Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

What is the real deal

C

Colin Hubick

Well-known member
Joined
Mar 23, 2007
Messages
629
Reaction score
0
Points
16
Location
Canada
I found this posting on creative cow talking about quality of debayer method, ie 1/2 , 1/2 high, full.

In this email from Graeme, he states the following.

Hi Jack,

"Well, I'm as qualified as anyone on how the RED / Color integration works.....

There are two modes of extraction of 2k from 4k used in Color. Both modes cut through Quicktime and access the R3D raw data direct.

2k Quick, is used for fast preview - it access the full 4k R3D data, but rapidly extracts 2k RGB from it. It's full RGB with equal resolution in all three channels. Only downside to this is you may, in some circumstances, see mild aliasing artifacts.

2k Nice is used for rendering (you can also render in 2K Quick by adjusting the render proxy settings) - Access the full 4K R3D and demosaics and downsamples to 2K in one go with a specially designed algorithm for absolutely superb sharpness and minimal aliasing. It's a very nice downsample from the full 4k and again, has full resolution in R, G, B.

In neither mode is their reduced colour resolution. No image components will come out soft.

As you know I wrote both the algorithms for getting 2k from 4k R3D, so I know exactly what I'm talking about.

Graeme"

original thread on creative cow

http://forums.creativecow.net/readpost/223/9895

Then there is this posting from Jim, which states something differently. As this was a point of disagreement with all involved, I would like to know the straight goods on this matter.

From Jim

"2. They shot 4K footage and did a half-res debayer to 1080P (half-res throws away 78% of the recorded information) and did not "un-sharp mask" the RAW footage because _________ (insert another myth here)."

..If half-res throws away 78% of the recorded information, how much is thrown away in half-high?

original thread on Reduser

http://www.reduser.net/forum/showthread.php?t=40067&page=2
 
Well, one factor that I can think of is that there is a difference between:

1) Doing a half-res debayer of 4K HD material to 1080p (which is mathematically easy and the result looks pretty good to me)
and
2) Doing a half-res debayer of 4K to 1080p, which is is technically 4K half-res to get you 2K and then downscaled to 1080p, and I don't think that looks quite as nice coming from some software programs, and requires more processing.
 
This was for a film-out, and the argument was primarily between a full debayer, and half-high debayer.
 
That Creative Cow is also dated April 2009. Since that time, there has been a color-science update, as well as an update to FCS that allows 4K in Color.

To go with the original question "How much is thrown away on Half-High?" I'm going to hazard a guess at "The same." Now, I'm not Graeme, but it seems that your two statements aren't mutually exclusive. In fact, they're talking about the same thing. Graeme's talking about how it gets to Half-Res (AKA What is done with the information *before* it's thrown away), whereas Jim is talking about the end result, which is that you have, in fact, thrown a lot of pixels away.
 
Before Color 1.5 the max working output rez of Color was 2K. (I seem to remember the difference we had between the 2K and 1080 license of FinalTouch, which was what Color was before Apple bought it was around $15K! Who knows what they would have charged for 4K).

I don't think Graham was actually saying that in the first Color plugin they were only doing a half-rez debayer when rendering? Rather he says that they debayer and scale to 2K in one process, which makes a lot of sense. I have often extracted outside of Color first as the plugin didn't have a DRX slider which is useful on clipped shots, and the Color secondaries key differentiation was messed up with Raw. But doing a "best light" first before stepping away from the raw.

Jim is if course correct that a full debayer is always better. The difference by the time you get to 1080 (our most regular deliverable) for me is apparent in shadow noise.

But there are still other ways you can mess up the picture after that! I'd rather have a half-high debayer decently scaled to 1080 than a full debayer badly scaled. I'm not sure if the Red plugin or Color handled the further scaling to 1080, but I always got much better results going to 2K in Color and then going 1080 outside of Color.

Since Color 1.5, I've done some comparisons of going to 4K and 2K and then scaling to HD and cannot see significant difference between the two when both are decently scaled to HD. Not that Color still doesn't have other issues but that's not Red's fault!

Best

James
 
What you're reading is two perceptions of the same thing. If you start with 4k, and go down to 2k (by whatever method at all) you're throwing away a goodly percentage of the recorded image. That is what Jim is talking about.

What I'm talking about is given that some people will want to go to 2k from 4k, what are the pros and cons of the various methods available - the best, to me, being a full demosaic to 4k then a traditional high quality downsample.

Graeme
 
Graeme Nattress said:
What you're reading is two perceptions of the same thing. If you start with 4k, and go down to 2k (by whatever method at all) you're throwing away a goodly percentage of the recorded image. That is what Jim is talking about.

What I'm talking about is given that some people will want to go to 2k from 4k, what are the pros and cons of the various methods available - the best, to me, being a full demosaic to 4k then a traditional high quality downsample.

Graeme
Click to expand...

Graeme,

I think that one of the least-talked-about questions with Red footage is this: What is the best path to take for downsizing from 4K Red footage to 1080 or even SD. Most of my work ends up on TV, so this is a big concern for me. What do you believe to be the best way to downsize (software, settings, whatever) to avoid moire artifacts and aliasing, and get the best image out of the original 4K?
 
Every app that does re-sizing has different filters. The RED apps have always had a good set of filter choices, as does Shake. Beyond that.... Be careful and test.

Graeme
 
nope...hd to sd.
before that comes your magic stuff.

anyway, makes great SD footage of RED, the best i did get so far

of course i dont own one, but a friend has two in his living room :)
 
Graeme Nattress said:
What you're reading is two perceptions of the same thing. If you start with 4k, and go down to 2k (by whatever method at all) you're throwing away a goodly percentage of the recorded image. That is what Jim is talking about.

What I'm talking about is given that some people will want to go to 2k from 4k, what are the pros and cons of the various methods available - the best, to me, being a full demosaic to 4k then a traditional high quality downsample.

Graeme
Click to expand...

Thanks Graeme, that answers the question for me. The primary debate was about if the quality you got from doing a half-res high debayer was equal to doing a full debayer. Easy question, but it was a debate.

Colin
 
You must log in or register to reply here.
Back
Top