Green artifacts Helium

[Steven Dean]

Got my Weapon Helium the other day, and have been testing...

I have been getting some strange green artifacts on some shots. Part of me is thinking it is compression, but I would like to see if anyone else has some insight here. Poor black shade maybe? (this was Auto)... This was shot at 75fps, at 8k WS, which defaults to 16:1 compression with the standard mags. Take a look at the rocks in the foreground, at at %100, or more, and you will see what I am talking about. Conversely, the sky does not have this. I have owned a RED since the Scarlet MX days, through Epic, Epic Dragon, and Weapon upgrades, and have never seen anything like this. It doesn't really have an effect on the overall gestalt of the image when not pixel peeping, but this could be an issue when shooting lighter colors, and definitely want to nail down what it is, whether it is higher compression, poor black shade, or something else. Definitely going to do some more testing, but interested to see if anyone has any insight.


Thanks!


R3D:

https://drive.google.com/file/d/0ByC...ew?usp=sharing




Edit - first off, (which I probably should have said straight away) - I love Helium!!! After it was delivered, I had a couple of hours to go and grab some shots quick. When I got home, I latterly yelled expletives for about 30 seconds in complete awe of the image I was seeing in RCX. The Helium upgrade is definitely worth it, and I am very happy with it - as others have said - Helium definitely crosses a “threshold,” of resolution, and color, that I only ever imagined. Tested as much as I could that evening, and next morning, had a commercial shoot - no problems, gorgeous image, etc… We had a Weapon 6k on hand just in case, but didn’t need it. I am exceedingly happy with Helium - just trying to nail this down moving forward, and am not attempting to be alarmist about it in the least.

Lower compression definitely minimizes the green artifact issues, but it is still there a little at 6:1. This is when super pixel peeping, which I don’t usually do, but new camera, new sensor - I’m going to do it… I ultimately love the image coming from the Helium sensor, and feel that this is not a huge issue, especially when not pixel peeping, however, working on larger budget commercial shoots, I wanted to find some insight. Will definitely be testing some more, and really need to get the higher data rate Mags so that I can choose lower compression at 8k high speed - lower compression helps a lot.

Helium is the “camera to beat,” at the moment, and probably for years to come. I love this forum, all of the helpful insights, and of course, the entire RED team! Seriously - I love the images coming out of Helium (more to come on that).

[Anton Shavlik]

I think it's a compression issue, at least it looks like macroblocking, or whatever the wavelet version of that is called since it's not neat blocks.

Helium images do seem to sit on a frothy magenta/green noise profile so I think the compression just made bigger bubbles

[Stacey Spears]

Hi Steven, Can you post an image with the artifact in question circled? I think I might see what you are talking about, but want to confirm. You can shoot the same shot in 5:1 at 24p just to double check compression.

[Aaron Lochert]

Definitely looks like color noise, muddied up by compression to me.

[Graeme Nattress]

Yes, it's over-compression. Try turning on DEB as that should clean it up a lot.

Graeme

[Daniel Stilling]

Compression for sure. But I also noticed that you shot it at 320 ASA.
I feel that you underexposed a little in the shadow areas, and that doesn't help with the issue at hand
By having the camera set at 320, you had to protect a little extra the highlights. By doing that, I feel that the shaded areas are a little underexposed as you protected the sunlight bouncing of the water and the sky.
Might not have made that much of a difference, but compounded with the high compression ratio, it might...

[Steven Dean]

I think it's a compression issue, at least it looks like macroblocking, or whatever the wavelet version of that is called since it's not neat blocks.

Helium images do seem to sit on a frothy magenta/green noise profile so I think the compression just made bigger bubbles

Anton - that seems to be the case, at least for the most part… It is still present in the newly attached file at 6:1 (bottom of my reply), though definitely not as bad, and in different areas.

Hi Steven, Can you post an image with the artifact in question circled? I think I might see what you are talking about, but want to confirm. You can shoot the same shot in 5:1 at 24p just to double check compression.

A couple of the areas highlighted - more in similar tonal ranges throughout (jpeg):

https://drive.google.com/open?id=0By...W1FVVI2UVBzTUU

Yes, it's over-compression. Try turning on DEB as that should clean it up a lot.

Graeme

Thanks Graeme! Have to say - I love Helium, and IPP2 is AWESOME!!! It’s really nice to have it in camera… I will give that a try, and see what happens. Thanks for everything that you do!

Compression for sure. But I also noticed that you shot it at 320 ASA.
I feel that you underexposed a little in the shadow areas, and that doesn't help with the issue at hand
By having the camera set at 320, you had to protect a little extra the highlights. By doing that, I feel that the shaded areas are a little underexposed as you protected the sunlight bouncing of the water and the sky.
Might not have made that much of a difference, but compounded with the high compression ratio, it might...

Daniel - yeah, still seeing what Helium can do as far as highlight retention, and I definitely “played it safe,” on this one, instead of pushing it, though I know that many ETTR instead. Maybe I need to change my practices as far as that is concerned. I agree - this is definitely “Exposed to the Left” in order to “hold” the highlights, which as you point out can introduce more noise in the shadows. This has been my general philosophy with landscapes, for better, or worse - I really hate blown highlights in landscape shots, which, of course, can introduce shadow issues. Landscape shots are pretty much the only time that I use this philosophy. Combine that with lack of ND on hand that was strong enough, this was my result. Definitely need to do more testing for sure.



Here is a shot from the same location at 6:1 compression… The artifacts on the rocks are not as bad, but the water has some of the same issues, as does the horizon line. Part of me keeps asking myself if this is CA, but I have never seen CA present in this specific green color, though it sometimes appears to be in strong contrast areas.

23.98 6:1 compression

R3D
https://drive.google.com/open?id=0By...XJWblQzaW1iQ0U

areas highlighted (jpeg):

https://drive.google.com/open?id=0By...lBaRG9tSllkMmc




Thanks for the replies everyone, I love the helpfulness of this forum!

[Anton Shavlik]

Hey, I've only checked the jpeg, but the highlighted areas look normal to my experiences with helium and ipp2. IPP2 is a much sharper debayer at the pixel level and you can see lone dark and bright pixels like you have in the wave you highlighted. I'm not sure why red's pixels seem to have more color to their noise than Alexa or dslr raw photos but it's totally normal.

Likewise to the horizon, these green/magenta leaning areas seem to appear where the sensor is up against really fine detail. I noticed most of this cannot be seen when the image is scaled to 4K.

My 2 cents at least

[Steven Dean]

Hey, I've only checked the jpeg, but the highlighted areas look normal to my experiences with helium and ipp2. IPP2 is a much sharper debayer at the pixel level and you can see lone dark and bright pixels like you have in the wave you highlighted. I'm not sure why red's pixels seem to have more color to their noise than Alexa or dslr raw photos but it's totally normal.

Likewise to the horizon, these green/magenta leaning areas seem to appear where the sensor is up against really fine detail. I noticed most of this cannot be seen when the image is scaled to 4K.

My 2 cents at least

Thanks Anton, I appreciate your insight!

[Aaron Lochert]

I would try a manual blackshade. I noticed less green in the lows on my Epic-W when I did it.

Also, if you're on the Beta version of Redcine-X, I noticed less green, again.

Release Build left (with IPP2 LUT workflow), Beta right (with built-in transforms):