Epic Vs Alexa Green Screen Test

[Esmaeil Neisi]

Ok guys We just finished an Alexa Vs. Epic Green screen test & I thought Maybe it will be interesting to post i wish it will be useful:
Hear is the Test Condition:
-The Epic shot with Ziees UP 65 & 20 mm
-The Alexa Shot With Arri Alura 18-80 Zoom At the same Focal length
-Side By Side at the same distance
-Alexa Shot logC on SxS 1920 x 1080 /24Fps/180 degree/5600 kelvin /ISO 500
-Epic Shot 5K FF /180 Degree /24Fps/5800 kelvin/ISO 500 /On SSD
-The Epic Files scaled to 1920x 1080 for fare comparison (i now maybe this shouldn't have done)
-All imported to AF and a Keylight 1.2 Applied and only the green screen selected for each file without any other adjustment.
would like to know what you guys think about it. not mean to say this camera or that is better just for sake of learning more stuff.
Any idea is welcome
BTW the Background Footage is A still from A lomo sample that shot by Gunleik Groven

 

[Esmaeil Neisi]

It Seems that The Keyers can't find the proper green screen ton on the Alexa Sample

 

[Patrick Tresch]

Do you use the keyer correctly? I'm sure Alexa is better than what is shown for key...

It's more a post issue than a camera one...

See you Esmaile

 

[Esmaeil Neisi]

Hear is more Samples

 

[Jeffrey Loewe]

Thanks for doing this test. While I won't be an Epic owner as of right now i'm excited to see the RED grab looks way better. Most of the cameras I work with and have shot for blue/green screen including HVX200's, EX3's, F3, and a 7D/5D dont look anywhere close to this, as to be expected. But compared to the Alexa even it looks better.

And since you mentioned they were shot side by side, at the same time, I'm assuming the same lighting since they were next to each other. I use keylight 1.2 for alot of my work, unless I use Mocha. And for some reason to me it looks as if the Epic interpreted the green color tone and interpreted it way better than the Alexa did. The Alexa didn't separate the colors as well since there are transparencies in each of the Alexa images. Based on this test, and I'd love to do more it seems as if this might be a weak point for the Alexa or maybe other variables could make a difference.

Thanks again for doing this test! I've been trying to find something like this for awhile.

 

[Esmaeil Neisi]

Patrick,
I know you can pull better key from Alexa footage but this test done only by selecting the green tone on the green screen for comparison in primitive way by
just selecting the green tone

 

[Patrick Tresch]

Have you applied exact the same keylight 1.2 effect to the other camera or did you make a new keylight 1.2 preset for the other camera?

Thanks


Pat

 

[Jeffrey Loewe]

Do you use the keyer correctly? I'm sure Alexa is better than what is shown for key...

It's more a post issue than a camera one...

See you Esmaile

I was wondering the same, I'd be shocked if the Alexa struggled this much with green screen shooting. But I have noticed using different cameras certain ones do struggle more than others.

 

[Esmaeil Neisi]

Thanks for doing this test. While I won't be an Epic owner as of right now i'm excited to see the RED grab looks way better. Most of the cameras I work with and have shot for blue/green screen including HVX200's, EX3's, F3, and a 7D/5D dont look anywhere close to this, as to be expected. But compared to the Alexa even it looks better.

And since you mentioned they were shot side by side, at the same time, I'm assuming the same lighting since they were next to each other. I use keylight 1.2 for alot of my work, unless I use Mocha. And for some reason to me it looks as if the Epic interpreted the green color tone and interpreted it way better than the Alexa did. The Alexa didn't separate the colors as well since there are transparencies in each of the Alexa images. Based on this test, and I'd love to do more it seems as if this might be a weak point for the Alexa or maybe other variables could make a difference.

Thanks again for doing this test! I've been trying to find something like this for awhile.

Thanks Jeffry
I agree with you it seems that the problem comes from how this cameras process the materials Epic 16 Bit internal Processing Vs. Alexa 12 Bit ,
and how knows if Alexa shoot Row on codex maybe the result would be better , thats some thing need's to be tested

 

[Esmaeil Neisi]

If any of you guys interested i can give him the green screen shot and he can test it in his one way no problem , but what i did is same for to camera even i scale the Epic photage to 1080 Some thing I shouldn't have done but still the deference is way beyond what i expected

 

[Eki Halkka]

I was wondering the same, I'd be shocked if the Alexa struggled this much with green screen shooting. But I have noticed using different cameras certain ones do struggle more than others.

The cameras are color corrected differently here - alexa has lower contrast and saturation than red, possibly also different white balance - and i'm rather confident that that's the main source of differences. Color match the two before keying, and the results should be rather identical.

 

[Eki Halkka]

If any of you guys interested i can give him the green screen shot and he can test it in his one way no problem , but what i did is same for to camera even i scale the Epic photage to 1080 Some thing I shouldn't have done but still the deference is way beyond what i expected

I'd be interested... so yes please, could you export a few frames from both?

 

[Esmaeil Neisi]

I just put a default keylight on each other then select the green background tone for each other individually.
Not the same keylight for to of them because each camera have its one characteristics

 

[paulherrin]

i hope you found this test useful. i'm not sure what you've learned by introducing a workflow that will never be used for the cameras. hopefully you learned something on the production end. i recommend doing post production tests with your real, usual workflow. an experienced compositor isn't going to need to worry about tests like this, but that doesn't mean you can't always shoot the plates better, the plate's cleanliness is always the bottom line.

 

[Esmaeil Neisi]

The cameras are color corrected differently here - alexa has lower contrast and saturation than red, possibly also different white balance - and i'm rather confident that that's the main source of differences. Color match the two before keying, and the results should be rather identical.

Eki ,
for best result the Alexa DIT said to record logC the difference you see between the too camera will be problematic when you apply the same keyer preset on both of them . i did not do so i just select the green tone fore each one individually so its not related to kelvin or CC ,
the kelvin measured by Sekonic c500 color meter and was 5600 alexa are rated at the exact kelvin but as i test my color meter i know there's about 200 kelvin deference between what Epic Sees And what colormeter sees so i rate the Epic on 5800

 

[Jeffrey Loewe]

Thanks Jeffry
I agree with you it seems that the problem comes from how this cameras process the materials Epic 16 Bit internal Processing Vs. Alexa 12 Bit ,
and how knows if Alexa shoot Row on codex maybe the result would be better , thats some thing need's to be tested

No worries!

Yeah looking at it again and thinking back to some of my projects, it could be camera related. I've done a test with a 7D vs an EX1 and an EX1 vs an HVX. And with the EX1 I got alot of transparency issues in the footage, not sure why. The HVX shot the cleanest. These cameras are far below the Alexa and Epic but particular cameras interpret footage for keying much better than others it seems.

I mean it could come in post but I've come across the same issue from having different cameras shooting alternate angles and we couldn't make it match. I think with some further testing and research you could probably get the Alexa to be closer to matching footage wise with the Epic, but its pretty incredible for the almost same set up you had the Epic blew the Alexa away. Its insanely clean particularly around the Hair of the individual and even on the Glasses/Coffee mug where those objects will usually grab a slight green reflection from the screen and cause a transparency like you saw with the Alexa grabs.

It looks as if RED did some rigorous testing in this area for it to shoot that well in the same situation as its primary competitor.

 

[Esmaeil Neisi]

it's not a mater of workflow and all this thing done in 16 Bit, you guys correct me if i'm rung but ProRess compression and bit depth Is some thing Critical for pulling good green key out , for sure you can have better key out of alexa but you need more work thats what i want to say

 

[Esmaeil Neisi]

No worries!

Yeah looking at it again and thinking back to some of my projects, it could be camera related. I've done a test with a 7D vs an EX1 and an EX1 vs an HVX. And with the EX1 I got alot of transparency issues in the footage, not sure why. The HVX shot the cleanest. These cameras are far below the Alexa and Epic but particular cameras interpret footage for keying much better than others it seems.

I mean it could come in post but I've come across the same issue from having different cameras shooting alternate angles and we couldn't make it match. I think with some further testing and research you could probably get the Alexa to be closer to matching footage wise with the Epic, but its pretty incredible for the almost same set up you had the Epic blew the Alexa away. Its insanely clean particularly around the Hair of the individual and even on the Glasses/Coffee mug where those objects will usually grab a slight green reflection from the screen and cause a transparency like you saw with the Alexa grabs.

It looks as if RED did some rigorous testing in this area for it to shoot that well in the same situation as its primary competitor.

Agree

 

[Timur Civan]

Some about this seems off.

There is NO WAY the Alexa's performance is that poor on a green screen. I've had HVX200 footage key better than that.

 

[Esmaeil Neisi]

According to logC if you apply a Rec 709 to it for sure you will see better result but the point is the tones that a keyer couldn't find properly in Alexa footage