OFFHOLLYWOOD DRAGON R3Ds & MORE

[Nick Morrison]

I really want to purchase an epic dragon this year. I have the money. I work with epic a lot. Alexa too. There is no doubt Arri is going to come out with a KICK ASS 4k camera. It's going to happen. BUT, it will be much more expensive than Dragon (i assume). And honestly, I think we will be sitting in exactly the same boat as we are now regarding Epic vs new 4k Alexa. But who cares, at this point, both will be spectacular image-wise, I can just afford to own an epic for a lot less money. That's what's cool...just thinking aloud...

I agree, visually...I'm sure the two may not be that different....however Dragon's small, gimbal/3D friendly size....and 6K resolution...are certainly two distinct features that hopefully give Dragon some significant staying power, no matter what Arri comes up with...

 

[Jake Wilganowski]

AND, if RED does come out with the iPad app that you can live grade on, THAT will be spectacular and is a BIG DEAL in my book, super cool, clients will love it too...

 

[Mark L. Pederson]

i have played with every single raw dragon clip i can get my hands on....and yes, the files are obviously larger and i also see better roll off and better compression....but when it comes to noise, shadows and ISO i am simply not blown away...at all....yes, things will get better and there is no doubt that dragon is an improvement over epic but i don't honestly see the leaps and heaps...

Hmmmmm ....

There are basically THREE ways to evaulate a camera IMO.

#1: Test it yourself. It's really important.

You will have things that you respond to - that you focus on and or care about more than someone else. What "looks the best" is a very, very subjective thing. I've helped facilitate countless tests since RED first shipped the RED ONE camera - some tests with extremely accomplished members of the ASC, etc. - and they ALL test slightly differently and look for and react to different things. Typically, they focus on the specific issues that concern them the most. You can learn some things from other people tests - and that's why I like to share files heer on REDUSER - BUT - no matter what the method of a camera test - given all the variables (lighting, subject, lenses, filters, post, etc.) - the best way to evaluate a camera is to test it yourself. The way you want to test it.

#2: Look at a wide body of work shot on that specific camera. This will help you know what the camera is capable of but not really a good way to compare one camera over another. And many times the talent makes the movie shine over the tool used - really talented cinematographers have made great looking films on much more limited, lower tech spec cameras than EPIC MX and ALEXA for exmaple.

Taking the example of EPIC MX vs. ARRI ALEXA - the two leaders in digital cinema acquisition -

Unless you have your head burried in the sand - there is now a significant number of very different films shot on EPIC MX and also on ALEXA that look fantastic. Even the biggest RED HATER in the world can not deny how good some features shot on EPIC MX look - and even the biggest RED FANBOY on the planet can not deny how good some films look that were shot on ALEXA.

So ... you have to admit that they are BOTH great cinema cameras. GREAT cameras. You might like one more than the other - or you may have more skin in the game with one over the other - or you might have had a bad experience with one of them - but these two cameras are clearly amazing tools that have both been used to create amazing films.

#3: Listen to the opinions of others, marketing, and whatever tests, reviews and sample files and tests that are released to the public.

Most everyone does #2 and #3 just based on their interest in a camera and/or quest for knowledge. And obvously not everyone has the opportunity to test a specific camera - but I think it's really important to recognize the value of testing a camera youself and trying not to "make up your mind" based on any material anyone (including me) chooses to share.

when it comes to noise, shadows and ISO i am simply not blown away

All three are clearly better on Dragon than the MX and MX wasn't such a bad camera (did I just put MX in "past tense "). Are you blown away by the ISO and noise on an Alexa? I think there may be a unlealistic expectation or the desire for something extremely "dramatic". I've seen plenty of noise on MX, Alexa, F55 and F65. I have also seen black blacks on material shot on all 4 of those cameras.

There are less than a handful of poeple in the world that have worked with - from camera to post - the quantity of RED footage that I have. It's what I have been doing since RED first shipped the RED ONE. And for me - it's "leaps and heaps" better. Trust me - after you rock a Dragon for a very short time - you won't even want to pick up your MX. You will see.

i would still love to see a good side by side epic dragon comparison but somehow i don't think we will see one too soon here.....

What is leaps and heaps for you Paul? Obviously I can't answer that - BUT ... I would like to know EXACTLY what you would like to see posted. I can't promise will have the bandwidth to accomodate as Dragons like to keep busy - But ... for example - if I could post 6 side by side shots - MX vs Dragon - tell me EXACTLY what you would want me to post.

 

[Nick Morrison]

What is leaps and heaps for you Paul? Obviously I can't answer that - BUT ... I would like to know EXACTLY what you would like to see posted. I can't promise will have the bandwidth to accomodate as Dragons like to keep busy - But ... for example - if I could post 6 side by side shots - MX vs Dragon - tell me EXACTLY what you would want me to post.

WOW. Yet another reminder Mark why you are a tremendous, tremendous Reduser. So remarkably generous.

Me personally, I'd be interested in seeing a well lit, controlled interview/glamour setup (a "beauty" look can be a nice approximation for everyday client work)....with some deep blacks and strong highlights in frame...and see them compared at a wide range of common (and uncommon) ISOs:

1) Both at 320 iso
2) Both at 800 iso
3) Both at 1200 iso
4) Both at 2000 iso
5) Both at 3000 iso

 

[paul schefz]

Hmmmmm ....


So ... you have to admit that they are BOTH great cinema cameras. GREAT cameras. You might like one more than the other - or you may have more skin in the game with one over the other - or you might have had a bad experience with one of them - but these two cameras are clearly amazing tools that have both been used to create amazing films.

i agree and will go with what bob said....and obviously the leap from 11 to 14 stops is bigger then 14 to 17...and it is great to have a camera that does what it does so well and i mean the epic and of course it gets harder and harder to improve on...
i guess what i mean to say in my post was that epic is not even close to as bad as some people make it out to be (even when compared to the alexa) and alexa is not the end of all cameras either...not even close....so if people get excited and jump on the dragon bandwagon even if they don't even know the difference or could tell the difference...its fine by me...
i know there is a difference and i can see definite improvement and i won't make a final judgement until all tweaking is finished anyway....
i would still love to see short comparisons of what Nick asked for...that would be awesome....i and a lot of other people would be incredibly grateful....

 

[Phil Holland]

Thank you Phil. I will have to test the Schneider Platinums a little more. I have been told they fall down in hot weather. Ever tried a hot day shoot with them and some nice cheap black t shirts?

When you say expensive filter, which expensive filter? Would Dragon play as nice with moderately expensive regular NDs (i.e. Schneider/Tiffen)?

Yep. That's where all IRNDs tend to have issues. It's just tinted glass that's designed to correct the color cast under normal conditions. The 100% best solution is the Hot Mirror/IRND Combo Filter.

I was mostly showcasing that both expensive and inexpensive NDs worked "as you'd like" with Dragon. There will be various levels of performance variance due to manufacturing techniques and materials however. The good news is so far every ND I've tested, cheap through pricey, all work. Even Vari-NDs.

There are less than a handful of poeple in the world that have worked with - from camera to post - the quantity of RED footage that I have. It's what I have been doing since RED first shipped the RED ONE. And for me - it's "leaps and heaps" better. Trust me - after you rock a Dragon for a very short time - you won't even want to pick up your MX. You will see.

100% agree. During even testing the prototypes I was hungering for Dragon more and more. Once you have a taste of something better, it's real hard to go back.

That new and good color, massive dynamic range, and the not-so-digital highlight roll-off are a real big deal from a creative and technical perspective. There are resolution gains obviously as well. I love having 5K in that S35 format now. That's going to be many will shoot with.


Just to chime in on a few other things I've seen pop up in this thread.

As for comparing systems, one thing that many are overlooking is what exactly the camera is seeing at it's base sensitivity. Which is a trickier thing to test. It is noticeably more on Dragon and that's where a lot of that subtle, yet relevant dynamic range can be found. In terms of workable latitude in post, that will mostly depend on how you expose. But, if exposed properly above the noise floor you have crazy freedom in pushing things around. If pressing into the noise floor, you still have a great deal of latitude, but you may want to mind any lifting out of the darks if you want to avoid revealing noise.

In terms of exposing, I'm mostly shooting at ISO 500-800 for most things because I like where values are placed and how much post freedom I have. I shot a bunch of ISO 2000 recently and that looks solid. I think the texture of ISO 3200 is also something very cool. There's something slightly different about the "energy" of the signal noise on Dragon in comparison to Mysterium-X. I think if you like the sort of texture ASA 500 film gets you, you may be into ISO 3200/4000 in particular.

I find myself mostly shooting between REDCODE 5:1 and 9:1 at the moment at the various 6K formats as well. Overcranking obviously diving into the higher ratios. Still testing various conditions on REDCODE, but overall I'd say it holds detail better than Mysterium-X. The overcranked material shows that the most in the shadows.

 

[Gunleik Groven]

WOW. Yet another reminder Mark why you are a tremendous, tremendous Reduser. So remarkably generous.

Me personally, I'd be interested in seeing a well lit, controlled interview/glamour setup (a "beauty" look can be a nice approximation for everyday client work)....with some deep blacks and strong highlights in frame...and see them compared at a wide range of common (and uncommon) ISOs:

1) Both at 320 iso
2) Both at 800 iso
3) Both at 1200 iso
4) Both at 2000 iso
5) Both at 3000 iso

Actually this is the most confusing request you can make, in practical terns...
And you would not necessarily learn anything from it. It is easily rigged to show that 320 =3000 (which it in practical terms actually is on the RED and mostly also on the Alexa)
To get info, you must hit it with different light levels in a controlled way.

To simplify your example, I will use these ISO numbers:

400 - 800 - 1600 - 3200

Which would mean that you make 4 exposures and in a controlled way take off 1 - 2 - 3 stops of light from the base exposure.

Then you would see how the same scene works at different exposure-levels (and yes: those levels would be normalized with an ISO correction, but the look, color, noise, and vibrance would differ a lot)

I did something like that, but it gives you a lot more informative if you do it yourself.

I totally agree with EVERYTHING Mark says. Also about the TEST IT YOURSELF. Tests like Marks and mine only have limited 2nd hand value, AND we test for different things, and thus the results may seem inconsistent.
To me they do not, as methodology is a big part of test-results.

And that is the thing I would like to add to Marks comment about testing yourself.

KNOW what you are testing for an how to extract the information you are after.

THAT could well be done with a general test-shoot and getting "the feel" of the camera.

But if you ask for specific technical info, you need to use a methodology that makes it reasonable to think that that kind of information is extractable from the tests you do.

And even then, you should reshoot the same test under different conditions (say daylight/tungsten. Colorful/contrastful xxxx/x2x2x2) scenes to be able to pin together a larger picture.

One could be shooting just a macbeth or a chroma du monde through the range. That way you could extract info about exposure vs resolution and how color-renderition might differ under different exposure conditions)

Also it is important to shoot images which are "relevant to what you want to do". As I (at least) usually do not shoot charts, and charts and pure tech scenes only give THAT part of the equation.

There is no ONE single test that tells "it all". Not even on noise. as just a small variable in how you expose can affect the result dramatically.

How you make nice blacks (as in noiseless) has to do with exposure. (light hitting the sensor) not "ISO settings". And that goes for ALL cameras.

So tests, like the ones I have published... Only have limited value. (I have mainly stopped publishing them, because I see that they are not interpreted in a way that gives sensible info to many of those who see them... )

Tests are good.

If you understand what it is that is tested for.

If not... They are just confusing.

I see "the same" camera in what Phil, "the Marks" and I have published. But I can see that that may seem confusing...

The "Why do my shots not look like his" question, is usually easily answered through: Well, you probably did not shoot like him.

Independently of camera.

And as Mark said, ALL the good cameras these days are very good. If I shot for a match (and not a tech test to see where they differ) I bet I could match them all pretty well on a given scene...

In that case I would test for "matchability"

LOL

It is all about what you test for...

Or in a more philosophical/science orientated general statement:

It is not about the answer you get to your question, but the way you asked the question in the first place.

The variation of results are close to infinite with just nuances in ways to ask. Even though you may THINK you ask the same question every time...

Thus, back to my point:
Methodology is a huge part to understand if you want to read any actual info from tests.

 

[Gunleik Groven]

i agree and will go with what bob said....and obviously the leap from 11 to 14 stops is bigger then 14 to 17....

Actually, the second leap is a lot bigger and more difficult and more rewarding... IF it is done by extended DR and not Gearing of the sensor and optical-path techniques that you yourself can apply anyway (the optical path tricks) if you know your filters. (Actually you can to SOME extent also simulate the sensor-gearing in post, if you know what you do...)

But THAT is because "a stop is not a stop"

And that is especially true in digital cameras.

What adds to the fun in the case of Dragon, is that fortunately "not all sensor tech is similar". It is not just about what the cams are technically able to capture, which is a tech parameter, but how comforting the image "feels" in different stages of the process.

The Alexas ability to appear easy and good in the capturing stage has been its greatest advantage. Whatever resulting image is possible to make with the REDs vs xxx
And I like that with the Alexa, Dragon gives the DP a lot of confidence in what he/she sees and does. Confidence is good and cannot be technically measured, but sometimes has huge artistic impact...

The Dragon has a lot of that "comfort" in the on-set image.
In addition to the tech advances, THAT cannot be properly displayed in a test...

I agree 110% with Marks comment that peoples perception on the Alexas and other cameras noise is a bit... exaggerated. I have found noise (as a single parameter when it is that I have checked for) is very similar at same exposure levels between the MX and the Alexa. AND I can easily why exactly THAT can make the Alexa seem smoother in some circumstances.

Confusing?
Not really. If you start to look into how they handle exposures. But "noise at ISO X", is NOT part of the answer. If that is your question, the answer is: THey are very alike.

Understanding sensor-tech and signal-chains, has little to do with pointing cameras at the scene and change WB and ISO or start tweaking the image in-camera.

To extract info of a camera, you must throw different light and scenes at the sensor. Not change the in-camera tweaks. (and that goes for all the "relevant" cameras in this discussion....)

And if you want to see how a camera works under let us say "halogen lights", you need to shoot them in "halogen lights"

It is actually as simple as that. Because the color-renderition and response to non full-specter lights can be very different due to sensor response and matrix adjustments.

There is really no easy answer to tests...

(and I have performed a few...)

 

[Ergin Ozturk]

As a epic owner i am looking at every single dragon footages and checking the forum for every day .. i am trying to convince myself to upgrade. but not yet..
ok. i do not care 6K+. 4k is more than enough for me.

I am not a fan boy of red or alexa.
i love alexa footages. from the beginning. alexa is good replacement of film. whenever i see some music video or feature shot with alexa i am in love.

Arri guys did good work on;
1-softer creamy look
2-dr and highlights
3-color science

Softer creamy look is not a big deal since it can be solved in post.

It seems Dragon now has same DR (or may be little more) with Alexa. It is good. and now highlights are more under control with dragon.

Color: That is the biggest difference between Alexa and Red. If you need low saturation or dominant color (blue, green etc) footages no problem you can get it with RED. (for example: "Girl with dragon tatoo"-Fincher). But if you need each color seperated and high contrast look then you need Alexa. Or you have to deal with lots of secondary color separations with RED.
if you need Alexa color example just watch "Drive" movie.

So color science is the biggest differentiator of Alexa now. RED is working hard. i hope they can do better.

 

[Gunleik Groven]

Sensor color-response on Dragon is a lot ritcher than on Epic, even as it is. And I see that (and I think partly how) it will improve a lot.

But it is there in the signal.

 

[Ergin Ozturk]

Sensor color-response on Dragon is a lot ritcher than on Epic, even as it is. And I see that (and I think partly how) it will improve a lot.

But it is there in the signal.

Gunleik as you said "sensor color-response" is the key. When i say new color science I do not mean RedGammaDragon since i believe none of the pro colorist using RedGamma2or3. most of them starting their job by using RedLogFilm. So question is what new dragon color science can give us as RAW or RedLogFilm.

 

[Gunleik Groven]

A lot, is what I think...

I can see how the saturationresponse throughout the range is, compared to Epic or ALexa. But I can also see that it is not yet fully tweaked.

"RAW" is not "RAW"...

 

[Phil Holland]

Gunleik as you said "sensor color-response" is the key. When i say new color science I do not mean RedGammaDragon since i believe none of the pro colorist using RedGamma2or3. most of them starting their job by using RedLogFilm. So question is what new dragon color science can give us as RAW or RedLogFilm.

While there is new color science coming, remember that Dragon's color right now is very different than Mysterium-X even when using REDcolor3 and the various gamma spaces. Based on conversations I've had with people in the outside world I think a few people are confused that since we're using the same color science that the look is the same between MX and Dragon, however that's not the case at all. Dragon is different and better (richer, more accurate).

This is primarily due to 3 things:

- The new Dragon Sensor technology and look
- The new OLPF that is optimized for Dragon
- The calibration and tuning of Dragon for the current color science

Honestly, right now there's nothing to really criticize about the color out of Dragon. It's rather good actually under a variety of conditions and out performs other systems in terms of neutral density IR contamination. In fact it's essentially solved that problem all together really.

I'm not using any IR protection with Dragon with NDs at all and I've tested up to 2.7 (9 stop reduction) . This is damn impressive in reality.

We know more is coming on the color front. However things are vastly improved from the get go and it's extremely noticeable to me.

Currently though, I'm getting the most interesting color out of Dragon using REDcolor3/REDlogFilm. It's nice and meaty.

 

[Filip Orlandic]

When i say new color science I do not mean RedGammaDragon since i believe none of the pro colorist using RedGamma2or3. most of them starting their job by using RedLogFilm.

Many colorists use Redgamma 2 & 3 for grading. The worst thing is that some guys still use these metadata settings: REC709/REC709 and REDspace/REDspace.

 

[Mark L. Pederson]

Currently though, I'm getting the most interesting color out of Dragon using REDcolor3/REDlogFilm. It's nice and meaty.

Agreed.

 

[Mark L. Pederson]

Actually this is the most confusing request you can make, in practical terns...
And you would not necessarily learn anything from it. It is easily rigged to show that 320 =3000 (which it in practical terms actually is on the RED and mostly also on the Alexa)
To get info, you must hit it with different light levels in a controlled way.

To simplify your example, I will use these ISO numbers:

400 - 800 - 1600 - 3200

Which would mean that you make 4 exposures and in a controlled way take off 1 - 2 - 3 stops of light from the base exposure.

Then you would see how the same scene works at different exposure-levels (and yes: those levels would be normalized with an ISO correction, but the look, color, noise, and vibrance would differ a lot)

My reaction exactly. But I think I can reverse engineer what he really wants to see - and yeah - you would need to add/subtract light from a "base" exposure - which also calls into question what method you want to use to get that "base exposure"

I have a good idea for infromative side by side MX vs DRAGON test files - let me see how next week shakes out -

 

[Hrvoje Simic]

if I could post 6 side by side shots - MX vs Dragon - tell me EXACTLY what you would want me to post.

These would be pretty informative. There is the 7th but these should cover all.


1+2 ) Color/color consistency/tonality

MS, interior

Child + man + older woman (zero make up on all) + color chart + fruit basket, against a black background, single source soft light moving from from one side to another, from unlit to unlit

HMI + Tungsten shots

2 x 5 sec



3) Detail / compression

WS, exterior

Forest, deep focus, slightly windy with moving foliage

3 sec



4) DR

WS, interior or exterior combo shot

Either interior - unlit room with a view outside the window or exterior - foreground building with unlit interior, background sunrise or sunset

1-2 sec



5+6) sensitivity

WS, exterior

Dusk, nature background, color chart foreground- no lighting

City at night, mixed street lighting, street and darker alley background, color chart foreground

2 x 3 sec



7) Motion rendition/shutter look

MS, exterior

Both cameras mounted in or on a vehicle facing sideways (cameras facing sideways not the vehicle, that is unless one considers it from the camera angle, in which case this comment is completely unnecessary and should be disregarded),
city street, buildings and people 10-50 feet away, 35 mm lens, vehicle slowly speeding up from 0-60 Mph.

10 sec




All shot with a slate.

 

[David Battistella]

Gunleik, Mark, Phil, Toia,

I don't want to get into an ISO discussion or a BASE ISO discussion. I think RED will eventually release what they believe the base sensor ISO (sweet spot) to be when the time comes.

What I am seeing in the R3D's is the tonality of how the sensor photo buckets suck up the light and redistribute that in the debayerd R3D file.
The gradients are smoother from dark to light, in between, and also color to color.

I am curious about what any of you might feel is the "sweet spot" right now, knowing that will change.

It would be nice to see a test like, lighting and metering a scene at 320 ISO F4 and then see what 320ISO looks like against metered light readings.
The same scene metered and lit for 800ISO F4 and see what that is like against metered light.
Same for ISO 1000 F4 (metered and lit) compare that.

This can easily be achieved by using ND on the fixtures, but I think it would show us just what happens to the tonality/color of a scene.

Key is to leave the shutter and aperture consistent and reduce the amount of light and increase the ISO to see how that is different in Dragon compared to MX.

I think a test like this would really show what Dragon does over EPIC X. My theory is that it would show that Dragon is more "pushable" or "elastic" or "forgiving in low light.


Dragon feels "right" and a lot of "harshness" for lack of a better word is gone (particularly when rolling up toward white (highlights). I use lo-con or dig icon on EPIC MX to create this similar kind of effect. Don't think I will need them much with Dragon. , But you can still get a lot out of EPIC MX.

Battistella

Here is a shot with locon to "stretch" the MX image.

 

[Mark L. Pederson]

I don't want to get into an ISO discussion or a BASE ISO discussion. I think RED will eventually release what they believe the base sensor ISO (sweet spot) to be when the time comes.

The thing is ... there is not a "sweet spot" ... there's a "sweet range". It really depends on what you are shooting and what you want to achive. I think the term "base sensor ISO" is not an accurate thing to say. If you think you are going to find ONE "sweet spot" to rate the Dragon for all situations and all desired results .... you are gonna chase your tail.

 

[Mark L. Pederson]

These would be pretty informative. There is the 7th but these should cover all.


1+2 ) Color/color consistency/tonality

MS, interior

Child + man + older woman (zero make up on all) + color chart, against a black background, single source soft light moving from from one side to another, from unlit to unlit

HMI + Tungsten shots

2 x 5 sec



3) Detail / compression

WS, exterior

Forest, deep focus, slightly windy with moving foliage

3 sec



4) DR

WS, interior or exterior combo shot

Either interior - unlit room with a view outside the window or exterior - foreground building with unlit interior, background sunrise or sunset

1-2 sec



5+6) sensitivity

WS, exterior

Dusk, nature background, color chart foreground- no lighting

City at night, mixed street lighting, street and darker alley background, color chart foreground

2 x 3 sec



7) Motion rendition/shutter look

MS, exterior

Both cameras mounted in or on a vehicle facing sideways (cameras facing sideways not the vehicle, that is unless one considers it from the camera angle, in which case this comment is completely unnecessary and should be disregarded),
city street, buildings and people 10-50 feet away, 35 mm lens, vehicle slowly speeding up from 0-60 Mph.

10 sec




All shot with a slate.

It's really f'in cold here in NY right now - I might have to adjust this list a bit