark test and it runs 87 write and 110 read. Completely out of ideas on how to run 30fps. 24fps works fine without a hitch. I'm running the latest nightly build.

Open to any suggestions.

budafilms

  • Hero Member
  • *****
  • Posts: 711
  • 5D Mark a1ex
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #435 on: June 01, 2014, 08:40:21 AM »
I have the exact same problem. I've read the raw recording guide and everything is set exactly as described. I even ran a card benchmark test and it runs 87 write and 110 read. Completely out of ideas on how to run 30fps. 24fps works fine without a hitch. I'm running the latest nightly build.

Open to any suggestions.

Try card warm up.

andresharambour

  • New to the forum
  • *
  • Posts: 47
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #436 on: June 03, 2014, 05:31:42 PM »
Hey guys, I'm new to the forum and I've been testing the nightly builds with my 5D mkiii to shoot raw. I'm loving it so far.

Here's a little comparison I made with some text in Spanish that could be useful to some of you.
(Couldn't figure out how to embed the video. Too long without posting in a forum)

This was shot with the latest nightly build (05-23-2014) using the mlv format.

Any suggestions are welcome of course!

I hope this was the right thread to post this, if not please let me know and I'll remove it.

Andres

joshuamk

  • New to the forum
  • *
  • Posts: 25
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #437 on: June 11, 2014, 11:14:20 PM »
Hi All,

I've been having a read and this is a great thread, its nice to see so many camera enthusiasts working together.

I have been doing some tests of my own of the 14bit RAW on the 5d3 and have come across an issue that puzzles me a little.
I started by shooting some basic videos around the office and wanted to see how much I could pull back in terms of highlight and shadow detail
and of course it is very impressive. However when filming the window I noticed that the more I claw back in the adobe raw after effects tools the more aliasing and flicker is introduced into the image. Of course being that this is the 5d3 I know i don't get this kind of aliasing when recording h.264.

Here is an example of what I'm looking at, specifically the crossbar on the window.
This video was recorded at 1600x1200 (I meant to record in 4:3 but I get the same effects in 16:9)
Also in after effects the highlights are at -100 and shadows at +100

Has anybody else experienced this or is this unusual?

Monsanto

  • New to the forum
  • *
  • Posts: 10
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #438 on: June 11, 2014, 11:22:01 PM »
Yeah you should never go past 50 on any settings. :-[ You need to expose better remember your ISO is baked in so by pulling up the exposure you will just introduce more noise and banding your stressing the DNG this is video not photo's. Watch those highlights too, otherwise you lose color detail. Just because its Raw doesnt mean you can work miracles in post. WYSIWG turn on zebras and use the waveform when you shoot, expose correctly and a little work in post will shine. Dont be afraid to use high ISO'S its better than pulling up the exposure in post.

joshuamk

  • New to the forum
  • *
  • Posts: 25
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #439 on: June 11, 2014, 11:35:56 PM »
Its okay, I'm perfectly aware of how to expose correctly ;) This is an exercise to deliberately see how far the dng's can be pushed in post. I find the best way to get a good understanding of possibilities of the whole workflow and find interesting optical effects are to conduct stress tests like this.

Noise and banding I expected to see however I just was a bit taken aback by how bad the aliasing became. Even at 50 its quite noticeable in this particular circumstance.

Thanks for your response.

stevefal

  • Contributor
  • Hero Member
  • *****
  • Posts: 528
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #440 on: June 12, 2014, 02:55:17 AM »
I've done pretty exhaustive testing with custom test images to try and get to the bottom of ACR's non-linear behaviors relative to highlight recovery and other exposure controls. These are the behaviors that causes flickering in video when the balance of bright and dark objects changes from frame to frame. I measured Process 2012, 2010, and 2003. Process 2012 offers the most aggressive controls but all of them will cause flickering due to some kind of image-dependent tweaking of curves or maybe microcontrast processing.

I've never seen the kind of aliasing you got on and below that horizontal window frame, but it might be an artifact of the exposure processing at extremes, and the geometry and lighting of that particular object. I don't see similar aliasing anywhere else in the clip.

Focusing only on Process 2012, the following exposure controls will cause non-linear changes and hence flickering:

Exposure +/- OK
Contrast +/- NOT OK
Highlights +/- NOT OK
Shadows +/- NOT OK
Whites + NOT OK (minor)
Whites - OK
Blacks +/-  OK
Clarity +  NOT OK
Clarity - OK

Processes 2010 and 2003 behave similarly, but with less dramatic flickering (due to less aggressive effects), and some differences from the latest. I can post those results if anyone is interested.
Steve Falcon

joshuamk

  • New to the forum
  • *
  • Posts: 25
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #441 on: June 12, 2014, 04:55:10 AM »
Hi stevefal,

Yes that is interesting. This particular window frame has an incredibly straight edge and seems to invite aliasing and therefore flickering after post processing in ACR. Of course these effects are very much less if you minimise the reliance on post-correction as monsanto points out.

I suppose my main curiosity is why this kind of artifact is so intense. As I stress the image I start to see the horizontal lines emerge to give off a similar effect to what you might find on dual iso images or on de-interlaced footage.

Just my two cents.

stevefal

  • Contributor
  • Hero Member
  • *****
  • Posts: 528
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #442 on: June 12, 2014, 07:33:22 AM »
I guess I'm seeing something else. I don't think it looks like dual-ISO aliased edges. I  see a kind of rolling moire along the upper face of the frame. Is there perhaps a line or grid pattern on that metal material?

As I look again, I don't really see how this could be related to the exposure issue. The artifact moves just like moire, but I don't see the pattern causing it. It's like to shooting though a screen.
Steve Falcon

a1ex

  • Administrator
  • Hero Member
  • *****
  • Posts: 12390
  • Quarantined
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #443 on: June 12, 2014, 08:56:15 AM »
Can you upload a few DNG frames?

joshuamk

  • New to the forum
  • *
  • Posts: 25
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #444 on: June 12, 2014, 12:43:40 PM »
Hi all,

there is no pattern on the window frame so yes a bit puzzling. Here are a couple of frames in my dropbox.

https://www.dropbox.com/s/jbhgi3euj0drfhb/M11-1956_00000.dng
https://www.dropbox.com/s/m7q647pbvfiwawk/M11-1956_00021.dng
https://www.dropbox.com/s/qmgeq1aqdoj4t2m/M11-1956_00054.dng

I'm also going to upload a similar davinci grade later today to compare with ACR.

Regards

mannfilm

  • Freshman
  • **
  • Posts: 88
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #445 on: June 12, 2014, 06:26:27 PM »
In Adobe CC, ACR is different in photoshop or after effects. There is flicker and artifacts in Photoshop ACR because it is treating each image as a separate image and changing  from frame to frame. There is no flicker or artifacts in After Effects ACR because it is treating it as video,  and applying the same changes to every frame - like a video editing app should do. 

joshuamk

  • New to the forum
  • *
  • Posts: 25
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #446 on: June 12, 2014, 06:33:09 PM »
Although I am using after effects :)

stevefal

  • Contributor
  • Hero Member
  • *****
  • Posts: 528
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #447 on: June 12, 2014, 08:49:45 PM »
There is no flicker or artifacts in After Effects ACR because it is treating it as video...

This is not right. AE/ACR CC still flickers: http://www.magiclantern.fm/forum/index.php?topic=5710.100
Steve Falcon

joshuamk

  • New to the forum
  • *
  • Posts: 25
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #448 on: June 13, 2014, 01:09:42 PM »
Here's an update on the clip I posted yesterday. It's good to get  to the bottom of this.

NEW CLIP:

ACR is absolutely the problem. I performed a simple correction in Davinci and get none of the same problems. I was a little shocked at how drastic the ACR settings actually are. To my eyes the output from Davinci is much milder even when trying to do a stress test. Overall I would still obviously like to have After effects as an option for post production but after this side-by-side I doubt I will touch it for RAW video again.


stevefal

  • Contributor
  • Hero Member
  • *****
  • Posts: 528
Re: Uncompressed 14-bit RAW video testing - 5D Mark III
« Reply #449 on: June 13, 2014, 02:52:39 PM »
I still use ACR for the first technical correction, because it has so many high quality features. I just keep away from too much of the fancy exposure controls. I use it for:

- white balance
- exposure
- some highlight/shadow
- lens correction
- noise reduction
- sharpening
- chromatic aberration correction (for specular highlights)
- gradient corrections

The last one is possibly overlooked by people. With ACR you can draw gradients over the image that contain exposure and other corrective variations from the main controls. For instance, if you have a hot light on the side of an image, you can bring its exposure down while leaving the other side of the image untouched. For me, this is a good alternative to aggressive highlight control. Check it out.

I then render in AE to ProRes4444 at 16 bits. These files are much smaller than the CDNG, but still very gradeable.
Steve Falcon