Epic-W Gemini Low-Light Test Footage + R3D

[AndreasOberg]

I wonder how ISO compares to the old MX. I just review lots of videos and tested what would happen if I would move the ISO up to 3200. Many of them do not display THAT much noise. However the colors are quite dull. As I understand Dragon was a bit worse than MX when it comes to noise, am I correct?
/Andreas

 

[Tehben Dean]

I wonder how ISO compares to the old MX. I just review lots of videos and tested what would happen if I would move the ISO up to 3200. Many of them do not display THAT much noise. However the colors are quite dull. As I understand Dragon was a bit worse than MX when it comes to noise, am I correct?
/Andreas

When I compared HELIUM and MX directly I found the low light noise was practically indistinguishable between the two but the HELIUM had more latitude in the highlights.

 

[Aaron Lochert]

Even at 6400, the mustache is a clear example of more detail, even though it's less resolution.

Detail is definitely being swallowed up by noise and I'd never shoot Helium that high for obvious reasons, but the focus was buzzed on the Helium shot. The plane of focus is behind his eyes.

 

[AndreasOberg]

Yes the focus was off in the shots so I wouldn't use this test to check sharpness or texture.

Honestly I would be surprised if Gemini has more detail than Helium, Helium has so much higher resolution.

When I compared HELIUM and MX directly I found the low light noise was practically indistinguishable between the two but the HELIUM had more latitude in the highlights.

That is really interesting. Was this at 100% crop or scaled down to 4k?
What I find with the MX is if I push the ISO higher than 1200 the colors are really bland so I would have to be pretty desperate to go higher.

I studied the Bulb test by Luigi and I cannot see that Gemini has a higher dynamic range than Helium. To me it looks like Helium has a higher range, I just see more gradients in the bulb on Helium. Maybe someone more experienced with dynamic range could chime in.

/Andreas

 

[Aaron Lochert]

I studied the Bulb test by Luigi and I cannot see that Gemini has a higher dynamic range than Helium.

That test is Dragon, not Helium.

 

[Tehben Dean]

That is really interesting. Was this at 100% crop or scaled down to 4k?
What I find with the MX is if I push the ISO higher than 1200 the colors are really bland so I would have to be pretty desperate to go higher.

I need to see if I can dig the shots up. Since they are vastly different resolutions I think I was just comparing them at viewing size.

 

[AndreasOberg]

That test is Dragon, not Helium.

I looked at the LL version so that explains the lack of dynamic range

 

[Jonathan Shugarman]

What was used for the noise reduction?

 

[AndreasOberg]

Vincent L has some interesting footage here:
https://vimeo.com/263659664
I see some strange fixed noise paterns in the image, esp for 3200 iso. Anyone else see this? Lack of black shading?

 

[Audy Erel]

Vincent L has some interesting footage here:
https://vimeo.com/263659664
I see some strange fixed noise paterns in the image, esp for 3200 iso. Anyone else see this? Lack of black shading?

On his Facebook he stated that he used Neat Video noise reduction plugin for all ISO 3200 and above, which I really think defeat the purpose of the video as lowlight cinema camera test (we always want to see the unaltered, real result).. And yeah, that static noise dark purple dots is really weird but maybe it's an artifact from the plugin.

 

[Prahlad Strickland]

The Gemini is awesome. I wish red color science could do what that Alexa mini does on faces. Def some magic going on there that I’m missing with the red sensor / color science.

 

[Michael Lindsay]

The Gemini does look impressive but I'd be very interested to see how it looks with a Skintone OLPF inserted?.?

Since it has sensitivity to burn it may make a good mid way compromise between tight accurate colours in shadows and LL performance...

 

[Sabovic Adis]

The Gemini is awesome. I wish red color science could do what that Alexa mini does on faces. Def some magic going on there that I’m missing with the red sensor / color science.

Is there anything ARRI's color science does on faces that You can't do yourself?

 

[Tim Daust]

The Gemini is awesome. I wish red color science could do what that Alexa mini does on faces. Def some magic going on there that I’m missing with the red sensor / color science.

do you even IPP2 Tho?

 

[Zeb B]

do you even IPP2 Tho?

IPP2 was a reinvention of RED imho. Like back when LightRoom went from V3 to V4 - a whole new raw process engine that allowed huge amounts of previously 'locked up' data to be utilized. Was able to revisit and remaster many of my older print files pushing and pulling without breaking them

 

[Kenneth C Merrill]

do you even IPP2 Tho?

I love me some IPP2, but it's still not quite up to Arri level. However, it's not too hard to make up the rest of the distance yourself these days.

 

[Kenneth C Merrill]

do you even IPP2 Tho?

2:19 in this video is a GREAT example of where Red still struggles. https://www.youtube.com/watch?v=Z540kmuQduI

Why is she so yellow? Fire gives off an orange glow, not a yellow one. Watch IPP2 footage of sodium street lights. It's that same yellow. Red still has a hard time catching nuanced hues like this.

Here's another example I shot myself:

Sony A7RII on the left. Red Epic-W Helium on the right. I LOVE the skin tones of the Red, but the blue is simply not accurate. The gel we had on the light was indeed a teal, not a true blue. The Sony nailed it (though of course Sony skin tones suck)--the Red totally missed the mark.

It's funny because I see people talk all the time about how Arri color is more stylized to look good while Red is going for "accurate." In my experience this is just making excuses, as I have found again and again that Arri's are much more accurate, and Red--even IPP2--struggles with the subtle stuff.

 

[Aaron Lochert]

Why is she so yellow? Fire gives off an orange glow, not a yellow one.

If you want orange glow all you need to do is white balance for that:

Straight IPP2 workflow, with a different white point and some saturation - that's all.

 

[Kenneth C Merrill]

If you want orange glow all you need to do is white balance for that:

Straight IPP2 workflow, with a different white point and some saturation - that's all.

You win this round

 

[Misha Engel]

2:19 in this video is a GREAT example of where Red still struggles. https://www.youtube.com/watch?v=Z540kmuQduI

Why is she so yellow? Fire gives off an orange glow, not a yellow one. Watch IPP2 footage of sodium street lights. It's that same yellow. Red still has a hard time catching nuanced hues like this.

Here's another example I shot myself:

Sony A7RII on the left. Red Epic-W Helium on the right. I LOVE the skin tones of the Red, but the blue is simply not accurate. The gel we had on the light was indeed a teal, not a true blue. The Sony nailed it (though of course Sony skin tones suck)--the Red totally missed the mark.

It's funny because I see people talk all the time about how Arri color is more stylized to look good while Red is going for "accurate." In my experience this is just making excuses, as I have found again and again that Arri's are much more accurate, and Red--even IPP2--struggles with the subtle stuff.

Don't forget the influence of the lens, lots of photo camera's do lens compensation (at least with some lenses), where movie camera's don't.
That is one of the reasons you have to test with the same glass when testing camera's.