All the Epic/Dragon Vs. Alexa/Sony threads are belong to us...

[John Marchant]

I think its worth considering more carefully what 16 stops and change means - the last couple of stops we're all concerned about here are found at the bottom of the exposure. Bear in mind that every stop is a doubling (or halving) of brightness. So go with me on a little thought experiment: Take a fantastic camera like an Alexa. Expose for the darkest discernible detail, the proverbial black cat in a dark room... thats a pretty subtle image. Now take away 75% of the light in that scene. That's the faint sprinkling of photons we're interested in.

It does not surprise me at all that these stops are tough to reveal, however Dragon is ALL about subtlety and fidelity and the presence of this data, even if hard to discern initially, is one of the critical elements at play when working to stretch your material to look its best.

 

[David Battistella]

I think its worth considering more carefully what 16 stops and change means - the last couple of stops we're all concerned about here are found at the bottom of the exposure. Bear in mind that every stop is a doubling (or halving) of brightness. So go with me on a little thought experiment: Take a fantastic camera like an Alexa. Expose for the darkest discernible detail, the proverbial black cat in a dark room... thats a pretty subtle image. Now take away 75% of the light in that scene. That's the faint sprinkling of photons we're interested in.

It does not surprise me at all that these stops are tough to reveal, however Dragon is ALL about subtlety and fidelity and the presence of this data, even if hard to discern initially, is one of the critical elements at play when working to stretch your material to look its best.

This articulates what I was talking about when it comes to the lower stops and the noise floor in the shadows. Thanks!

Battistella

 

[James_Mills]

Slate says T4 1/2 on a 50mm UP, 5KHD, RC 3:1

Posted RLF, ISO 250

I have even more convincing ones not at hand on my laptop - this shot was done without leveraging the whole sensor area (this is actually quite a crop from the 5k, so a little grainy). The later tests I did, I built a proper light-tight hood for the Xyla-21, rather than just using a darkened room, and shot a variety of glass and f-stops, filling the 6k width with the chart. I curtailed these further test though when last year it became apparent that more significant changes were coming, and I didn't want to blow a significant effort testing something which would shortly be redundant

So because you got these at 250 does it in turn disprove these other theories about needing to expose for 2000?

And if you got this without post effects, would the Alexa measure more on this same DSC scale? if so, why doesn't Alexa claim to have 17.3 stops of dynamic range?

I remember these tests, but I'm confused as to why they're not reflecting the results of this shootout.

 

[Kemalettin Sert]

from todays non scientific mx vs carbon dragon DR test
Dragon https://www.dropbox.com/s/2aks1ao7yh0furj/A001_C002_0218UB.0000618F.R3D
MX https://www.dropbox.com/s/6wqw2wbzwa8quny/B001_C001_021893.0000000F.R3D
also there is no IR on both cameras.you can see IR pollution on MX chip.

 

[Matt Ryan]

from todays non scientific mx vs carbon dragon DR test
Dragon https://www.dropbox.com/s/2aks1ao7yh0furj/A001_C002_0218UB.0000618F.R3D
MX https://www.dropbox.com/s/6wqw2wbzwa8quny/B001_C001_021893.0000000F.R3D
also there is no IR on both cameras.you can see IR pollution on MX chip.

Thanks Kemalettin!

 

[James_Mills]

from todays non scientific mx vs carbon dragon DR test
Dragon https://www.dropbox.com/s/2aks1ao7yh0furj/A001_C002_0218UB.0000618F.R3D
MX https://www.dropbox.com/s/6wqw2wbzwa8quny/B001_C001_021893.0000000F.R3D
also there is no IR on both cameras.you can see IR pollution on MX chip.

I wonder what caused the green tint in the Dragon file. Is this the same one we were worried about earlier in the thread?

Good test btw

 

[Mike P.]

I guess what I'm saying is if we have 16.3 stops, I'd like for it to be apparent at all times of capture if I am giving the shot that much range...*etc*

This is pretty much what I was saying, but jumping to the idea that, if exposed for ISO2000, would all stops be apparent because you'd have (roughly) equal under and over exposure (which is what Alexa does at ISO800). The 16 stops is there all the time, but when you're like ~11 stops under mid-grey, there's no way you'd dig out those stops without throwing your mid-tones completely out of wack and/or having more noise than necessary and/or clipping the highs earlier than necessary.

from todays non scientific mx vs carbon dragon DR test
Dragon https://www.dropbox.com/s/2aks1ao7yh0furj/A001_C002_0218UB.0000618F.R3D
MX https://www.dropbox.com/s/6wqw2wbzwa8quny/B001_C001_021893.0000000F.R3D
also there is no IR on both cameras.you can see IR pollution on MX chip.

This is a great example, but having an Alexa example would have been even more telling. Given the DR of the scene, I bet both Alexa and Dragon would have DR to spare, but given Alexa's over/under balance, it'd be interesting to see if it looked any better.

On the flip side, I think there's going to be a lot of arguments comparing DR vs. Latitude moving forward, and that confusion is going to prevent people from picking up a Dragon (in the same way people shied away from MX.) Again, it looks like we're having to futz around with the r3d files to make them "look as good Alexa", which is something current Alexa users will *not* switch platforms for (e.g. having to do more in post to get a comparable image).

 

[Evin Grant]

I guess what I'm saying is if we have 16.3 stops, I'd like for it to be apparent at all times of capture if I am giving the shot that much range. I shouldn't have to dig it out in post, it should be there in the RC3,RG3 shot. Does that make sense?

Matt, try to remember that RG3/RC3 were designed for a different sensor and dynamic range, is it really so surprising that a curve designed for 13.5 stops is having a hard time displaying 16.5?
Don't get your panties in a twist about the Alexa either, those that will come out guns blazing for Alexa wouldn't consider the Dragon even if it had twice the DR! Also I've shot a ton of Alexa footage and that last stop in the highlights is all roll off, it's pretty but there is really no useable detail there. I'm not sure if this has to do with the sensor low-con or software but it's not useful in the same way that the top stop is even on an MX. On the Dragon you have it both ways, you can tweak the curve to roll off smoothly into clip or if you've controlled your exposure you can have detail right to the top of the range. This is not something you can really see on charts, but you feel it in Dragon footage immediately. As to weather you need to tweak in post to get the most out of it, I think this is a non-issue. You have to do the same in Arri RAW and you have to at least add a LUT to view LogC Prores so how is an Alexa any easier?

 

[Matt Ryan]

Matt, try to remember that RG3/RC3 were designed for a different sensor and dynamic range, is it really so surprising that a curve designed for 13.5 stops is having a hard time displaying 16.5?
Don't get your panties in a twist about the Alexa either, those that will come out guns blazing for Alexa wouldn't consider the Dragon even if it had twice the DR! Also I've shot a ton of Alexa footage and that last stop in the highlights is all roll off, it's pretty but there is really no useable detail there. I'm not sure if this has to do with the sensor low-con or software but it's not useful in the same way that the top stop is even on an MX. On the Dragon you have it both ways, you can tweak the curve to roll off smoothly into clip or if you've controlled your exposure you can have detail right to the top of the range. This is not something you can really see on charts, but you feel it in Dragon footage immediately. As to weather you need to tweak in post to get the most out of it, I think this is a non-issue. You have to do the same in Arri RAW and you have to at least add a LUT to view LogC Prores so how is an Alexa any easier?

The RC3,RG3 has supposedly been tweaked for Dragon. So theoretically it has been calibrated for the 16.3 stops correct? Hence the sensor recalibrations and color tweaks etc.

The difference is when you drop the rec709 lut which is standard in most coloring programs, you still get a great, contrasty image, with lots of DR without having to "dig it out." I'm playing devils advocate here because I want to continue pushing for more vs. the camera is great now mentality. The other thing is how much longer till we get Dragoncolor/Gamma? The camera has been sold and shot with for a while now. I wish it were here now for tests like these...

Don't take it the wrong way, I'm not discounting the Dragons achievements or it's superiority. My statement is that if we could see the 16 stops on set and immediately in post without having to "dig" (or in film terms push and pull), there would be no debate from anyone whether Dragon has more DR than Alexa...

 

[Kemalettin Sert]

This is pretty much what I was saying, but jumping to the idea that, if exposed for ISO2000, would all stops be apparent because you'd have (roughly) equal under and over exposure (which is what Alexa does at ISO800). The 16 stops is there all the time, but when you're like ~11 stops under mid-grey, there's no way you'd dig out those stops without throwing your mid-tones completely out of wack and/or having more noise than necessary and/or clipping the highs earlier than necessary.



This is a great example, but having an Alexa example would have been even more telling. Given the DR of the scene, I bet both Alexa and Dragon would have DR to spare, but given Alexa's over/under balance, it'd be interesting to see if it looked any better.

On the flip side, I think there's going to be a lot of arguments comparing DR vs. Latitude moving forward, and that confusion is going to prevent people from picking up a Dragon (in the same way people shied away from MX.) Again, it looks like we're having to futz around with the r3d files to make them "look as good Alexa", which is something current Alexa users will *not* switch platforms for (e.g. having to do more in post to get a comparable image).

Alexa is gonna be on job until next week. i ll do same test with Alexa when i have the camera.

I wonder what caused the green tint in the Dragon file. Is this the same one we were worried about earlier in the thread?

Good test btw

Btw its not new OLPF.

 

[James_Mills]

Btw its not new OLPF.

Yeah, but people claimed in this thread earlier that it wasn't merely an old OLPF issue. Unfortunately this is another nebulous concern that will need sorting once the new OLPF is standard.

Alexa is gonna be on job until next week. i ll do same test with Alexa when i have the camera.

I'm excited to see this, thanks.

 

[Medavoym]

I honestly see a lot of dancing around the issue at hand.

Questions are answered with more questions - it's a bit annoying.

Can I ask a straight question? I need a straight answer. Either from Phil, Geoff, Gunleik etc - knowledgeable people who intimately know both cameras. So...

Based on these tests... and your personal experience with the two cameras... WHAT CAMERA HAS MORE DYNAMIC RANGE? EPIC DRAGON or ALEXA?

Clear answer please. No dancing around. We're not talking subjective art here... we are talking hard facts.

DRAGON or ALEXA?

Thanks.

 

[Mark Wilkinson]

We already got that answer. Alexa.

Now ask about which OLPF. Now ask about noise. Now ask about shoulder rolloff. Now ask about detail in the shadows. We're discussing the details of the issue.

No reason to be testy about it...

 

[Mark Wilkinson]

Shoulder

Shoulder

Any chance you shot this protecting the highlights? By 4 stops or something? It would be instructive to see if the dark detail held up since Geoff's example had so much in the toe and so little in the shoulder...

This is the Dragon image - take a look at the shoulder rolloff in the parade - pretty hard corner into clip.

This was RLF, BTW.

 

[Martin Stevens]

I honestly see a lot of dancing around the issue at hand.

Questions are answered with more questions - it's a bit annoying.

Can I ask a straight question? I need a straight answer. Either from Phil, Geoff, Gunleik etc - knowledgeable people who intimately know both cameras. So...

Based on these tests... and your personal experience with the two cameras... WHAT CAMERA HAS MORE DYNAMIC RANGE? EPIC DRAGON or ALEXA?

Clear answer please. No dancing around. We're not talking subjective art here... we are talking hard facts.

DRAGON or ALEXA?

Thanks.

The Alexa has the dynamic range of color film negative and the Dragon has MORE dynamic range than color film negative,
so therefore the Dragon has more DR than the Alexa.

This is not opinion, it is science fact.

 

[Martin Stevens]

Any chance you shot this protecting the highlights? By 4 stops or something? It would be instructive to see if the dark detail held up since Geoff's example had some much in the toe and so little in the shoulder...

This is the Dragon image - take a look at the shoulder rolloff in the parade - pretty hard corner into clip.

This was RLF, BTW.

I see it as a soft clip....but after the new OLPF is added the clip will be very soft indeed.

 

[Mark Wilkinson]

The Alexa has the dynamic range of color film negative and the Dragon has MORE dynamic range than color film negative,
so therefore the Dragon has more DR than the Alexa.

This is not opinion, it is science fact.

I'm sorry, Martin, but I don't think the tests in this string support that. Can you point us to some published results that do?

I think the undisputed facts are that the (old OLPF) Dragon has significantly less noise and that Alexa is capturing more detail in the highlights.

 

[David Battistella]

Clear answer please. No dancing around. We're not talking subjective art here... we are talking hard facts.

DRAGON or ALEXA?

Thanks.

Ha ha.

Asking for a straight answer on Reduser is like asking the government to do your taxes.

Battistella

 

[Matt Ryan]

I honestly see a lot of dancing around the issue at hand.

Questions are answered with more questions - it's a bit annoying.

Can I ask a straight question? I need a straight answer. Either from Phil, Geoff, Gunleik etc - knowledgeable people who intimately know both cameras. So...

Based on these tests... and your personal experience with the two cameras... WHAT CAMERA HAS MORE DYNAMIC RANGE? EPIC DRAGON or ALEXA?

Clear answer please. No dancing around. We're not talking subjective art here... we are talking hard facts.

DRAGON or ALEXA?

Thanks.

In that specific test, Alexa.

 

[David Battistella]

In that specific test, Alexa.

See, there is a qualifier in even the simplest answer.

Who's better? The Seahawks or the Broncos? Well on Super Bowl Sunday, the Seahawks.

battistella