Epic-W shoot. The good things and issues with fringing and noise.

[Alberto Guglielmi]

I just finished a shoot for a new client. First time I put the Epic-W on some serious pace. I had some issues and some good things.

I shot pretty much everything at f/5.6, ISO800, and with Sigma ART 50mm as the main lens. Few with the 35mm ART. I shot mainly at 1/125th because I had to grab stills as a priority. Standard OLPF.
I also shot with my Leica M some stills for each setup.

Basically the quality of the files and the resolution is magnificent. Very comparable if not better than the Leica. Different feel on some things, but the quality of the file is identical, with more resolution.
Color is different, and I find the Lecia more pleasant out of camera. RED I had to move it around a bit, but I find a place I like. See next post for some samples of good images.
I am not a colorist, but I find that:
- DC2 and RG3 or RG4 is too much
- RedWideGamut and redlog + LTU that John Schick posted (THANKS). A bit too green in the skin tones for me.
- DC and RG3 with Green at 0.9 and Blue at 0.95 render the best skin tones.
Again I am not a colorist, and I welcome other opinions on this. I would love to try Graeme's new color on it.
See next post for some samples of good ones.

Issues:
See attached files. With light coming directly in the sensor, and wide angle resolution for full body, I was getting noise.
One shot in particular, black dress, really did not resolve properly and I have a lot of noise in it. Yes, a difficult light situation, but the Leica handled both much much better at the same settings. It is a difference between usable and not so usable.
UPDATE. I solved the fringing. It was too much sharpness in REDCINEx.

I welcome suggestions on better settings I should have used, both during shooting and in post.
You can find R3Ds here.
https://www.dropbox.com/sh/y0ffp0w0m7rumsh/AADIPwNvD6xxzknY3TJ2AcQTa?dl=0

Right now I feel that the Helium, due to the smaller pixel size, tends to create more issues on those situations (strong backlight and focus on small portion of the frame).
I am quite happy with the camera, and I hope that these are just quirks that are going to get solve soon with the new color and firmware.

 

[Alberto Guglielmi]

Some good shots. I love how the camera reacts tot he dynamic range.
Light was mainly practical, with some little units around.

 

[joey richey]

these are nice alberto! thanks for sharing.

using A007_C034_1122Q6_001 as a baseline, it looks like your red and blue channels are the noisiest. the green channel is really nice though, so i would suspect you needed a bit more light on the skin to be clean on all 3 channels. if you go down around 320-400, you can see the red and blue channels even up pretty quick. I'm not an expert either, but that fringing looks lens-dependent to me... Graeme would know much better than us on that for sure.

also, i find my M240 sometimes does better in the red channel than even the phase IQ245 back when under the same lighting conditions!

 

[Alberto Guglielmi]

Thanks Joey. I'll look into it. Fringing was too much sharpness in RedcineX. I just updated the post.
And yep, M240 is quite a good camera.

 

[joey richey]

yah, thats how i checked the fringing... over sharpen and wait on the green and purple party to begin!

 

[Aaron Lochert]

Had fun playing around with these.

As a starting point, I used REDWideGamutRGB and REDGamma3 + John Schick's saturation limiting LUT to bring it to REDColor4.

 

[Marc Wielage]

Had fun playing around with these.

As a starting point, I used REDWideGamutRGB and REDGamma3 + John Schick's saturation limiting LUT to bring it to REDColor4.

I don't think a LUT is necessary. I generally go with DragonColor and RedLogFilm and just color correct in a program like Resolve or Baselight. I don't find any saturation limiting is necessary except in cases with clipping (or near-clipped) reds, like car tail lights, and even then, there are ways to just soft clip them or pull back on that secondary a little bit on a case-by-case basis. The problem with the LUT is that it's fairly destructive and has no means for manual adjustment.

 

[Nawaf Alsabhan]

Thanks for posting the R3D's Alberto, I'll check them out soon..
For the Noise in the Black Dress, do u remember what the Redcode was set at?

 

[Alberto Guglielmi]

Had fun playing around with these.

Thanks Aaron. They look great. Did you move the rd/magenta down a bit. The second (girl on lounge chair outside) looks very pleasant to me. Better than what I did. If you care to share your settings it would be great. I am not an expert on treating RED files, and anything I can lear helps...

Thanks for posting the R3D's Alberto, I'll check them out soon..
For the Noise in the Black Dress, do u remember what the Redcode was set at?

8:1 compression.

 

[rand thompson]

Alberto,

I see you prefer Dragon Color/REDgamma 3 for Helium. Is there A reason you prefer those settings over REDWideGamutRGB /Log3G10 ? Do you still have out of range colors with those settings? I am still learning how to use .R3D files and would be interested because I've seen other red owners use REDWideGamutRGB but instead of LOG3G10 they use one of the REDgammas.

[/url]Untitled000864007 by rand thompson, on Flickr[/IMG]

[/url]Untitled00086411 by rand thompson, on Flickr[/IMG]

[/url]Untitled000864002 by rand thompson, on Flickr[/IMG]

[/url]Untitled0008640013 by rand thompson, on Flickr[/IMG]

 

[Aaron Lochert]

Thanks Aaron. They look great. Did you move the rd/magenta down a bit. The second (girl on lounge chair outside) looks very pleasant to me. Better than what I did. If you care to share your settings it would be great. I am not an expert on treating RED files, and anything I can lear helps...

For that shot, in Resolve:

500 ISO
6000 K
-5 Tint
-0.2 exposure
REDColor 4 (I used RWG + that LUT, but just redid it via Marc's suggestions and it looks almost identical. Best to avoid LUTs when not needed)
REDGamma3

Quite simple really. What I did was balance the color and exposure so that the scene looked right to my eye overall, ignoring her skin. I've had better luck this way so far with Helium than trying to white balance the image according to the skin. At that point, the skin was a bit dull and green so I did a secondary on it to add a little magenta in the shadows and push some orange in the highs. I then did another secondary on everything blueish in hue, and shift those blues a bit more towards cyan and saturate them slightly, since I typically like my blues to be a little greener than what you'd consider accurate. Lastly, I brought up the blacks just slightly to better match the low contrast look of the other shots.

 

[Alberto Guglielmi]

Thanks Aaron!
I guess I need to start using resolve...

 

[Karim D. Ghantous]

Alberto, I don't know much about coloring either, but I liked the look of your photos anyway. And they're nice photos, too.

The ones lit with natural light looked beautiful. The interiors were fine but a couple were not quite right. Also, I do not like to see crooked lines when they should be straight. It's very distracting.

 

[John Schick]

I don't think a LUT is necessary. I generally go with DragonColor and RedLogFilm and just color correct in a program like Resolve or Baselight. I don't find any saturation limiting is necessary except in cases with clipping (or near-clipped) reds, like car tail lights, and even then, there are ways to just soft clip them or pull back on that secondary a little bit on a case-by-case basis. The problem with the LUT is that it's fairly destructive and has no means for manual adjustment.

I would agree that LUTs are not always necessary and some do indeed introduce terrible artifacts (there are a lot of bad quality LUTs out there). Though, I'd also like to point out that not all LUTs are created equal. The saturation rolloff LUT I built avoids a lot of those issues, and doesn't introduce any unwanted banding or color clipping of its own.

The benefit of the saturation rolloff LUT in this case is it can do non-linear gamut mapping - something you can't get with a linear matrix. Also, once the colors do clip in REDcolor4, it can become very hard to dial the saturation back later (outside of linear light). Another plus is the LUT matches REDcolor4 very closely (as well as DRAGONcolor2), allowing for consistency when intercutting wide gamut shots with other REDcolor4 material - saving time compared to manually grading them to match.

Regarding being destructive, as long as the saturation LUT receives the expected input (REDWideGamutRGB / REDlogFilm) and is interpolated properly in the software, I think you would be hard pressed to find any issues with it (compared to starting with REDcolor4 / REDlogFilm).

I know LUTs have been getting a bad rap lately, but I would encourage those interested to try the LUT out for themselves and draw their own conclusions. Don't throw the baby out with the bathwater, as they say:
http://www.reduser.net/forum/showthread.php?151259-Saturation-Rolloff-LUT

 

[Graeme Nattress]

THere's nothing inherently wrong with LUTs properly used for their designed task. All LUTs are is a representation of underlying math, and when that math is appropriate to the colour space and linearity of the image, and the LUT is sufficiently precise and properly interpolated they're a very efficient and portable way of implementing that math.

Graeme

 

[Nawaf Alsabhan]

Hi Graeme, Could you check the R3D file for the black dress for the noise that Alberto was talking about?

Is it because of the strong highlight behind her?
I just want to learn Helium's limitation so we can avoid the noise when possible.

 

[Graeme Nattress]

It appears to be mostly compression noise on the red and blue channels. It could be helped by going to a less compressed REDCode than 8:1.

Graeme

 

[Nawaf Alsabhan]

It appears to be mostly compression noise on the red and blue channels. It could be helped by going to a less compressed REDCode than 8:1.

Graeme

Good to know, thank you Graeme.