Welcome to our community

Be a part of something great, join today!

Register 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 :)

Helium weird bokeh / clipped highlights (RED tech please look at this)

J

Jason Han

Well-known member
Joined
Sep 25, 2015
Messages
203
Reaction score
0
Points
0
Location
Toronto, ON
The first thing I noticed with the Epic-W Helium apart from the astonishing low light capabilities, were these nasty hard clipped bokeh's. I did extensive tests with several cameras, and ran tests in post and found that using dragoncolor instead of Dragoncolor2 (suggested by Phil Holland) and REDWideGamutRGB, helped reduce some of these clipping issues, but does not eliminate them. Photos on the left are from Tim Daust's tests (reference: https://vimeo.com/189478583), photos on the right are from my tests (reference: https://vimeo.com/188865905). Both most likely wide open, I believe that's when the problem occurs (shooting wide open at spotty highlights). Because the colorspace seems to have improved these problems, is it safe to assume helium color will likely solve these problems?

bYKF1fH.png
JRznsVN.png

ByCJxwf.png
R9ZmZp9.png
 
Look at the images in Linear light. That will tell you if clipping is from the sensor or the gamma/color science.
 
I see the same thing. Hoping it's color science. Until we know I sugest not to expose to far to the right on helium.
 
I know experiencing the same and have been thinking I'm going crazy with my exposures so like you suggested Bjorn I will be exposing ETTL until Helium colour is released.
 
Glad to hear I'm not the only one dealing with this. Here is a raw shot, completely clipped

tAjuwby.png
FfhiLoP.png
 
You only have to be careful not to clip the highlights - they don't clip by themselves, they get clipped. :biggrin:
 
Šabović Adis;1683597 said:
You only have to be careful not to clip the highlights - they don't clip by themselves, they get clipped. :biggrin:
Click to expand...

Have you tested it? As this is not like when dragon clip, This is a bit different.
 
Šabović Adis;1683597 said:
You only have to be careful not to clip the highlights - they don't clip by themselves, they get clipped. :biggrin:
Click to expand...

Even if these highlights are clipping, their falloff should be much softer. The hard edge on these is what concerns me, not the fact that they're clipping. I really hope it's just a color science issue. It appears to be mainly on highlights that are red and blue, not white or other colors.
 
I agree Johnathon, it's the lack of softness when clipping. It's so so different to dragon.

Ive also had a bit of an issue with splotchy transitions between perfectly exposed and a stop under on skin. I have sent in a couple of r3d frames to red for clarification in case I'm missing something.
 
A few files for testing... + I believe a lot of the these artifacts are from pushing to hard from a DC2 RG4 starting point.

OUTSIDEBARTEST_dc2_rg2_CC.0000330.jpg

r3d = https://www.dropbox.com/s/qebc5j6vxudhsms/A001_C046_1019MQ.0000330F.R3D?dl=0
tif 8k as shot DC2 RG4 = https://www.dropbox.com/s/dj3336mopezwx19/OUTSIDEBARTEST_dc2_rg2_AS_SHOT.0000330.tif?dl=0
tif 8k simple CC still DC2 RG4 = https://www.dropbox.com/s/6wjau1f1vf3a2ph/OUTSIDEBARTEST_dc2_rg2_CC.0000330.tif?dl=0
tif 8k RWG L3G10 = https://www.dropbox.com/s/jpsrw1hrhx9p42q/OUTSIDEBARTEST_WGR_L3G10.0000330.tif?dl=0

A001_C034_1019ZQ_DCRG2_CC.0000103.jpg

r3d = https://www.dropbox.com/s/moz8xyzk3dcvylo/A001_C034_1019ZQ.0000103F.R3D?dl=0
tif 8k as shot DC2 RG4 = https://www.dropbox.com/s/ifzs369c36aadt6/A001_C034_1019ZQ_DCRG2_AS_SHOT.0000103.tif?dl=0
tif 8k simple CC still DC2 RG4 =https://www.dropbox.com/s/vheiaiglk5d49ef/A001_C034_1019ZQ_DCRG2_CC.0000103.tif?dl=0
tif 8k RWG L3G10 = https://www.dropbox.com/s/n2jnvwl456j9vm1/A001_C034_1019ZQ_WGR_P3G10.0000103.tif?dl=0
 
Björn Benckert said:
Have you tested it? As this is not like when dragon clip, This is a bit different.
Click to expand...

What's there to test - if it burns out it's burnt. And yes, I had some Helium R3Ds to play with, some of them had clipped highlights and some not.
It's always the same: know Your stuff, Your gear, don't overload it and don't be pushy...:biggrin:
 
Jonathan Stevenson said:
Even if these highlights are clipping, their falloff should be much softer. The hard edge on these is what concerns me, not the fact that they're clipping. I really hope it's just a color science issue. It appears to be mainly on highlights that are red and blue, not white or other colors.
Click to expand...

I'd say it's up to gamma space, regarding the "fallof", not the color science, but yes, let's wait for the Helium color space - it can't hurt to have the "apropriate" one...:biggrin:
 
 
 
Jason Han said:
Glad to hear I'm not the only one dealing with this. Here is a raw shot, completely clipped
Click to expand...
Jason, which color space are you using in all of the above posted pics? As Michael points out, if they are not clipped when you view in linear light, then they (detail) are there and its just the color space option being selected. If they do not clip in linear, are they (Again, detail) also present in RedWideGamutRGB? If so, then the new color space stuff will handle this.

In the mean time, I would suggest everyone send samples to Graeme to look at while we wait for the new color space to be released.
 
PastedGraphic-3 by Björn Benckert, on Flickr

PastedGraphic-4 2 by Björn Benckert, on Flickr

I think we can rule out problems at the sensor level and instead sit patiently until new color science is baked and ready. Dragon color is simply not helium color :)

Edit: and here is another screen grab at 800ISO zoomed in on the screen. To me the first one with wide gamut and a lut looks accurate, the second one is completely of with dragon color and then last one I guess is accurate as well, don't know how the screen looked in real but it seams right that it would go from blue to more white when doing an extreme iso or exposure change.

PastedGraphic-5 by Björn Benckert, on Flickr

Edit again:

Below Just to see how it deals with color balance here is one at 3200K still looks accurate in the highlights with redwidegamut Where as the old color science simply don't handle the pure blue burnout of that the screen. (the second of the two)

So whats at discussion here is color science that as I understand is still in the works.

PastedGraphic-6 by Björn Benckert, on Flickr

PastedGraphic-7 by Björn Benckert, on Flickr
 
I have seen multiple clips where DC2/RG4 looked clipped but putting into RWG/L3G10 they were clearly not clipped. Applying the Lutify.me LUT to these clips makes them look similar to the DC2/RG4 but almost all the clipping is gone. If there are no traffic lights on when captured, then there is no clipping of the raw signal. Also setting the ISO to a higher value of course helps.

Here: https://lutify.me/free-redwidegamutrgb-log3g10-log-conversion-luts/
 
Last edited:
Tim Daust said:
here you go -

View attachment 89241
r3d = https://www.dropbox.com/s/1wttzwqfc4c9kc1/A001_C041_1019GY.0000213F.R3D?dl=0
tif 8k as shot = https://www.dropbox.com/s/96jhirx3dgdrd96/A001_C041_1019GY_AS_SHOT.0000213.tif?dl=0
tif 8k CC = https://www.dropbox.com/s/rbhqncklbc9rtbo/A001_C041_1019GY_Cc.0000213.tif?dl=0
tif 8k RWG L3G10 = https://www.dropbox.com/s/zj5aruhprr9dm53/A001_C041_1019GY_RWG.0000213.tif?dl=0
Click to expand...

Thanks Tim for posting this and so glad it looks like something that can be fixed (thanks Bob and Bjoern).
I got a lot of events (at night) coming up with heaps of artificial lights. So that has the potential to look awful...
Now I can't wait for the new colour science. Will do some overexposure test with colourful highlights. I looked at Mark Toias video again
and I can't find these kind of blown out highlights, looks mostly just like CA from the L-glass. Maybe he used the skin tone olpf?
 
Björn Benckert said:
Edit again:

Below Just to see how it deals with color balance here is one at 3200K still looks accurate in the highlights with redwidegamut Where as the old color science simply don't handle the pure blue burnout of that the screen. (the second of the two)

So whats at discussion here is color science that as I understand is still in the works.
Click to expand...

Thanks ! I really need to export my video again using RWG /Log3G10 - What Transform/LUT are you putting onto the r3d there after you switch to RWG/L3G10? Its matching pretty good to the original but highlights retained. Cant wait for new color science.


Stefan Kraus said:
Thanks Tim for posting this and so glad it looks like something that can be fixed (thanks Bob and Bjoern).
I got a lot of events (at night) coming up with heaps of artificial lights. So that has the potential to look awful...
Now I can't wait for the new colour science. Will do some overexposure test with colourful highlights. I looked at Mark Toias video again
and I can't find these kind of blown out highlights, looks mostly just like CA from the L-glass. Maybe he used the skin tone olpf?
Click to expand...

Yes sir! just trying to do whatever I can to contribute. So thankful to have these cameras.
 
You must log in or register to reply here.
Back
Top