Messages

We have had development of this feature over the past months over here at the S3Dcentre. Once we have been able to confirm this data as a reliable workflow, we will be publishing our results on the forum with the developers of ML and on our website for the stereoscopic cinema community.

get in touch with us, patryk.

Hey Tom,

Get in touch with us.

Patryk

[The improvement we can get using your raw module in comparison with the H.264 is amazing]

Hey Magic Latern Community & Development Team:

We are two researchers (Patryk Stasieczek & Paolo Pennuti) at the Stereographic 3D Centre at Emily Carr University of Art + Design (Vancouver, Canada).

We are doing DSLR 3D tests using your ML raw module to shoot 3D raw footage with our Canon 5D III's.

The improvement we can get using your raw module in comparison with the H.264 is amazing... our understanding is that it can be even more surprising for 3D purposes (the immersive quality of stereoscopic 3D footage can really benefit from sharper and more detailed RAW footage).

We successfully recorded 3D using Magic Latern in H.264 videos using two 5D3 DSLRs synced with the Pocket Wizards mod  (http://3dfilmfactory.com/index.php?option=com_content&view=article&id=93:gen-lock-canon-5d-mark-ii-cameras-and-shoot-3d) but we experienced that it is impossible to use this trick to sync two 5D3's running on the latest ML builds when we shoot in raw. We think that because ML is putting forward an incredibly fast writing speed for RAW image-sequences, we aren't able to trigger a shutter to sync between two cameras.

Considering the lighting bolt and the motion detection modes that you added to the last versions of ML, we have many reasons to think that you can be technically capable of a software improvement in order to "genlock" two Canon 5D3 cameras. The major issue we are looking to resolve is motion synchronization in the footage in order for it to be immersive and true to reality in both right and left eyes.

If it is possible, it would be great for us to start to communicate with you in order to find a solution to this problem, as we feel that if ML embraces the current trend in S3D technologies, there will be more investment towards Magic Latern as a standard build in contemporary stereographic cinema.

Relatively many people are shooting 3D videos using Canon DSLRs, and the stereographic 3D community is ready for something to sync their DSLR cameras.

We are currently working on our comparison footage between ML RAW S3D and ML H264 S3D, and it will be posted in this forum discussion soon.

Can you help us (we would be glad to support this kind of research with our tests)?

Best,

Patryk and Paolo
HQP, Graduate Researchers at the S3D Centre at Emily Carr University of Art & Design
http:/www.s3dcentre.ca

Hey this does not offer a solution to working in Camera RAW ML footage. The Pocket wizards offer a 'seemingly' workable solution to shooting h264, but not in Camera RAW. Even in h264 there are sync issues when triggered with pocket wizards.

[This 100% clip from a .RAW sequence converted in raw2dng, and then exported as apple pro res 4444 in After Effects CS6.]

Hey, I have noticed that our the raw2dng.app inserting vertical lines into digital negatives. If i export in Apple Pro Res 4444, the vertical lines are still present, but if I convert the dng .RAW video sequence into an h.264 mp4 video file, it compresses them away.

I am not sure if this is our Mark III sensor, or an issue with compression. See Images Below.


This 100% clip from a .RAW sequence converted in raw2dng, and then exported as apple pro res 4444 in After Effects CS6.
[/b]

100% clip from .RAW sqeuence convereted in raw2dng, then exported as h.264 mpeg (.mp4) in After Effects CS6.
[/b]



Any information would be great, thanks.

My colleague and I are doing some testing on Canon Mark III running on Magic Lantern.

For 3 of our exposure testing shots, we encountered some issues with the .RAW file not being recognized by raw2dng.

Recording Methods:
File 1) We let ML record RAW until there was no memory left in the card, live view mode disabled.
File 2) We recorded for around 2 minutes, live view mode disabled.
File 3) We recorded for just over 1 minute, live view mode disabled.

File Size:
File 1) is 4.29 GB on disk (4,294,967,295 bytes)
File 2) is 4.29 GB on disk (4,294,967,295 bytes)
File 3) is 4.29 GB on disk (4,294,967,295 bytes)

It seems strange that the files are all the same size, when we decreased the number of frames with a shorter recording time.

When we drag these .RAW files into raw2dng, the .app creates the folder, but then does not extract the .dng sequence.



We are shooting on:

Canon EOS 5D Mark III
Lexar 16gb 1000x CF

I have searched the forum for the file-sizes and haven't found any information.
Would it be also an issue that the live view mode is disabled? Would disabling this feature during the recording process somehow corrupted the .RAW file data somehow? I am unsure.

Any help you can send in our direction would be amazing.
We are going to be testing out some 32gb cards in the coming week because our write speed is giving us issues on the Lexar 16gb 1000x.

Thanks.