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

N

nacamera

Well-known member
Joined
Apr 18, 2007
Messages
294
Reaction score
0
Points
16
Age
64
Location
Milwaukee, WI
Website
www.nacamera.com
Hello all,
I am having recurring color rendering issues with my Epic-X. The camera is being used for retail clothing commercials, so the accuracy of the color for the various clothes is critical. Lighting is tungsten. There have been Hot Mirror/ND combo filters used sometimes, but in this case it was a clean 17-80mm Optimo being used. The camera seems to have major issues rendering various shades of green clothing. Colors come off looking almost grey, or teal will look forest green, etc.
It does not appear to be correctable in RedCine-X.
Just wondering if anyone else has had a similar experience?

Thanks,
 
The only recent major issues have been with the new Macbook Pro EFI updates, update 8.2. Beyond that more detailed information might help with troubleshooting any possible issues.
 
Green turning grey sounds like an ir issue, especially if it gets muddy along the road.

Could you post a single r3d snap, or mail me one to gunleikgroven at gmail dot com?.

Cheers

G
 
.r3d still

.r3d still

can you post a .r3d still please... Maybe your Temperature is correct but your tint setting is off or maybe there is another factor ...if we could look at the actual .r3d it would be better
 
Actually or Additionally if you post a screen grab of your work-space in redcine-x or other in order to see all of your settings. For reasons unknown to me or if it's sensor related in that the red's sensor i understand is daylight rated.....i almost never shoot with tungston lighting. BUT, my issues are related to underwater and even still i find I can bring back colors to where they should be. But not knowing any of your post settings does not allow us to evaluate and give feedback and all of the usual apply: did you blackshade and allow camera to come up to working temp etc....
 
Looks like a magenta tint + possibly ir issue.

The tint on different epics seem to vary... How to put this... A bit...


No real issue, but it scares a lot of photogrphers i know.
 
Sorry, I don't have access to the R3D files, or anything more than this JPEG. As I had said, this has been a recurring issue for months over many clothes/fashion based shoots. It definitely is a camera issue as we brought a second Epic-X body to the set, with the exact same results. The footage from a past shoot was given to a 25+ year veteran colorist, and he couldn't bring a piece of clothing around to the proper shade of green no matter what he tried. He described it as a "blind spot" in the cameras color rendering range. I am working further with Red to get this figured out. I'll post the answers here once I get them.
 
No ND filters were used. It was straight tungsten lighting with the iris set to the appropriate stop correction. Therefore no IR contamination or at least no more than the camera's built-in IR filter should be able to handle.
 
PLease keep us posted here. I have found that the Epic renders certain colors in a very strange way. It's not the most color accurate camera by any means, and maybe not the right tool for your job because of this.
 
As a veteran fashion catalog photographer, I've seen this many times of the past 13 years (with digital capture). This can be related to IR, as certain fabrics reflect IR, while others do not. Not all digital sensors are the same. It seems the Canon sensor does a much better job at eliminating (blocking) the IR, while the Epic lets it pass through. It would be in your best interest to use an IR cut filter in front of your lens. Even in the days of film, we always had to color match certain colors prior to printing. All of this is correctable, and your colorist should be able to select the sweater, and dial in the correct adjustment.

Von
 
My comment is a bit offtopic and related to personal red-green partial colourblindness. Here is a weird thing. When I first compared the colour charts linked to in a preceding response, they looked identical. Then on closer inspection I see the green box reproduced as grey as described. Then I find if I look directly at the pink or brown to the left, the grey box changes to green, then back to grey if I return my stare directly upon it.

IR contamination as a cause makes sense. I found that with stone fruit tree foliage, a look that just did not appear quite right. The upper surfaces of the leaves in direct sun reproduced faithfully. The undersides where one's eye would expect to see a darker green from flow-through of the light, seem to reproduce as a greyish tinted lighter colour. One would expect a deeper shadow if the light level from the lower side of the leaf was below what the sensor would see. This was on a f16 day with the SI2K. The TruCut IR750 filter seems to yield more faithful greens and orange clay/laterite rock colours. This is purely a subjective impression and no science involved.

It never entered my head that indoors lighting should get the TruCut filter treatment as well but for tungsten lamps, it makes a lot of sense. I therefore very much appreciate this topic having been raised to my own selfish benefit.
 
Last edited:
Green turning gray and muddy is usually an ir issue. It's kinda the strongest ir warning you can get. I usually do not worry much for ir as i usually recognize it in the preview if it an issue. But then again, i allways shoot the cam flat, so ir and other issues are not surpredsed or confused by metadatasettings...
 
If I understand IR contamination correctly, you have to cut down the visible light by around 3 stops or more before it becomes evident. There was NO ND filter used in these shots. The iris on the camera was just set to whatever----probably a 5.6-8 and they hit record. Is IR contamination even possible under these circumstances? Doesn't the Epic have SOME built in IR abatement?
This problem has occurred on SEVERAL other shoots in which we used a Formatt Hotmirror ND combo filter with the same results of the green being off in much of the clothing.
 
But that is not neccesarily right.

I have taken out 7 stops without IR issues, and I have had IR issues without any filtration.

IR is not a constant.
Nor is the issue.

And I am not a huge fan of allways filtering for it "just in case" as I find it rather easy to spot on set, and I think the IR filters aren't all that swell for skintones...
But that again... it is spot'able because i shoot the cam flat , and don't let metadata confuse what is recorded...

But that's just me.
 
We use Formatt Filters ND's and Hot Mirror, we also use a Pancro Hot mirror IRND. The Formatt Hot Mirrror is available in two different types which IR attenuation is set diferently on each filter, I dont recall the wave length, one of them is shown here.

http://www.formatt.co.uk/showimage.aspx?id=wf3000000036

May be the format Hot Mirror is not agressive enough in dealing with the IR if this is the issue.

we also dont like using IR filtration where we don't have to.
 
nacamera said:
No ND filters were used. It was straight tungsten lighting with the iris set to the appropriate stop correction. Therefore no IR contamination or at least no more than the camera's built-in IR filter should be able to handle.
Click to expand...

Don't be too sure about that. With hot tungsten, especially in the sub 3000K range, you stand a chance to have a lot of reflected IR. Certain types of clothing materials are like mirrors for IR.

Not to pick apart the methodology here, but I have to wonder why this is being shot at 2900K? We also have to take into account the comparison with the 5Dii. It's not that it blocks IR any better, but it has rather advanced WB and spectral monitoring and compensation. There's a chance it sees that sweater as grey too, but compensates for the lack of green in the frame that it is able to detect. We could tell a lot more about what is going on if you could post a frame or two from the R3D file. If you don't have access to it, perhaps you can get whoever does have access to post this?

If it's a constant issue with more than one camera, then I'm going to say it's almost certainly IR with that lighting setup. I would recommend pushing the light temp closer to 4000K. At 2900K, you're killing your blue channel information anyway and sacrificing a lot of chroma info and detail. Sure the camera does handle tungsten well, but the sensor is still biased toward the daylight range -- about 5000K. I'm not sure I would recommend a hot mirror IR filter in this situation, it seems the lighting design just needs to be addressed. Shots with the 5Dii or other DSLRs will benefit too. Just because they don't show the same anomalies right out of the camera, doesn't mean the issues don't exist.
 
nacamera said:
If I understand IR contamination correctly, you have to cut down the visible light by around 3 stops or more before it becomes evident.
Click to expand...

No, you don't need to cut visible light to see it. IR is picked up by the sensor, the OLPF does block some level of IR, but not all. If you're shooting a certain fabric that reflects IR, then it can be an issue and can muddy your colors. Some materials will reflect more IR than visible light. No ND required to see it.

Additionally, the 2900K lighting setup is killing your blue channel.

All that said, color tweaking matching and all out replacement is commonplace in fashion photography, magazine / print ads, etc.. So makes one wonder about color intent here... If the sweater came out the same color on the R3D file as the 5Dii file, would it be used that way or would it get changed? Who's doing the post work or "developing" of the R3D files and how far can they manipulate the sweater color before it breaks? What are the camera's color / look settings, etc..? Kinda hard to help from a JPEG file. But to me this still looks like a 2900K + IR issue.
 
You must log in or register to reply here.
Back
Top