Dragon/Epic comparison

[Gunleik Groven]

Hi.

Because of the other thread, I will post a few comparative images, but I will show off the developed images from both cameras. No R3Ds as of yet.

Hang on

(And do not post a million answers before the images are up...)

G

 

[Gunleik Groven]

Ok.

This is a quickie, Indoor with mixed light.

Unfortunately I could not get a higher exposure than the first one here.

Exposure ranges from T1.3 1/25 to T16 1/6400

Lightsources were mixed.
LED @5600, an Arri 300W Tungsten directly into the cameras.
and a small Halogen nightstand lamp.

And of course my beautyful NSFW friend Espen!

Here comes the first.

Image 1 is base exposure.
Developmentsettings "as shot" RG3/RC3 800/5600

First "development setting to look at: log, as that would be natural at the start of a Grade
Same as above @ RLF

Third I look at the highlights only: Linear/320

Developed @ Linear/320

As you can see, I did not really have enough light to break the Dragon available for this test....

Link to fullsize downloadables jpegs for each exposure will come...

Links to fullsize "100%" compression setting jpegs:

http://www.gunleik.com/dragonepic/jpg_fullsize_01.zip

12 frames r3ds

http://www.gunleik.com/dragonepic/r3d_01.zip

 

[Björn Benckert]

realy nice Gunlik, Please can you not post a crop out of his shirt or something that is the least exposed. And post in pixel 1:1 side by side from each. I guess its doable from your jpgs that are to come but still.

 

[Gunleik Groven]

Thx. Good to try to keep this thread uncluttered without many other than my comments for now.

Feel like I should say one thing of significance before I prepare the next batch of images:
This does NOT make the Epic bad!!!

I will probably have ample time to shoot Epic the next few months, and I will be a happy camper doing so...

I would just have slapped in an ND9 and been fine.


Should also note that what you see around the letters on the iPad is CA from a superspeed @ T1.3

 

[Laco Zamba]

Try DRX

 

[Gunleik Groven]

Try DRX

I NEVER had any issues with magenta on Epic. THAT is not the issue.
But the added highlight tolerance and colorhandling is kinda interesting IMHO.
Point here is more like what is in the image "out of the box"

Please, no more comments on this thread until I am done... with the examples, or I will just ask the mods to take it down.

(and I will spend a few hours. Do not complain! I could just have said, like I have done before; Dragon is "better", but that silly example in the other thread is so misleading, that I think it is ok to rectify a bit...)

 

[Gunleik Groven]

Next... "Base exposure"

5 stops down from the above examples.

T5.6/800/5600/RG3:

And

T5.6/800/5600/RLF

And here are the fullsize JPGs

http://www.gunleik.com/dragonepic/jpg_fullsize_02.zip

12 frames r3ds

http://www.gunleik.com/dragonepic/r3d_02.zip

 

[Gunleik Groven]

Now it gets interesting...

Let us go two more stops down

First


T5.6, 1/200, 800/5600 RG3:

And of course...

T5.6 1/200, 800/5600, RLF:

Then, let us compensate back to base exposure with ISO, thus go to 3200


T5.6, 1/200, ISO 3200/5600, RG3

And finally:

T5.6, 1/200, ISO 3200/5600, RLF

Now... I will not force you to see what I see, but if you download and compare the 8-bit full-size jpgs, there are particularly 2 things I would like to bring your attention to

1. tonality in skin
2. Noise on the iPads frame...

Here are the fullsize jpgs

http://www.gunleik.com/dragonepic/jpg_fullsize_03.zip

Added

Stillframe R3Ds:
http://www.gunleik.com/dragonepic/r3d_03.zip

 

[Les Dittert]

1:1 crop image

1:1 crop image

Thanks for taking the time to run this test Gunleik. Here is a crop from the ISO3200 . Dragon has less noise. I blew up the MX sensor to better match the size.

Dragon on left.

 

[Gunleik Groven]

Thanks Les.

It does indeed...
And a lot more sensitivity in both ends.

Look at the guys eye. Tonality - difference is striking.

I am debating whether I should post 2 stops down from this, too. But decided against, since I fear many people will not know what they are watching.

And I think that when we se DragonGamma/DragonColor, these posts will be irrelevant.
I have a hunch or two about what we will see when those arrives.
(No inside info. Just an idea after testing this and a few other things)

If someone from RED steps in and either takes this thread down (because the colorscience is not done) OR gives the "go" for posting these R3Ds, I will gladly comply to both.

Last thing I would like to say.

I took this all the way to

T16 1/6400 exposure.
That is 16 stops below the 1st exposure. (And I did not fully overexpose that first image on the dragon...)

On the Dragon at 4000 ISO, I have an image.
The Epic has nothing except for the lights.

The +16 stops claim seems more than possible from this initial test. The Dragon should had nothing either, here.

I will push a bit of highlights in snow tomorrow, but do not expect to see hours of tests from me outside that.

OK

Let the discussion begin.

Keep it calm and be polite.

Cheers

G

 

[Kwan Khan]

NIce and Thanks

 

[D Fuller]

The most interesting is, as you say, the difference in tonality. It is striking.

The highlight detail and shadow detail are nice-more than nice, but kind of expected after shat we've seen already. Where it becomes very interesting is the gentleness with which the shadows and features of his face are rendered (in this rather harsh light). I didn't expect that so much. With a bit of work on the curve, this would seem to offer tremendous control in the look of skin tones. I love seeing that.

 

[Gunleik Groven]

Well, to me it is all about fidelity, and nuances in colors.

THAT is really what the Dragon means to me.

But the added latitude and fatness of the images are two other huge bonuses, as is 6k when you want it.

I basically want it...


The best two analogies for me still of the difference between Dragon and any other camera I have worked with:

First is the difference of recording a voice or a grand piano with either a Shure SM 57 or a Neuman U87
You probably can get all the notes and the dynamics recorded with both, but it will sound very different.

Second is the feeling I had when I got my DrumWorkshop Drums.
I could not play louder or faster with them, but gee... I really sounded better...

 

[Björn Benckert]

Thanks for taking the time to run this test Gunleik. Here is a crop from the ISO3200 . Dragon has less noise. I blew up the MX sensor to better match the size.

Dragon on left.

Les, When you scale up the MX you also scale up the noise I think a pixel to pixel comparison should be just that then if the subject differ does not matter much. or?

 

[Nick Pasquariello]

I'm surprised by two things.

First, that the MX is so much more magenta, while the Dragon is looking greenish. Would love any insight anyone has into that. Those are CRAZY different tints, assuming that the lens, lighting, and position are all exactly the same (and I have no reason to believe that those three things are not exactly the same).

Second, I'm surprised that, looking at the blowups that Les provided, I don't really see that much of a difference in noise. The dragon certainly holds more detail in the highlights, and more detail in the darks, but noise for noise, I'm not seeing the difference.


Oh, and Thank You to Les for providing the blowups, and an EXTREME Thank You to Gunleik for providing an Apples to Apples comparison. Incredibly useful for those of us that can't actually get our hands on a Dragon just yet.

 

[Gunleik Groven]

Les, When you scale up the MX you also scale up the noise I think a pixel to pixel comparison should be just that then if the subject differ does not matter much. or?

Agreed.

As the cameras stand on the same point. there should be no point in scaling the Epic.

You see in the eaxmples the practical difference between 5k and 6k in an 1:1 image (if you download the zips)
And which effect that has on FOV

 

[D Fuller]

Well, to me it is all about fidelity, and nuances in colors.

THAT is really what the Dragon means to me.

But the added latitude and fatness of the images are two other huge bonuses, as is 6k when you want it.

I basically want it...

I see that. I don't think I had considered the importance of the fat negative before (That comes, I think, as a result of the dynamic range.) Even in B&W film days, it was always easier to get the print I wanted--especially if I was going to manipulate it--when the negative I started with was both fat and well-exposed. I see that happening here.

It's nice.

 

[Björn Benckert]

So dragon to the right and MX to the left... Not that it matters much as the noise floor is just about the same. Do not know what that means but thats how it is. I do not know much about colour science.

Edit: added an eye 1:1...

There are differences for sure, to me those that show the strength of the dragon is more towards highlight and mids in the low end it looks quite similar. this is me judging from looking at these frames only, and nothing wrong with that I rather have massive highlight protection then being able to shoot in the dark so to speak. again it looks better in the low end swell but down at the noise floor they do look pretty much identical to me in these tests. Curious to see what the dragon color science will bring, again I do not no to much about such things but in my mind an even noise flop has less to do with that and more with calibration, blackshading and actual physics of the sensor.

So all in all from looking at these pictures it looks like the photosites, if looking at them as buckets, are standing on the same base / floor but the dragon buckets does not fill up as quickly.

 

[Gunleik Groven]

You really see the difference 16 stops under the first exposure.

But this was a quickie.

If RED is OK with it, I will post R3Ds, but as I said, I think I see where Dragoncolor/Dragongamma will make a difference, so I am not superkeen on it until that is done.

Essentially, though (and you can see that tendency in my last post) the Epic goes (as expected) into black a lot earlier than the Dragon, and struggles a lot more with nuances, even at lowest compression.

I am glad I did not do anything like this when I had the proto, as the camera has really evolved since late August, and I expect it to still do so over the next few months.

But what a start!

 

[Bob Gundu]

Noise doesn't scare me. Artifacts do however. Dragon has finer pixels and less artifacts. Boom… next. I personally see a vast improvement in all aspects of the sensor. If this is not evident and you need to see this at a pixel level, the upgrade may not be warranted in your case. Nothing wrong with that. It's the same argument about making GH2 footage look like RED footage. Sure you can, but it's a calculated and skillful process. But the Dragon really shines in situations where there are extreme conditions and lesser cameras will quickly show their limitations. Dragon footage to me just feels like a fatter negative.