First Day with Dragon

[Matt Ryan]

copy. Now, only if they would send that e-mail.

As an example here's something I just shot on the old olpf Dragon using available light only. No noise reduction, no manipulation, just moved ISO slider to 2000ISO from 320. So go into your shoots giving the sensor as much light as possible for the low end and you can really push it.

 

[Robert Ruffo New]

What few people ever talk about is that noise reduction will work much better on a higher resolution source. The more pixels you have, the more you can apply noise cleaning in post and while not risking a plastic effect. So shadow rendition AFTER noise reduction will be very different between Arri and Dragon.

RedGamma 4 and Dragon Gamma (4) are also very aggressive in terms of shadow crush, which is why the 1600 ISO only looks as bright as Arri at 800 - you are seeing differences in the development curves not the recorded data. Using RedlogFilm you would have very similar results to Arri in Log at same ISOs.

 

[Doug Beatty]

Noise level is different from exposure level. This shot was exposed for the highlights outside, not for lowlight. If you shoot for low levels and open up you will not get this amount of noise because you are not closing down to expose for the highlights. Does that make sense? You can not look at this shot and say well the Dragon isn't good in low light. Yes, it is worth the upgrade cost by far.

For the sake of a very specific argument, let's say you are shooting a moody scene at 800 and a 2.8/4 split on Alexa and you have exposure on a face, key side. The window curtain that illuminate the face is 8 stops over and fill side falls off into shadow using a simple hard side bounce out of frame.

On Dragon, to maintain your exposure, you either need to add another soft source for fill or bump up to 1600 ISO because you are on standard speeds and cannot open up to a 2/2.8 split

1) I'd be interested to see how the window performs on Alexa vs. Dragon. My inclination is that you could pull down the highlights and get details on the fabric of the curtains if you wanted.

2) Alexa would have less noise on the fill side detail right out of the camera.

I didn't know Alexa has a full stop more sensitivity in ISO when comparing to Dragon, so thanks for educating me on that.

As an example here's something I just shot on the old olpf Dragon using available light only. No noise reduction, no manipulation, just moved ISO slider to 2000ISO from 320. So go into your shoots giving the sensor as much light as possible for the low end and you can really push it.

This makes a ton more sense. Goalposts are your friend, and Dragon footage really seems to be the most elastic on the planet when you play between the lines.

 

[Mike P.]

You know what though Matt, the exact same logic has always been the case for MX as well (actually any sensor, really). When I shoot at T1.3-T1.5, MX stills ends up being clean at ISO1600... Truth is, you're just saturating the sensor with enough photons when you're that wide open... It's like pushing day time exteriors to 3200ISO -- it still looks clean.

In my own tests I've come to the conclusion that Dragon is "cleaner" but more so because of the downscaling benefits going from 6k to 4k or 1080p (as opposed to being 4k at 4k or 4k to 1080p on my Scarlet).

 

[Doug Beatty]

You know what though Matt, the exact same logic has always been the case for MX as well (actually any sensor, really). When I shoot at T1.3-T1.5, MX stills ends up being clean at ISO1600... Truth is, you're just saturating the sensor with enough photons when you're that wide open... It's like pushing day time exteriors to 3200ISO -- it still looks clean.

In my own tests I've come to the conclusion that Dragon is "cleaner" but more so because of the downscaling benefits going from 6k to 4k or 1080p (as opposed to being 4k at 4k or 4k to 1080p on my Scarlet).

Another good point, I'd rather not shoot at 6K just to be able to down scale the noise on high ISO footage for cleaner UHD or 1080p output.

The penalty you pay for such high resolution on a relatively small sensor is the fact that each photon is less light sensitive and I think that's why Alexa at 800 looks better in low light scenes than Dragon at 800.

I have no doubt Red could make a lower res sensor (if they wanted to) that would allow you to get clean exposure in the dead of night for a 2K output, but they are staking their claim that resolution and dynamic range trump low light performance since there are so many tools readily available in post to process out said noise.

 

[Matt Ryan]

You know what though Matt, the exact same logic has always been the case for MX as well (actually any sensor, really). When I shoot at T1.3-T1.5, MX stills ends up being clean at ISO1600... Truth is, you're just saturating the sensor with enough photons when you're that wide open... It's like pushing day time exteriors to 3200ISO -- it still looks clean.

In my own tests I've come to the conclusion that Dragon is "cleaner" but more so because of the downscaling benefits going from 6k to 4k or 1080p (as opposed to being 4k at 4k or 4k to 1080p on my Scarlet).

Have you shot Dragon and MX outside at night?

I have never gotten clean MX results pushed to 1600ISO in low light let alone 4000ISO like I have with Dragon (and at such a low color temp)... I wouldn't have even gotten super clean images at 800ISO for the most part unless I crushed the blacks completely.

Even in low light you can open up your exposure to let in more light and have the ability to push it further in post whether by opening a stop or augmenting with light. If you do not take that precaution then you probably shouldn't be shooting with a digital cinema camera. My advice is to always shoot in low light with a lower ISO to force yourself to get a better exposure. Shooting at higher ISO's leads you into that false exposure area where you think you have good exposure in monitor but realize you not only have noise but in reality you are underexposed. A lot of RED shooters make that mistake and later claim of noisy footage. The Dragon to me is not only cleaner, it is far better at handling low light than MX. Thats not even mentioning better color response and dynamic range. Once you shoot Dragon it's hard to go back.

 

[Björn Benckert]

Another good point, I'd rather not shoot at 6K just to be able to down scale the noise on high ISO footage for cleaner UHD or 1080p output.

The penalty you pay for such high resolution on a relatively small sensor is the fact that each photon is less light sensitive and I think that's why Alexa at 800 looks better in low light scenes than Dragon at 800.

I have no doubt Red could make a lower res sensor (if they wanted to) that would allow you to get clean exposure in the dead of night for a 2K output, but they are staking their claim that resolution and dynamic range trump low light performance since there are so many tools readily available in post to process out said noise.

I compared alexa and dragon quite a lot to me it's pretty obvious that the dragon sensor gives me more to play with at the same stop and exposure time. Then on top of that dragon got the trump of having HDRx. Low light is one thing, but when shooting in the dark then usually it's the contrast of the scene that is the problem you toast your highlights if you expose for the dark areas or loose your low end if you expose for the highlights.... Then, to me, the HDRx feature makes dragon quite incomparable to all other cameras on the market.

And why do we even talk iso, iso is post capture gain no matter if it's set in camera or not. The only thing that matters is how far the image can be pushed and still look good in it's mastring format. We shot with alexa cameras for the last 2 months as our dragon been in for service and I can only say I'm really looking forward to have the dragon back.

 

[Jacobo Martinez]

Matt Ryan, when you select any iso on any camera, let say, iso 800, shouldn't all cameras, if we would put a grey card in front, measure the same in a waveform?? Iso is a measurement of sensitivity, should 800 be always the same? I know they can vary a small amount like ire 44-50 but 1 stop seems too much.
If for example, I use an incident meter and i get an exposure of f4, if I shoot with a 5d, alexa, c500, shouldn't they appear almost the same exposure? or this is something unique to dragon? or is it alexa that is 1 stop over? I am really confused.
I mean, even if we exposed dragon at iso320, it would always be one stop underexposed. that means that if you would like to expose at 320, you would put you incident meter at iso 200 to compensate. is this true?

I usually shoot at iso 320-400. Really dark scenes at 250 without clipping. this is with epic MX.

 

[Matt Ryan]

Matt Ryan, when you select any iso on any camera, let say, iso 800, shouldn't all cameras, if we would put a grey card in front, measure the same in a waveform?? Iso is a measurement of sensitivity, should 800 be always the same? I know they can vary a small amount like ire 44-50 but 1 stop seems too much.
If for example, I use an incident meter and i get an exposure of f4, if I shoot with a 5d, alexa, c500, shouldn't they appear almost the same exposure? or this is something unique to dragon? or is it alexa that is 1 stop over? I am really confused.
I mean, even if we exposed dragon at iso320, it would always be one stop underexposed. that means that if you would like to expose at 320, you would put you incident meter at iso 200 to compensate. is this true?

I usually shoot at iso 320-400. Really dark scenes at 250 without clipping. this is with epic MX.

In theory they all should match, but how you view them, the color space and gamma space will make them all slightly different. The MX is the same when compared to the Alexa, about a stop less bright at the same ISO. You should always calibrate your meter to each individual camera because they all vary slightly. Check out the OffHollywood test to back up this idea that Dragon/MX is a stop less than Alexa at equivalent ISO (appears it has been removed from Reduser). Anyways, this is more apparent when you shoot the two side by side.

 

[luigivaltulini]

I compared alexa and dragon quite a lot to me it's pretty obvious that the dragon sensor gives me more to play with at the same stop and exposure time. Then on top of that dragon got the trump of having HDRx. Low light is one thing, but when shooting in the dark then usually it's the contrast of the scene that is the problem you toast your highlights if you expose for the dark areas or loose your low end if you expose for the highlights.... Then, to me, the HDRx feature makes dragon quite incomparable to all other cameras on the market.

And why do we even talk iso, iso is post capture gain no matter if it's set in camera or not. The only thing that matters is how far the image can be pushed and still look good in it's mastring format. We shot with alexa cameras for the last 2 months as our dragon been in for service and I can only say I'm really looking forward to have the dragon back.

I have the same thoughts

No grade only i have applied Kodak Lut

[/URL][/IMG]

[/URL][/IMG]

 

[Jacobo Martinez]

Matt, I imported 1 frame of the ARRI raw and 1 frame of the dragon in to premiere CC to do a side by side comparison. I turned both to LOG and this way with a waveform, at 640 match perfectly, I think it is the LUT that is affecting perceived inconsistant exposures. please correct me if i am wrong.
One mayor difference I saw is that the dragon is clipping, alexa is not.

 

[Matt Ryan]

They are close but Alexa's LogC is slightly less contrasty than Redlogfilm (Also look at the left side of the waveform on both.). The levels might be very close, but judging waveform the low end on Dragon is noisier and it clips a tiny bit compared to Alexa in this shot. That's not to say it's the case for every shot, but this specific example.:

 

[Matt Ryan]

Just so we can be 100% clear is this shot with the old or new olpf/calibration?

 

[paul schefz]

i am trying to follow this discussion but somehow....not sure but alexa seems to show almost a stop more in the shadows looking at this specific test and exposure....
i also downloaded Matt's super speed dragon clip and honestly can't believe how much noise there is.....at iso 320?
not really saying it is not "better" then epic but i guess the question is how "bad" is epic? because mine really isn't....i don't have that much noise, and i usually shoot at 800...yes, my highlight clip (since i always try to protect the shadows) but it really depends on the lenses quite a bit how "bad" the clipping looks....

i am still waiting for my email to send my epic in for dragon upgrade....do i expect it to improve an already very capable camera? yes...10K worth?

i really, really, really wish RED would post some official dragon footage for everybody to download and play with....for one it would really help to work out workflow issues or what to expect in terms of performance from an existing workstation but really it would also help actually judging the image....everytime anyone puts up a test it turns into a 50 page discussion on why the test is flawed and why dragon is of course WAY more awesome then this specific test shows....the worst thing about all this is that nobody in their right mind would want to put up some test or footage because of the aftermath....

 

[Matt Ryan]

i am trying to follow this discussion but somehow....not sure but alexa seems to show almost a stop more in the shadows looking at this specific test and exposure....
i also downloaded Matt's super speed dragon clip and honestly can't believe how much noise there is.....at iso 320?
not really saying it is not "better" then epic but i guess the question is how "bad" is epic? because mine really isn't....i don't have that much noise, and i usually shoot at 800...yes, my highlight clip (since i always try to protect the shadows) but it really depends on the lenses quite a bit how "bad" the clipping looks....

i am still waiting for my email to send my epic in for dragon upgrade....do i expect it to improve an already very capable camera? yes...10K worth?

i really, really, really wish RED would post some official dragon footage for everybody to download and play with....for one it would really help to work out workflow issues or what to expect in terms of performance from an existing workstation but really it would also help actually judging the image....everytime anyone puts up a test it turns into a 50 page discussion on why the test is flawed and why dragon is of course WAY more awesome then this specific test shows....the worst thing about all this is that nobody in their right mind would want to put up some test or footage because of the aftermath....

I'm curious what clips you are referring to and what you are viewing the files on? On a calibrated display in almost every one of those shots there is no noise and unless you were there in those light levels it's hard to understand this was as the sun went down behind the hills...

 

[paul schefz]

I'm curious what clips you are referring to and what you are viewing the files on? On a calibrated display in almost every one of those shots there is no noise and unless you were there in those light levels it's hard to understand this was as the sun went down behind the hills...

i am watching the clip from your dragon super speed thread...about a 7 sec clip....watching this on a calibrated 4k monitor in RCX...no settings changed....am attaching a screen shot this is full at 100%......at 320 i would expect less noise....


EDIT...impossible to see what i am seeing on my screen from this screenshot, sorry...

 

[Matt Ryan]

i am watching the clip from your dragon super speed thread...about a 7 sec clip....watching this on a calibrated 4k monitor in RCX...no settings changed....am attaching a screen shot this is full at 100%......at 320 i would expect less noise....


EDIT...impossible to see what i am seeing on my screen from this screenshot, sorry...

You can't really judge that shot though since exposure peaks at 60IRE. I wouldn't even judge an official RED R3D. Everyone shoots differently so it's not really relevant unless you do it your own way. If that makes sense. The early offhollywood fashion stuff I thought was pretty noisy, but using it myself it seems less noisy. But maybe it's my way of shooting/exposing.

In the OP's example the Alexa is for sure cleaner on the low end.

 

[paul schefz]

You can't really judge that shot though since exposure peaks at 60IRE. I wouldn't even judge an official RED R3D. Everyone shoots differently so it's not really relevant unless you do it your own way. If that makes sense. The early offhollywood fashion stuff I thought was pretty noisy, but using it myself it seems less noisy. But maybe it's my way of shooting/exposing.

In the OP's example the Alexa is for sure cleaner on the low end.

matt...i understand and obviously did not download the clip and post: "noise, noise everywhere!"
i am absolutely with you, everybody shoot differently to get out of the camera what they want and how they want it....which is one reason these tests or comparisons always end in 50 page discussions......
what i am still on the fence on is how much of an improvement dragon is and i guess that will be impossible to tell until i upgraded and then it is too late....knowing what we all know the 10k probably is a no brainer right now....but 10k can buy some nice glass or goes a long way towards a second body....
i guess i am still waiting/hoping for that clear cut...here it is....side by side this is why and where and how much dragon differs from epic....and i guess i will never see it...

regardless: i have no need for alexa at all even if it was the same price as dragon or even epic....so this comparison does little for me other then wake that nagging doubt again about how much better dragon is then epic...

 

[Daniel Soderberg]

Great sample Matt! You probably saw this coming, but what lens did you shoot those stills on?

As an example here's something I just shot on the old olpf Dragon using available light only. No noise reduction, no manipulation, just moved ISO slider to 2000ISO from 320. So go into your shoots giving the sensor as much light as possible for the low end and you can really push it.

 

[Matt Ryan]

Great sample Matt! You probably saw this coming, but what lens did you shoot those stills on?

My personal favorites, Zeiss Super Speeds MKII.