Back In Business (new image from sensor)

[Jarred Land]

Well the recent events didnt stop Red.. In fact it didn't even slow them down. It actually gave Jim some time to create an incredible new color matrix, and the results are pretty stunning.

Here is a chart straight off the camera this morning with only Jim's matrix applied.. no levels color correction etc.

There is a digital Macbeth integrated to the JPG so you can compare.

http://www.dvxuser6.com/uploaded/2/1159385660.jpg

[Brook Willard]

Ooo...

[Kinematic]

My heart just stopped............
.
.
.
.
.
.
.
.
Oh! There it goes again. God bless those pacemakers!

Great work Jim and thanks Jarred for posting this!

[insanityfw]

Are you kidding me? I'm no pro, but my eyes don't lie... and that looks pretty on-the-money to me.

The hits just keep on coming. Thanks you guys.

[Homersapien]

Color looks great, I also like the DOF on the bg and the metal stand, seems to me to tell us alot more than some of the stills of the car/women etc.

Keep em coming, looking sweet!

[esmilis]

this is awesome!

[Poi Boy]

Great; some level adjustments and it is stuningly good to go !
Aloha
-A

[craigbowman]

LOL! You know you've got the bug when you get palipitations from a colour chart comparison.

[scienceguy_ae]

Well the recent events didnt stop Red.. In fact it didn't even slow them down. It actually gave Jim some time to create an incredible new color matrix, and the results are pretty stunning.

Here is a chart straight off the camera this morning with only Jim's matrix applied.. no levels color correction etc.

There is a digital Macbeth integrated to the JPG so you can compare.

http://www.dvxuser6.com/uploaded/2/1159385660.jpg

Ok, that's incredible. The only thing that I can tell is that it's representing all the colors a tiny, tiny bit warmer–which is actually the better direction to go, IMHO.

[ericyoung]

:thumbup: Cor Blimey Guv'nor. That's a beauty and no mistake!

Wow. So nice it made me go pseudo-Cockney, and I ain't no Cockney!

(BTW, what is it with the stupid bowties and tweed jackets that us Brits are supposed to wear in all American films with a UK setting, not to mention the red buses and telephone boxes which are in every shot! It gets a bit annoying).

[imgentertainment@mac]

damn Jim that is really close. Good work.. You do know about color there is no doubt about that.

[Jarred Land]

Its amazing watching Jim tweak this stuff... he really does have a handle on colorimetry.

[Anders Holck]

Here is a composite of the charts, for easier comparison. The synthetic targets are the small superimposed squares.
Remeber that both charts must be viewed in a color aware application (like photoshop) as they are in the Adobe RGB 1998 colorspace.

Composited chart (http://www.andersholck.com/chart.jpg)

It's looking very good already!

[andersh]

When you have both the input and output values, it seems to me to be just a matter of solving the equations rather than tweaking?

Edit: Just tried to solve for the equation [r, g, b, rr, gg, bb, rg, gb, br, rgb, rrr, ggg, bbb, rrg, rrb, ggb, ggr, bbr, bbg, 1]*A=[RGB]' (it was what I could come up with, but maybe the log-colorspace should be used). That got the (min, max, mean, median) distance between the 24 colors down from (0.0077, 0.1419, 0.0462, 0.0466) to (0.0008, 0.0194, 0.0087, 0.0089), which is nice, but also introduced some ugly artifacts in the yellow, lime and light orange.

[Haakon]

Its amazing watching Jim tweak this stuff...
Must be nice to be able to just hang around the RED offices all day... :-P

Looks amazing, Jim. The news from your camp just gets better and better.

[Graeme_Nattress]

Thanks for that, Anders. We've put some tweaks into the REDCINE and should be getting much better matches pretty soon. The other thing, is that there's a contrast curve on the image Jim made, which may account for the level, rather than the chroma differences you're seeing, in greyscale tracking for instance.

Graeme

[Jarred Land]

Here is a composite of the charts, for easier comparison. The synthetic targets are the small superimposed squares.
Remeber that both charts must be viewed in a color aware application (like photoshop) as they are in the Adobe RGB 1998 colorspace.

Composited chart (http://www.andersholck.com/chart.jpg)

It's looking very good already!

thanks for the composition.. nice to see the overlay.

just to let you know, since posting this image, Graeme has made some revolutionary tweaks that has made the two even closer..

[mcgeedigital]

That is just sick!

[Jannard]

Truth be known, I would be handcuffed without Graeme. He is providing tools that make this a lot easier. Certainly is a lot of fun, though.

Jim

[WarrenS]

... since posting this image, Graeme has made some revolutionary tweaks that has made the two even closer..


Sounds intriguing--just what is a "revolutionary tweak"?:thumbup:

[stevesherrick]

Jim, you got yourself a good team. Graeme has been doing excellent work with his various plugins, which I own and love. But most importantly, he has always been about helping people out and giving great support to his products. He's always just an email away. The fact that he is on the Red team is very encouraging.

Steve

[evinsky]

http://apog.nolablogs.org/archives/images/drool.gif
(Drooling)

[acrochordon]

I don't know much but I am guessing the most important things are the 18% grey and the skin tone, and they look perfect to me. (I am looking at them on my 23" Cinedisplay, which might not be a perfect screen, but I think it is good enough.)

You guys are going to take it out into natural sunlight too, right?
I want to see the required shots of Mannequin eyelashes when it comes time to test the zoom lens. :)

[andersh]

It would be nice to have a natural looking scene shot with the same settings to test.

[agwah]

This is amazing, have you done tests at higher iso ratings as well

[Terence Lehane]

1st point:
It says the picture has been colour corrected.

Colour correcting a Macbeth chart can't be that difficult. So what's all the fuss about?

2nd point:

Cor Blimey Guv'nor. That's a beauty and no mistake!


My mother's a londoner so that's how I recognized your accent.

[Graeme_Nattress]

Colour correcting a shot macbeth to look like it's supposed to look is actually quite tricky. There's a lot of math in there, and a lot of art. We're trying math+art to get it to look good, and be accurate.

Graeme

[im.thatoneguy]

Perhaps part of the RedCine toolset should be a software app which lets you load up a Macbeth tiff and a shot from on set, and then just draw square A draw Square B (A shot, B Tiff) Then let it get you close, and from there offer some refinement tools. It seems like the math would be a relatively simple plot interpretation app.

I actually don't know if this would be useful at all, I'm not very well versed on LUTs.

[Graeme_Nattress]

It's not a LUT thing, it's a colour matrix thing, which is based upon the colorimetry of the sensor dyes on the colour filter array, and the target RGB space you're aiming for. This will be calibrated in two ways:

1) each sensor is calibrated in the factory to match a standard sensor spec
2) the standard spec is calibrated to the target RGB space.

Once this is done, all you need to adjust is white balance. We do all the hard work, so you don't have to.

Graeme

[Brook Willard]

Well, from a technical perspective, the use of a MB chart shouldn't actually be necessary from shoot to shoot. Lets assume that a given sensor in a given camera has been tested and locked down with a particular custom color profile to read the MB chart accurately in a controlled situation.

At that point, you have a camera that will output what it sees properly. Reds won't be too red, greens won't be too green, etc. Then all one should have to do from shoot to shoot is read a basic 18% grey card. Technically, if you have grey locked off at grey in your particular environment and the sensor is locked off to read colors correctly according to its particular color profile, you should be set.

This could happen in camera, I suppose. I'd love to be able to put my 18% grey card up on set, point the camera at it in my selected neutral color temperature and click the "this should read 18% grey" button. The camera adjusts itself to that grey in that situation [keeping in mind its own color profile] and I'm ready to rock and roll.

Ok, basically I'm talking about white balance but with a grey card. Ah, habits from telecine.

[edit: Ok, I wrote this while Graeme was writing his reply... and he already answered the question I was asking myself]

[Jarred Land]

I'd love to be able to put my 18% grey card up on set, point the camera at it in my selected neutral color temperature and click the "this should read 18% grey" button. The camera adjusts itself to that grey in that situation [keeping in mind its own color profile] and I'm ready to rock and roll.

Ok, basically I'm talking about white balance but with a grey card. Ah, habits from telecine.

[edit: Ok, I wrote this while Graeme was writing his reply... and he already answered the question I was asking myself]

this is actually a good idea brook.. because it can also tell you if your exposure is correct as well, not only your color balance.

[Brook Willard]

That would work in *some* cases, but not all. Here's my example:

Lets say I've lit my scene for tungsten. I've used four different gel colors from four different directions, the set looks ridiculous. I put my grey card up and what am I going to see? Four colors on one card [and it *won't* be tungsten]. What I do in a situation like this is kill the gelled lights and hit my grey card with a tungsten 1K [slate light].

Since I've lit the scene with tungsten as neutral, the grey card will read as neutral with a single neutral light. I'll expose the card properly to keep it in the middle of the dynamic range. Then I pull the slate light, turn my lights back on and enjoy the [non-neutral/tungsten] colors that they bring to the scene. The final exposure will almost certainly be different.

So in *that* situation, it won't work for exposure. It'll work for color, but that's just for the telecine operator. The whole setup will only work because I'm lighting the scene with knowledge of what will read as white/neutral and what will read as a varying temperature.

Lets say that I don't use a slate light [in a situation with a more neutral color scheme]. Lets say it's a day exterior without any lighting of any kind. The grey card will work for both exposure and color but *only* if the card is held in the same place as the subjects. On a wider lens, I'll have to have the card close to the lens to fill the whole frame [it only really works if the frame is full on the card]. If the card has to be in a different plane than the subject, the exposure may be different.

Anyway, this post has gone on too long. It could work sometimes, but it certainly won't work in all cases.

It'll always work for color if you do it properly.

[Jarred Land]

yes of course.. if you test with different level/color of lights your not going to get the correct exposure with that test... but your really not going to get an accurate color balance either using that method... as if you light your card with a different light than your set, your not taking into consideration color bounce, stray shifts or lamp inacuracies. A much easier way would just to use a color tempurature preset... set it to 3200K (or whatever your source is) and shoot.

As for the last comment about needing to hold the grey card close to the subjects, you do that anyways in the "traditional" method of exposing with a light meter. for wider lenses, you would need to employ a focused reading, much like using a spot meter.

I get what your doing, but it just seems like a hellofalot of work to do something that should be rather simple using color temperature presets.If you wanted to be more accurate, you could use a color meter to depict the correct kelvin of your source.

[Brook Willard]

I should note that my post was referring to a film shoot. No grey card/color chart = bad juju. I completely agree that using presets should do the trick most of the time for digital.

I was also referring to a rather complicated lighting setup in the first instance.

[Jarred Land]

ah there you go.. of course shooting film would make that more appropriate, i will completely agree with that.

but then again... you could use a specific temperature balanced stock. :)

[Brook Willard]

I always use temperature-balanced stock. Regardless, there's no excuse not to shoot a grey card at the head of each roll. Potential variances in the celluloid, chemical makeup, development, etc.

If I was on a big show and used the same colorist the whole time, I'd trust them to match it roll to roll. Unfortunately, most shows I work on are small, potentially use multiple colorists and - unfortunately - sometimes I don't get to supervise telecine. The card saves me from all potential color disasters.

But for some reason I'm talking about film in the RED forum.:)

[im.thatoneguy]

It's not a LUT thing, it's a colour matrix thing, which is based upon the colorimetry of the sensor dyes on the colour filter array, and the target RGB space you're aiming for. This will be calibrated in two ways:

1) each sensor is calibrated in the factory to match a standard sensor spec
2) the standard spec is calibrated to the target RGB space.

Once this is done, all you need to adjust is white balance. We do all the hard work, so you don't have to.

Graeme

Sorry I wasn't very clear in where I was going with the question.

My question mostly relates to matching cameras. With film or a locked white balance in video it's easy. You shoot one way, then you can grade all the color from each of the cameras pretty much the same since you're starting from the same color settings.

If you white balance two cameras differently, trying to reverse engineer something as simple as one having white balanced to a cooler lit card can be a nightmare.

Would it be possible to export say an ASCII definition of the WhiteBalance and an inverse color function to calibrate two seperate cameras.

Let's say Camera A applys a rudimentary color equation to the Raw sensor data: R^2 but Camera B is using 2R. If a program was fed both pieces of footage and told to match B to A it would know what the difference between the two equations was and properly match the footage via a new LUT.

Once the footage gets into a grading application, often the gray card alone isn't enough information to mathematically match the two balances, since a camera's white balance isn't a linear RGB scaler. (warmer/cooler)

I guess what I'm asking for is the ability to white balance in post. Would this be possible or am I just way the hell off base. (Easily possible.)

All of this is mostly "outdated" since the 10bit color space of RAW will allow me to simply shoot a default setting such as 5600k and then grade somewhat universally later without too much color loss. But it'd still be nice if I need to shoot in a heavier compressed 8bit format where I want to get as close as humanely possible in camera before it gets burned to ones and zeros, but isn't possible to perfectly synchronize on white balance.

[Graeme_Nattress]

You'd just put the same white balance number into both cameras - say 5600k. Then they'd be practically identical, or, you shoot raw, and type the "right" number into both clips in the REDCINE.

Graeme

[Haakon]

You'd just put the same white balance number into both cameras - say 5600k. Then they'd be practically identical, or, you shoot raw, and type the "right" number into both clips in the REDCINE.
Graeme,

Does this mean that we will be able to input whatever white balance temperature we want, numerically, as opposed to being limited to selecting from presets or having to "set" the white balance optically? If so, that's awesome. What kind of range will the camera have in terms of minimum and maximum white balance values?

I know part of the point of shooting RAW is the ability to adjust white balance in post - while I still don't completely understand the concept yet, are you able to shift it as much as you like in any direction without any kind of image degredation at all? Will the REDCINE software allow us to carry out these functions? Thanks in advance.

[Anders Holck]

Hopefully we'll at least have something like a fast 3200k/5600k/Custom setting, and a variable kelvin/tint for the custom setting.

This setting is used as the live viewing setting and as the starting point in REDCINE.

If you have never tried RAW processing, have some fun with Adobe Lightroom. It's a free download, and the 120 MB download includes a few RAW images to play with. http://www.adobe.com/cfusion/entitlement/index.cfm?e=labs%5Flightroom
Note that Lightroom has been in active development over a few years and not <8 months. Exactly what controls we'll have available inside REDCINE is not known, but I don't expect it to be as extensive as this....

[Stuart English]

Anders, you are exactly correct in what you suggest as the workflow. The EVF or LCD drive signals are white balanced to permit on set viewing, and that value is attached as metadata to the REDCODE RAW data stream.

In REDCINE, the white balance value can be used as your starting point for whte balancing the footage. The great thing about RAW though, is that your white balance in a shot isn't quite right, or if it is varying shot to shot, just change it.

[Haakon]

The great thing about RAW though, is that your white balance in a shot isn't quite right, or if it is varying shot to shot, just change it.

I get that part of the concept, but I'm curious as to how far you can push it:


are you able to shift it as much as you like in any direction without any kind of image degredation at all?

[Graeme_Nattress]

Current plan is named presets + user number for WB colour temperature setting. As for range and precision, that's TBD.

You can't shift totally in post without degredation, but you can shift an awful lot without any nasties. Best way to see for yourself is to shoot RAW DSLR stills and process it Lightroom or Aperture.

[Haakon]

Graeme,

Thanks. I'm a PC guy so no Aperture for me, but I'm checking out Lightroom as we speak. I've got a great dSLR, so I'll do some playing around this weekend and see if I can educate myself a little more on the RAW front.

[Graeme_Nattress]

Superb - I think you'll love it. I do. I shoot everything RAW.

[im.thatoneguy]

*Smacks forehead* Oh yeah I had completely forgotten RAW doesn't apply the white balance until you "develop". Stupid Stupid Stupid.

I know you're probably still trying to work this out, but once we've chosen our preferred whitebalance how do we import this into an application such as Shake, Nuke or Fusion? Will we be able to import it just as any other codec or will we have to process the footage in REDCINE to an uncompressed Cineon style file format? It would be nice to be able to just apply all of the RedCine modifications to the RAW data using a custom LUT created by REDCINE which is equivalent to the processing without acutally modifying the data necessarily yet.

I like the idea of the low bandwidth compressed RAW, but don't want to lose a lot of quality if I transcode to a useful comprssed format.

[avocade]

Considering how long it takes to update the white balance on a single RAW photo in Aperture on my laptop, you guys must do some serious performance-tweaking to get the RAW-stream to play in realtime. Especially considering that all dSLR's I've used to date shoot less than the 11 Mpixels you get out of this bad boy.

Or will the "online" RAW editing (hopefully in FCPro) only be possible on a Octo-Core Mac Pro? The timing for when those will probably be out seems to coincide nicely with the RED shipping date...

[Graeme_Nattress]

Well, Aperture 1.5 seems a lot faster :-)

Of course, you only need to white balance all 12mp if you can see them all on your monitor at once :-) or when you render an effect. Not much way around that, but we'll write the code and optimize it to go as fast as it can, for which we have some tricks up our sleaves.

Graeme