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 :)

ACS Weapon test

Alexa XT vs. Carbon Fiber Weapon Test From Australia

Alexa XT vs. Carbon Fiber Weapon Test From Australia

Maybe I missed this before, but this from the Australian Cinematography Society and posted over on News Shooter.

"The ACS Technical Committee and FXPHD have done an “Introductory Video” looking at RED’s Carbon Fiber Weapon. The video looks at the new features, performance levels and where the new camera sits in the production ecosystem. Ben Allan ACS narrates the test video."

Here is the article. http://www.newsshooter.com/2016/01/...ted-by-the-australian-cinematography-society/

I feel like I have to be looking at this wrong. Does the Alexa really hold that much of an advantage over Weapon in the highlights?
 
They used standard olpf though not highlight
 
The only way to measure a C-mos sensor is to see where it clips and how many stops below that you can use. From looking at this test one can tell that the red cameras with the tested olpf clips 1 stop before alexa, but the red cameras also hold much more in the low end.

So in the test alexa clips at +6 stops in a similar manor as Weapon and dragon does at 5+ stops.

Then in the low end it´s much more difficult to define from looking at compressed vimeo feed but it´s quite clear that both red cameras with standard OLPF has about 2stop more detail in the blacks.

So from looking at this all I see is that dragon and weapon with standard olpf has about one more stop of DR than the Alexa. Not only that, the reds are also more light sensitive.

Now, one could argue that such great test results for RED could be package a bit differently... people that does not pay attention might miss the point and think Alexa reached the better results... To me the big confusion here is that they start from "rated iso". To reach greater understanding forget about rated iso instead talk about clipping point and acceptable noise.
 
Phil Holland said:
interesting perspective and I enjoy the back and forth discussion a lot. There's a lot I can agree with on those fronts.

One thing that is extremely time consuming to test is to check across the entire usable ISO range of over exposure and underexposure versus noise character. Examining the effect of ISO 250-2000 in particular on all three OLPFs is important and effects the results of tests like these. Good to do this with all of the OLPFs as well as that does effect the Highlight Roll-Off to Clip as well:




Another thing I didn't hear brought up is format size and resolution, which is understandably not the focus for everybody in the world. Arri Open Gate on the Alexa is around 7.5 megapixels at 3.4K versus 19.4 megapixels at 6K. If considering the S35 format you'd be comparing 2.8K versus 5K essentially as well. Makes a difference for some.

Totally agree about testing. It's long, labor intensive, and detail oriented work. I certainly need time between tests to shoot pretty things to get me back to normal.
Click to expand...

so what? As you can see there is at least 1 stop difference on highlights.Even skintone wouldnt matter much against Alexa.
 
Bjorn is correct in saying it is best to forget the rated ISO when it comes to considering dynamic range and think about clipping point to acceptable noise. In our test we tested the DR on a single shot and the cameras may behave differently with another shot but IMHO the Alexa has slightly less than a stop advantage on the Weapon but the comparisona of noise levels in different cameras is indeed rubbery. While the different OLPFs change sensitivity I am unconvinced that using another OLPF would substantially change the DR results. Please feel free to carry out specific testing of this and prove my OLPF hypothesis incorrect .

The dynamic range advantage of the Alexa is not huge and probably puts the Weapon over or equivalent to every other camera out there. In years past DR of electronic cameras has been a big issue as they were poor compared to film but at this level of cameras DR has got so good it's not such a primary concern. The ASC tests were not designed as an Alexa vs Weapon test. The Alexa's advantage in DR is a tick to the Alexa but there are many other advantages and considerations that come into play and either the Weapon or Alexa are fantastic choices. It's not productive to think as one as better than the other. Rather one camera may be better for a project for specific reasons. Frankly if you can't make sweet picture love with either of these cameras you are in the wrong game.
 
Bingo!

you have to look at ALL the stops and not just where it clips on the high end. the STH (i think) would probably give the dragon a half stop more in teh highlights, that is what it is meant for, but as you see th elow end suffers, as it does with the Alexa.

David



Björn Benckert said:
The only way to measure a C-mos sensor is to see where it clips and how many stops below that you can use. From looking at this test one can tell that the red cameras with the tested olpf clips 1 stop before alexa, but the red cameras also hold much more in the low end.

So in the test alexa clips at +6 stops in a similar manor as Weapon and dragon does at 5+ stops.

Then in the low end it´s much more difficult to define from looking at compressed vimeo feed but it´s quite clear that both red cameras with standard OLPF has about 2stop more detail in the blacks.

So from looking at this all I see is that dragon and weapon with standard olpf has about one more stop of DR than the Alexa. Not only that, the reds are also more light sensitive.

Now, one could argue that such great test results for RED could be package a bit differently... people that does not pay attention might miss the point and think Alexa reached the better results... To me the big confusion here is that they start from "rated iso". To reach greater understanding forget about rated iso instead talk about clipping point and acceptable noise.
Click to expand...
 
Kemalettin Sert said:
so what? As you can see there is at least 1 stop difference on highlights.Even skintone wouldnt matter much against Alexa.
Click to expand...

There's at least one stop more in the RED shadows...

The Alexa system places a lot of range in the shoulder.

What's going on behind the display gammas or even the log gammas* at the linear HDR data level, is more interesting/relevant.

*the log curves of RED, ARRI and Sony all differ slightly.
 
I can not tell how much longer the red stick is but I can tell its´s longer. Saying +5 alexa is burning as +4 weapon is possibly even a bit unfair to weapon. As I think weapon at +4 1/3 has about the same burn as alexa at +5.

Then in the lowend it´s even more difficult, judging noise etc. But we need to remember this test is without signal processing on the red side, on the alexa side what do we have?

No matter signal processing or not I see further into the dark on the red. Is it one stop or is it more? I could argue for two stops, some would say 1.5 but if you say less than one I say you are lying or need to look again... But remember this is with color science luts applied. One could argue that looking at the lowlights trough such filters is not really the correct way of doing things.

Then with that in mind. What camera got the longest stick?
And further more what works better in the dark? Again remember we are not talking lowlight olpf here, we are not talking ADD, we are not talking HDRx. Just standard olpf and nothing more.

Skintone olpf would bump the red stick to the right, Lowlight olpf would push it further to the left. ADD or other signal processing would clean up the lowlights and possibly open up for more latitude in that end.

So simply one could say that what is compared is the worst of weapon with the best of alexa. Still I judge the red stick to be about 1 stop longer. Please explain if you see anything different...

Regarding the difference in where in the DR range the 800 ISO is lands? Well 800 iso is only a recommendation from red and arri it´s not a mid point for any of those two cameras. So why measure form it? Especially as the latitude very much slides back and forth over that 800 line depending on olpf and signal processing you use.



24088454949_12f961de98_o.png
[/url]PastedGraphic-6 by Björn Benckert, on Flickr[/IMG]
 

Attachments

  • PastedGraphic-6.jpg
    PastedGraphic-6.jpg
    12 KB · Views: 0
Björn Benckert said:
I can not tell how much longer the red stick is but I can tell its´s longer. Saying +5 alexa is burning as +4 weapon is possibly even a bit unfair to weapon. As I think weapon at +4 1/3 has about the same burn as alexa at +5.

Then in the lowend it´s even more difficult, judging noise etc. But we need to remember this test is without signal processing on the red side, on the alexa side what do we have?

No matter signal processing or not I see further into the dark on the red. Is it one stop or is it more? I could argue for two stops, some would say 1.5 but if you say less than one I say you are lying or need to look again...

Then with that in mind. What camera got the longest stick?
And further more what works better in the dark? Again remember we are not talking lowlight olpf here, we are not talking ADD, we are not talking HDRx. Just standard olpf and nothing more.

Skintone olpf would bump the red stick to the right, Lowlight olpf would push it further to the left. ADD or other signal processing would clean up the lowlights and possibly open up for more latitude in that end.

So simply one could say that what is compared is the worst of weapon with the best of alexa. Still I judge the red stick to be about 1 stop longer. Please explain if you see anything different...

Regarding the difference in where in the DR 800 ISO is put? well that is only a recommendation from red and arri it´s not a mid point for any of those two cameras. So why measure form it and the latitude very much slides back and forth over that 800 line depending on olpf and signal processing.
Click to expand...

this is a good point Björn.
 
Brett Harrison said:
There's at least one stop more in the RED shadows...

The Alexa system places a lot of range in the shoulder.

What's going on behind the display gammas or even the log gammas* at the linear HDR data level, is more interesting/relevant.

*the log curves of RED, ARRI and Sony all differ slightly.
Click to expand...


Can not agree more, why are they even talking about "rolloff" as these sensors has no such thing. They all go linear to the clipping point. The only "rolloff" existing is done in color science and is up to the colourist, or who ever develop the raw files. So comparing the images when slided along different such curves does not mean much.

Posted this in the other thread about the same test. Possibly a moderator can fold the two threads into one, no?

PastedGraphic-6 by Björn Benckert, on Flickr
 
yep, i checked now normal exposure on this test.
The level of face on Alexa, be less exposed than a Weapon, and as showed Bjorn, Weapon , burning earlier in highlight.
All be very close in performance ..

test_zpsqlcyyr7o.jpg
[/URL][/IMG]
 
I still think the most interesting aspect of the test is how much quieter the noise is on the Weapon Dragon sensor in comparison to the Epic Dragon sensor. Interesting point you make Luigi that the Weapon and Alexa are very close in performance as this is exactly what the colourist observed.
 
Tom Gleeson said:
I still think the most interesting aspect of the test is how much quieter the noise is on the Weapon Dragon sensor in comparison to the Epic Dragon sensor. Interesting point you make Luigi that the Weapon and Alexa are very close in performance as this is exactly what the colourist observed.
Click to expand...


The sensor is the same.

So its the difference in teh bodies, but the sensor is the same.

Dragon sensor, Weapon body
Dragon sensor, Epic Body

Let's not further confuse things. If I send my Epic Dragon back to RED they are putting that sensor into the Weapon body. Let's be clear. The boards and electronics of the weapon offer some more performance, but sensor is the same.

Every ARRI camera has the same sensor since the first Alexa.

David
 
I think Tom means camera hardware when he says sensor. The higher bit rate that comes with weapon is something we all can relate to. we all know how almost double the data rate simply gives less noise. The additional improvements in boards etc will show when shooting in the same compression settings for both cameras. I have personally not done or seen such tests so I can only guess. But data rate should be a big one.

Also what I have not tested is shooting redlog10bit process... is it confirmed that the downscale is done pre / without the r3d compression. If so that should render some really clean blacks.
 
David Battistella said:
The sensor is the same.

So its the difference in teh bodies, but the sensor is the same.

Dragon sensor, Weapon body
Dragon sensor, Epic Body

Let's not further confuse things. If I send my Epic Dragon back to RED they are putting that sensor into the Weapon body. Let's be clear. The boards and electronics of the weapon offer some more performance, but sensor is the same.

Every ARRI camera has the same sensor since the first Alexa.

David
Click to expand...

Pretty sure they're not swapping your sensor. You get a new one.
 
David Litchfield said:
Pretty sure they're not swapping your sensor. You get a new one.
Click to expand...

I'm pretty sure you're incorrect, as in, they do swap the sensor and you do not get a new one.
 
You must log in or register to reply here.
Back
Top