OK - the RCX vs Color vs Redline/Clipfinder debayer noise-test when preparing images

[Gunleik Groven]

Yup. That's exactly the point of these tests.
Preparing the ground for online of a feature + two shorts and looking a bit into the glassball for a series that won't enter post before a year, but will be shot now on M.

And the tests are not finnished.

But these initial tests have given me some new info I wasn't prepared for.
Among others that Color starts to hold up to RCX out of the RED-room.

I am still not totally sure that the signal chain in Color from RAW holds up against DPXs. Before I did this test, I would have written it off as an option.

Now I will have to test.

Whiteballancing have been a problem.
Now it has less bad affect, and possibly more good than bad results - as long as it is done in post. But B30 seems to have solved a lot of the on-set pitfalls...

Mythbusting.
For myself

And others who want to have a look.

I still think Redline looks amazingly good, though as a startingpoint for a grade...

And I am glad if I can leave RCX out of the equation for the time being, as there are no good way to conform to it today.
(And three of the projects are current and with deadlines and have to relate to today)

RedRocket support in Color, with RCX as the RedRoom feeding Color a clean 12 bit RGB image that could be reconfiguerd from RAW in realtime would have been.... schweeet...

Need to test more...

 

[Gunleik Groven]

Small update:
I have got confirmation that I can post RAW samples of what we do.

The next batch of tests will be done from thursday this week to wednesday next week.

Will post material then.

Tests to be done:
Charts... (yeah, I know how fun that is...)
Taking the RAW signal through the Color Rooms vs taking DPXs through the Color rooms with and without display-LUTs applied.

Please come with suggestions for tests, and I'll look into them.

Cheers

Gunleik

 

[Nook Kim]

Very assuring test findings for Color users like me. Thanks for sharing your findings, Gunleik. Now, I'd love to see what we can expect from the same test images through Scratch and SpeedGrade. I know they're in their own league price-wise, but I am just curious to compare for my future upgrade.

Cheers to your next test!

 

[Gunleik Groven]

I will be posting r3D's alongside my next test, but a firmware bug has made me decide to look carefully through which I can post and which will have to be re-shot.

Then people with Smoke, Baselight, EQ, Speedgrade and Scratch can share their findings and post to the same threads with the same settings from the same RAW files.

Could be interesting for comparisons.

Thanks!

Seems to me that there's a couple I can post, like the key-test.

Thanks!

 

[Justin McAleece]

Thanks to users like you!

Thanks to users like you!

you're a champion Gunleik. Many of us would like to do these kinds of tests but lack the time. Your'e a patient and generous guy for providing us all this info.

 

[Gunleik Groven]

Thanks.

I'll add rocket-renders tomorrow + some alternative ways of developing the files to get less noise, but I can't post the raw files with this bunch of tests, as they are to huge and would kill the server...

 

[Hans von Sonntag]

Then people with Smoke, Baselight, EQ, Speedgrade and Scratch can share their findings and post to the same threads with the same settings from the same RAW files.

Could be interesting for comparisons.

Thanks!

Seems to me that there's a couple I can post, like the key-test.

Thanks!

Hi Gunleik,

thanks for your generous tests. I know how much time such tests eat and hence I'm very grateful that you took your time, and knowledge of course.

To add a few points:

- Problems with scaling algorithms can best bee seen when the camera is pointed to a checker-board.

- Moreover, scaling to an equal common denominator (scaling 4K 16:9 to 2K 16:9 or 4KHD to 1080p) won't show the amenities of one algorithm over another. I suggest to scale 4K2:1 to 1080p.

Sometimes a softer scaling algorithm is more meaningful than a sharper one, eg. scaling for interlaced PAL broadcasting. Hence, a general judgement can't be done here.

- Smoke does not read R3Ds natively like Scratch, SpeedGrade, Premier Pro etc.. do. With Smoke you use a separate application that renders R3Ds to DPX into Smoke's assigned directory (hard import) using Red's SDK. RedLine does not need to be installed. I don't know which SDK build is used.

- SpeedGrade reads the R3Ds natively through Red's SDK. The newest SDK is now integrated in beta versions of SpeedGrade 2010. A service pack integrating the newest SDK (Build 30) for the current SpeedGrade 2009 version will be released soon.

If you want to now how these two above mentioned applications handle the SDK and scale footage PM me, I have access to both of them.

Hans

 

[Gunleik Groven]

Thanks Hans!

To see scaling tests, go to the B30 thread:
http://reduser.net/forum/showthread.php?t=42476

I've posted the r3d alongside the zone-chart for others to find other results.

And:
Yes. these tests take time and Norwegian Broadcasting Corp have been very generous to both let me do these tests and let me post the results and the RAW files alongside them.

Gunleik