Welcome to our community

Be a part of something great, join today!

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 :)

Epic Vs Alexa Green Screen Test

Esmaile Neissi said:
Eki ,
for best result the Alexa DIT said to record logC
Click to expand...

He's right - that does get the best result. But you need to go from LogC -> normal gamma in After Effects. Also, you need to color balance out the green tone.

If you want something to work right out of the box, use 709 mode in Alexa.

This has nothing to do with 16bit sensors or anything like that.

Just the Alexa is not being decoded properly.

If you use Epic and you output in XYZ color mode, with RedLogFilm tones, with Tint at -40, you would not get a good result with Epic out of the box.

Just want to help.

Bruce Allen
www.boacinema.com
 
For sure the log to rec709 yeld beter result for eliminate confusing i will post a version with corrected gamma but still you can see the defrence between the 16 Bit and 12 bit tones
 

Attachments

  • Alexa01_00047.jpg
    Alexa01_00047.jpg
    91 KB · Views: 0
Actually a lot of people here are sort of missing the exact point of the test. (Some of you do get it, and have made useful suggestions!)

The cameras were shot as close to the same as practical, and then these are the results of default settings for the keyer.

You can in fact pull much better keys with Alexa of course. For that matter you can pull better keys with an XL-1 or any of the much reviled DSLR cameras. But not at default settings.

Why does this matter?

The short answer is that it barely matters at all.

The Epic is delivering a good result without touching the key controls ... its suitable for live keying, and very fast turnarounds with lots of shots. The default settings are giving a result that for many projects can be used as final. (Not all projects of course!)

With all these other cameras the compositor needs to take time and fiddle settings in post to get a good basic key.

So ... this was a pretty straightforward key test. With some gamma changes and a uniform tweak Alexa will give much better results than we've seen here. That's two steps for each shot though that are not required for Epic footage.

Think about a more difficult key, for example Saruman or Gandalf in white robes and reflective swords/props on green and blue screens, with a wind machine going. Where do you want compositor time spent when you are paying for it?

The same can be said of other systems. If they had shot a 7D that day alongside default settings give better results than the defaults on Alexa. There is a flip side though .... You can only clean up the key so much on the 7D before you have to start scoping. You can clean up the Alexa to within a hair of perfection even on the most demanding shots - just like Red and Film.

Still, this is a flaw in the Alexa system ... a very minor flaw, but its there, and you must plan for it.

Alexa delivers results more like Epic and R1 when you use ARRIRAW recording, albeit in 2K.

What the reporting of this test does not account for is any decoding settings in the Epic footage. Was that the default camera look merely scaled and transcoded to ProRes, or was there any correction done? If there was correction ... you are not seeing a true comparison.

So ... my experience?

Alexa keys fine, and current workflows expect that you'll have to do "something" to pretty much any footage entering the pipeline or that you will spend more time on the key.
 
Last edited:
Thanks For Getting My Point Alexander,
Thats Exactly what i meant from this test,
BTW no Correction on Epic footage St rat From Redcinex to tiff then Scale in AE to 1920x1080
 
Hi Esmalle,

Thanks for doing this test. One question, would it have been more fair if you left the Epic in RedLogFilm so it would be in the same colorspace as the Alexa? To say there was no correction on the Epic or Alexa isn't quite fair if you left the Alexa in SLog and applied REDcolor2 to the Epic. As it shoots in RAW you are forced to apply an initial color space to it.

Or the opposite, Epic in REDcolor2 and Alexa in REC709. Still not correction applied, just in the same space.

Still, very glad you did this test and please don't let the comments frustrate you. Like you, we all want to get to the truth, and there is some good advice coming forward.

Cheers!

Jim Hare
Sydney Australia
 
I have worked alot with Alexa and it isn't bad with green screen, I feel this test isn't very well put together.
 
What you're saying isn't entirely true. I love the Red, but there is absolutely no reason to fake bad results for the Alexa in a test like this.

Saying that you're keying Alexa in Log and Red in Linear because ARRI recommends you shoot in LogC is just dishonest because Red recommends you choose RedLogFilm which is pretty much the same thing. It isn't an extra step to convert LogC to linear compared to Red - it's exactly the same. The Alexa 709 key should be shown next to the Epic Linear key to be fair.

Second, this isn't shot as close to the same as practical - they use different lenses. One on a prime lens the other on a zoom. Aside from lenses having different color rendition properties the difference in sharpness between a prime and a zoom can mean the difference between the keyer seeing a hair and a circle of confusion.

I love Red but this is a poorly done test that is essentially meaningless. I'd do Gunleik a favor and take his name out of your original post so it doesn't seem like he's involved in this kind of "test".

Alexander Ibrahim said:
Actually a lot of people here are sort of missing the exact point of the test. (Some of you do get it, and have made useful suggestions!)

The cameras were shot as close to the same as practical, and then these are the results of default settings for the keyer.

You can in fact pull much better keys with Alexa of course. For that matter you can pull better keys with an XL-1 or any of the much reviled DSLR cameras. But not at default settings.

Why does this matter?

The short answer is that it barely matters at all.

The Epic is delivering a good result without touching the key controls ... its suitable for live keying, and very fast turnarounds with lots of shots. The default settings are giving a result that for many projects can be used as final. (Not all projects of course!)

With all these other cameras the compositor needs to take time and fiddle settings in post to get a good basic key.

So ... this was a pretty straightforward key test. With some gamma changes and a uniform tweak Alexa will give much better results than we've seen here. That's two steps for each shot though that are not required for Epic footage.

Think about a more difficult key, for example Saruman or Gandalf in white robes and reflective swords/props on green and blue screens, with a wind machine going. Where do you want compositor time spent when you are paying for it?

The same can be said of other systems. If they had shot a 7D that day alongside default settings give better results than the defaults on Alexa. There is a flip side though .... You can only clean up the key so much on the 7D before you have to start scoping. You can clean up the Alexa to within a hair of perfection even on the most demanding shots - just like Red and Film.

Still, this is a flaw in the Alexa system ... a very minor flaw, but its there, and you must plan for it.

Alexa delivers results more like Epic and R1 when you use ARRIRAW recording, albeit in 2K.

What the reporting of this test does not account for is any decoding settings in the Epic footage. Was that the default camera look merely scaled and transcoded to ProRes, or was there any correction done? If there was correction ... you are not seeing a true comparison.

So ... my experience?

Alexa keys fine, and current workflows expect that you'll have to do "something" to pretty much any footage entering the pipeline or that you will spend more time on the key.
Click to expand...
 
Jake you seems too angry about this test ,
First of All , we didn't do this test intentionally to not be fair , the reason we key that way because of the recommendation of the Alexa DIT guy so its his fault not me
Second , I post another version with Rec709 and you can take a look at it , you are exactlly dont get the meaning of this test its not related to Circle of confusion or resolution, Circle of confusion is related to the size of pixel and size of sensor and stuff like that please don't miss leading people with your worlds
We discus the Bit depth and i can guaranty you if we even use the same UP on alexa the result will be like the one i post hear
 
About Gunlik ,his a well respected pro guy and he is not need you to defend him & this Test not related to him , I just wanted to put Some blurry clean footage az a background and that sample was a good one that's it.
you better find your self a another place to say your beautiful worlds.
 
Hi

Esmaile
Thanks for adding me to your test! :-)

This conversation is getting a bit out of hand, maybe.

Interesting test. Thanks.
Thanks for the kind words! :-)

Fredrik was only pulling a joke. Mostly on me. That's fine.
The studio is actually Fredriks workplace.

No hostile vibes at all.

As to gamma-discussion and all that jazz.

it is one of the interesting ones.

I have shot a greenscreen piece on the Epic with tungsten recently. seems to key just fine in both REC and log.
Some software is optimized for one, others for the other.
AFAIK when f. ex. nuke goes "linear" from a log image, it maps the data linearily internally before processing. Not like "linear" rec, which is basically an image compensated to look good for the eye, not neccesarily the math... :-)

Keep it up.

All these comparisons and obsercations are interesting.
And it's fun to finally imagine myself and Fredriks workplace in kuwait.
I could do worse.

If you want to host some of these frames somewhere so that our geeky community can do theur own developments and pull their own keys, just tell me...

Cheers!

Gunleik
 
Jake Bastian said:
What you're saying isn't entirely true. I love the Red, but there is absolutely no reason to fake bad results for the Alexa in a test like this.
Click to expand...


You quoted me and said what I was saying wasn't true. Which part exactly? Because you didn't reply to anything I said.

Here are the key points: (colored for separation)

Alexander Ibrahim said:
Why does this matter?

The short answer is that it barely matters at all.

Still, this is a flaw in the Alexa system ... a very minor flaw, but its there, and you must plan for it.

So ... my experience?

Alexa keys fine, and current workflows expect that you'll have to do "something" to pretty much any footage entering the pipeline or that you will spend more time on the key.
Click to expand...

So, exactly why are you accusing me of faking bad results? First off I didn't do the test. Second ... there aren't bad results in this test! Aside from the obvious colorspace issue (A user error, which Esmaile fixed with some later images), Alexa is showing rather well here!

What is so egregious about my comments about Alexa and this test that you singled me out for criticism? Are you complaining just because I see a very minor advantage for Epic, one I said "barely matters at all?"
 
Hey gunleik, Nice to see you hear and thanks sorry to use your plates dont mean any thing just because i'm a big fan of Anamorphics and have a set of them.
Fredrik not Joking you are one of perfect guys in this forum for sure and i personally learn a lot from your Tests And for sure will send you some of my tests to check them out on your special way of Analiese.
Would be pleasure to see you guys hear how knows maybe in the future will happen.
 
Sorry Alexander, the only part of your message I was referring to was that they were shot as close as practical (which I did reply to). The rest of the message was directed at the poster. I also didn't mean to imply that the results were faked - more agreeing with you that they are essentially meaningless because they are shot with different lenses and using different gamma settings. Apologies if I offended you.

But I stick by my original post that this isn't useful or interesting as a camera comparison test and would be better framed as a gamma comparison one for chroma keyers.

Alexander Ibrahim said:
You quoted me and said what I was saying wasn't true. Which part exactly? Because you didn't reply to anything I said.

Here are the key points: (colored for separation)



So, exactly why are you accusing me of faking bad results? First off I didn't do the test. Second ... there aren't bad results in this test! Aside from the obvious colorspace issue (A user error, which Esmaile fixed with some later images), Alexa is showing rather well here!

What is so egregious about my comments about Alexa and this test that you singled me out for criticism? Are you complaining just because I see a very minor advantage for Epic, one I said "barely matters at all?"
Click to expand...
 
You must log in or register to reply here.
Back
Top