Nice analysis! Good geeking, Art!

-mike

Posted by Mike Curtis  on  09/10  at  07:16 PM

---

Thanks for your hard work in exploring RED. The truth will out!

Posted by Rob  on  09/10  at  09:19 PM

---

Thanks, gentleman. Just trying to make the camera a little less of a mystery. It’s been a fascinating journey.

Posted by Art Adams  on  09/10  at  11:21 PM

---

This is an interesting article and finding. One thing comes immediately in mind. What about if you rolled back to build 18 or 16 and repeated the same test? That should yield some further information.

Posted by .(JavaScript must be enabled to view this email address)  on  09/11  at  01:56 PM

---

Great piece, Art. 

Of course it reminds me of why I love working in B&W;!

Jeff “wants a GRAY Camerz” Kreines

Posted by Jeff Kreines  on  09/11  at  02:15 PM

---

Such Chroma Matrix values can be very useful for compensating for the color of the light and the coloring of the subject. If they boosted the blue by adding some green then they may have needed to add some red to the green and green to the red to avoid the flesh tones from going magenta and make them more orange.

Fuji reduced the density of their masking layer in their 500 speed film when it first came out to reduce grain and improve the the ISO.

I have such chroma masking controls in my “freeish” 6K+ DI system’s color correction, its a download from my web site. It can be useful for improving RED TIFF frames or film scans from faded prints or color negatives, you subtract extra blue from the green and red, green from the red and blue, and red from the green and blue to get brighter cleaner colors, how much depends on the mixing in the frame being compensated, I have waveform monitors in the software to help with the adjustments.  So you could un-do the Build 20 mixing of the colors if you wanted to get brighter colors under tungsten lighting, and use my sharp/soft filter to blur the blue to reduce the blue noise, which I have tried on RED shots before to deal with underexposure of the blue sensors.

Posted by www.DANCAD3D.com  on  09/11  at  07:41 PM

---

Thanks, Art!  I’m getting ready to shoot RED with a lot of tungsten lights…
-Graham

Posted by Graham Futerfas  on  09/12  at  03:02 PM

---

You might do a test and see if there’s a lower EI that works better noise-wise for tungsten than for daylight. Monday I’ll go through my tests and see if I can figure that out. Tungsten just seems to be a lot noisier overall.

Can you use an 80D or 1/4 CTB filter?

Posted by Art Adams  on  09/12  at  03:35 PM

---

Hi Art—

Great article, can I ask for a bit of clarification?  Is this color mixing being done in the colorspace conversion (when interpreting bits as redspace or other), or is it baked into the raw data?

If it’s baked in (that is, not just a metadata setting), what’s to stop this color-bleed when shooting under 5600K light with a predominantly yellow-ish subject?  (In other words, if the camera detects and underexposed blue channel to perform this color-mashing, what happens when we want to underexpose the blue channel?)

Perhaps I’m not understanding… or perhaps this new build has some evil gotchas…?

Thanks for your time, and another great article.

Cheers,

Ryan

Posted by .(JavaScript must be enabled to view this email address)  on  09/21  at  03:44 PM

---

I’ve since discovered that RED isn’t doing any color mixing at all. Not intentional, anyway. My theory was that they were doing this to reduce color noise, because who would consciously do it for any other reason?

One of my next two articles will show that it’s not color mixing at all. It’s cross-talk between blue and green under tungsten light, and it’s a sensor issue, not a software issue.

Posted by Art Adams  on  09/21  at  06:07 PM

---

Hm… well, I’m intrigued, and I’ll look forward to the next article—but I’m still confused about how the camera ‘knows’ it’s under tungsten light—I mean, it only sees reflected light, so how can it possibly distinguish between a 5600K source bouncing off a yellow subject, and a 3200K source reflecting off a whiter subject?

And if Red’s to be believed, it can’t possibly be a color temperature setting, since that’s only metadata… right?

...uh, right?

Posted by .(JavaScript must be enabled to view this email address)  on  09/21  at  07:12 PM

---

That’s what I tested: whether there was metadata being encoded that had an effect on the camera’s colorimetry that changed its response based on white balance. As best I can tell there isn’t—the color issues between tungsten and daylight are hardware related, not software.

Posted by Art Adams  on  09/21  at  07:24 PM

---

Name:

Email:

Location:

URL:

Smileys

Remember my personal information

Notify me of follow-up comments?

Submit the word you see below: