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 :)

Epic-X Color Rendering Issues

Sorry Jeff, I think we should be able to shoot green sweaters in tungsten lighting and expect the camera to reproduce the colors in an acceptable fashion. It's not always practical or aesthetically desirable to use daylight sources, especially in mixed lighting locations when you are working around existing color temps.

Why can't anyone admit the Epic just doesn't handle some colors very well? I've posted in several threads before about the problem epic exhibits with certain makeup tones. Have you ever used deep saturated gel colors? The camera certainly doesn't render party gel colors faithfully either. Color science is the primary improvement I think could be made with the camera. I'd almost rather the red team focus on this over higher iso or even dynamic range in future sensors and upgrades.
 
Sure Eric, you can look at it that way, but CMOS sensors don't handle tungsten lighting as well as daylight sources. Doesn't matter if it's RED, ALEXA, Canon 5Dxxx / C300, etc..

I'm still betting they can get the color they want with proper processing and grading. They are shooting RAW, after all. And Canon "RAW" is still exposure-baked and color-corrected / compensated out of the camera, R3Ds are not...
 
It does matter if it's a red, alexa, or 5D. I haven't experienced nearly the amount or severity of color rendering problems with the Alexa or 5D as I have with the Epic. I love my Epic, but the color could be improved. Sometimes it seems like redusers make excuses for the camera, which prevents legitimate issues from being heard, responded to, and eventually solved in future iterations of hardware and software. Cameras show colors differently. Maybe you CAN get the green in the sweater back with some heavier/more precise grading, but the truth is that his other camera rendered the color accurately without any issue. It is fair to ask that the RED team take note and try to improve this.
 
lower than 4400k it looks like (to me that red has dropped the float curve on the blue channel in favor of a multi. Which isn't bad. Just looks like that.

I am sure I am wrong Graeme.

There were some build that looked like pure float curve, but to me it looks like it's been abandoned.

OK

THIS is just observations from a number of images analyzed, and NOT a fast/true knowledge.


Just sayin'

But it adds to the issue discussed here.

To check if it is really a IR issue, try grading from a 5600k RLF frame at 800.

That'd be my first check, actually.

To me, it looks like resolve and other apps handles a few color issues better than the SDK, when fed a "flat" image (Log r linear, not REC and whiteballanced)

This not as a general advice, but as a troubleshooting advice.

Agree with Jeff on all points, btw...
 
Here is a test I shot myself with the same green sweater shown in my earlier post. I used a single 1k Fresnel, metered at 2940k. Camera was in factory default with the WB set to 3200k. As you can see, the sweater that is quite green to the eye, is straight up grey here. Any more thoughts? Will adding blue to the lights (and raising the color temp. to say 4000k), somehow solve this? Tungsten BTW is being used for budgetary reasons.
https://dl.dropbox.com/u/91462025/A001_C001_1003SU.0000009F.R3D

Thanks,
 
Haven't started playing, but we're seeing crazy IR properties with that sweater. I can see through it. Anyone wearing that sweater who happens in front of a night IR camera or typical security camera may as well not be wearing the sweater at all. ;)

I'll take a closer look at the R3D here in a minute, but what I'm sticking with what I've said above at this point.
 
Played some more, only way to fix this in post is to isolate the sweater and re-color.

I'm with Hans, I would file a ticket. I still think it's a combination of your lighting setup and the IR properties of this sweater (in this example), but either way, the camera should not be that susceptible from IR, even at 100% reflectivity, from a single fresnel. I would definitely file a ticket or just email one of the RED guys directly.
 
Red is very aware of my situation, and is working on solutions. I just thought I'd throw this out to the world. You can't get enough opinions when it comes to crazy stuff like this. Regarding "filing a ticket" I'm guessing you're implying there is a problem with this particular camera. My OTHER Epic-X yields the exact same results, so it is a hardware/software issue.

Thanks,
 
RAW deal

RAW deal

IR filters are intended to kill wavelengths over 750nm (or thereabouts) with zero effect on visible spectrum color rendition. Unfortunately, none are perfect so each camera system design is a matrix of compromises. I'm not surprised that specific colors, particularly with synthetic fabrics in the mix, render oddly when "developed" nominally from the RAW data set. I would love to see the green sweater in question, same lighting, photographed side by side with a color swatch on neutral material that appeared identical to the naked eye. Another great test would be to shoot the same sweater with daylight balanced lighting and a hot mirror.

FWIW the RAW data set in the r3d file is perhaps "raw-er" than what most manufacturers give the user access to. Strength or weakness? Like most engineering issues its likely some of both :-)

Cheers - #19
 
You must log in or register to reply here.
Back
Top