canon 5d conversion workflow

[Lior Molcho]

Hi

can anyone outline in simple steps a successful workflow for converting 5d files to cineform on a pc?

( I seem to find a lot of math talk and few workflows ) also, please mention build or Beta version of cineform being used

Thanks!

Lior

[David Newman]

After you have copied your 5D source from the flash media to your hard drive, follow these steps.

1) Launch HDlink that comes with all NeoXX products.
2) Set your preferences to progressive and quality to High. (only need to do this once.)
3) Select the convert tab.
4) Click "Select Files..." or "Select Folder..." to import your camera media (5D or almost anything.)
5) Click "Start" and after a short time you're done.

The new AVIs will be in the same folder as you source; if you want to go elsewhere use the Capture Location option in preferences.

All pretty simple.

[Lior Molcho]

Hi David
thanks for the quick reply!

will that also solve the Gamma differences? when playing the rendered files to cineform , they appear darker and the blacks are "crushed"

Thanks
Lior

[David Newman]

Should be (and is) the opposite, we uncrush the image, restoring to standard YUV ranges and fix the colorspace form 601 to 709 -- this happens automatically and it is a good thing. However there is a bug in Vegas 10 that is display to the black at the wrong level, so you may be seeing that. Or you maybe using Quicktime Player what has the lovely feature of playing with the wrong gamma. The converted files are correct, so if you seeing a clipping of black or white images element, have a look at how the image it being displayed.

In the HDLink log window you should see:

10-bit correction for Canon 5D and 7D Luma range and color space
Upconverting 4:2:0 source to 4:2:2

[Mike McCarthy]

I switched to AME as soon as Adobe fixed the DSLR importer in CS5, because back then HDLink was still converting files in the Quicktime decoded colorspace. Obviously any app that uses QT to decode the DSLR files (Like CS4) gets dramatically incorrect result. How does your processing compare with the way CS5 decodes the files? Will the outputs from converting DSLR to Cineform in AME CS5 be the same as using HDLink? If not, will they be close?

[David Newman]

Mike,

We have never used QuickTime to decode Canon DSLR files, we originally recommended tools like CoreAVC, then the behavior depended on how you had that set up. For nearly a year we have had our own engine, that fixes all the issues, well before CS5 was even out. We have continued to enhance the DSLR pipeline. CS5 does get the color, gamma and contrast correct, just as we do, yet its 4:2:0 to 4:2:2 or 4:4:4 up-conversion is not as good and there are more RGB transform involved. Also I imagine their processing path is not as paranoid as ours.

HDLink does the following:

Source -> decode to YV12 (native 4:2:0) -> up convert to 4:2:2 10-bit with range correction to 64 to 940 and colorspace convert to 709.

I expect CS5 is going a 601 full range 0-255 range YUV decode to RGB32 (8-bit), this is where the 4:2:0 to 4:4:4 happens. It is 4:4:4 data that is sent to AME before is likely encoded back to 4:2:2 YUV 709.

The CineForm approach does it best to maintain a YUV colorspace.

I'll post images that show the quality difference when I get more time.

[Lior Molcho]

Works like a charm!
thanks David, apparently i was using an older version of cineform HD link

[David Newman]

Mike & Lior,

Instead of posting images here, I did a blog entry on the subject.
CineForm Insider: Why use an intermediate for DSLR video?

David