Welcome to our community

Be a part of something great, join today!

Register 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 :)

"Dueling with DRAGON: RED's 6K Sensor Takes on Older MX in This EPIC Camera Shootout"

JeanChristophe Dupasquier

JeanChristophe Dupasquier

Well-known member
Joined
Dec 17, 2013
Messages
160
Reaction score
0
Points
16
Location
Switzerland
"Dueling with DRAGON: RED's 6K Sensor Takes on Older MX in This EPIC Camera Shootout"

Someone at No Film School has been reading reduser.net as it seems :)

For pixel peepers and everybody else:
http://nofilmschool.com/2014/07/red-epic-dragon-vs-mx-sensor-raw-camera-shootout/

Dynamic Range and Rolloff, also compared to the Arri Alexa:

Filters, IR and color shift:

Color science, skintones, contrast ratios, low light, (red) noise, compression:
 
Last edited:
I think some of these are old tests... or all of them.
 
14 is the new 16, I guess... This is very similar (the same) to my initial results with the new OLPF back in Feb.

I'd be very curious to see how MX would do with a DEB-equivalent helping to clean up the (more useable) blacks... Maybe get an extra stop of highlight protection with similar noise/useable shadows?
 
Anthony Berenato Jr said:
The article states New OLPF.

Really surprised with the underexposure test. Something doesn't seem right.
Click to expand...

It's right. Dragon with new olpf is not as great in lowlight as the old olpf. That's why a lot of us have been asking for an olpf that splits the difference.
 
I had a Dragon shoot recently for a spot involving some green screen and practical rain and vfx, shot 250 inside and outside. The problems occur when not shooting for the sensors strong points. I have to say it is so weird now switching to raw view and seeing 800. That takes some getting used to. But at 250 Dragon is super clean, even inside.
 
Damn, this is pretty rough. I think a lot of the misinformation about Dragon comes from the initial specs given and even the test films first launched with the camera ( Old OLPF ). It's simply a totally different camera on both those terms, especially seen in these tests. That dynamic range test doesn't lie though, hard to see 16.5 stops in that, or am I reading it wrong?
 
The dynamic range test comparing Alexa with Dragon illustrated the gains in highlight performance but the shadows seemed drastically lacking(on the new OLPF).
Pluses when compared to MX sensor
14 stops of dynamic range(across several different tests now with OLD/NEW OLPF, FIRMWARE, Color Science),
Better color/ skin tones,
Higher resolution
Minuses when compared to MX sensor
limited sensitivity
higher data rates
 
The thing is, in this test, when comparing Dragon with the MX, there doesn't seem to be two more stops of latitude. It looks as they have taken two stops from the shadows and put them in the highlights.

In fact in looks like the Dragon has one more stop in the highlights and two less in the shadows.

There has to be something off in this test, because that's simply not what their latitude graph shows or what you see here:

https://www.flickr.com/photos/123436930@N07/14457270568/in/photostream/
 
What I don't understand are these low light tests, where they compose the frame, light it, expose it right, then go on to add ND's and push it in post. Who does that in real life?
What I want to see is a frame with real ratios: Skintone exposed right, some highlights in the background, but also dark areas. Jus a contrasty scene with a lot of ratio. I want to see the noise there. Rated at 800 for accurate exposure on the skin tones, and see how the rest of the frame fares. No pushing in post to get back to a "normal" exposure...
 
Matt Ryan said:
It's right. Dragon with new olpf is not as great in lowlight as the old olpf. That's why a lot of us have been asking for an olpf that splits the difference.
Click to expand...

Definitely understand, but I'm still most surprised at how crushed the blacks are in the Dragon underexposure test. I was really expecting to see a milky image with a lot of noise, not a very contrasty image with no shadow detail other than pure black at a stop and 2/3rds under.

I'll have to see if they uploaded R3Ds when I get home. I've seen underexposed New OLPF Dragon images on REDUser that were almost three stops under and the shadows didn't flatten this bad when the image was brought up (granted the extreme shadow detail is many more stops under in these examples)

If the difference between usable shadows in MX and Dragon reacts this differently, I'd expect the Dragon OLPF to have a coating you can barely see through. Something still feels off here, but I don't own a Dragon and I haven't had the experience of learning how it reacts to light.

Really sorry if I come off like a jerk! Just rather interested in what the OLPF is doing with the light here for future knowledge.
 
The compression test was bogus. First, he says that he's fine with ISO 800 and OK with ISO 1600, but would not use ISO 3200 unless somebody held a gun to his head. He then shoots the compression test at ISO 3200 which creates all sorts of noise (no surprise). The then judges that 12:1 compresses noise in a bad way and that 18:1 compresses noise in an unacceptable way. A better test would have compressed either ISO 800 or ISO 1600. If *those* compressed unacceptably, fine. But if the lower noise resulted in low enough compression artifacts then he could have reported "honestly, 12:1 is perfectly usable on DRAGON" or even "18:1 is perfectly useable on DRAGON". Mark Toia's results certainly show that 12:1 and even 17:1 can be pretty amazing. Intentionally breaking the image quality and then making a recommendation that *compression* is to blame for unacceptable quality is bad science.

Another problem I have with these tests is that he compares DRAGON to EPIC at 1:1 pixel resolution, which basically negates the 6K vs. 5K advantage. To truly evaluate these cameras--including noise--everything should be finished at 4K. If that results in lots of noise being pushed down below perceptible levels, isn't that the point? There *are* tradeoffs making the photosites of the 6K DRAGON smaller than the photosites of the 5K EPIC. Comparing them at 1:1 instead of 4K kinda defeats the whole concept of "Shoot at >4K to finish at 4K".

I was also confused about the initial exposure test of the Sekonic chart. Were these truly shot at REDLogFilm? Or were they shot at RG4? The S-curve tells me RG4. He said they were all log, but I didn't see the evidence that he got that right.
 
can´t confirm lots of stuff - in my real tests outdoors Dragon is better or at least equal with MX, i can´t confirm that MX has more in shadows, ... i surely will test again with all the new experiences of the last weeks (DEB, Black Shading)

https://www.flickr.com/photos/123436930@N07/13929414249/

https://www.flickr.com/photos/123436930@N07/13919107117/

Fact is, you must be careful (more careful) with Dragon and Black Shading and temperatures ...
after reading here the last 2 months, i am always surprised how good my Dragon outputs look in comparison to a lot of other footage posted here.
i would never go back to MX honestly, as said before a lot of times - colors and highlights are (for me) much more important as noise in higher ISO - as i shoot 90% of my stuff under controlled light situations...
 
Michael Tiemann said:
Another problem I have with these tests is that he compares DRAGON to EPIC at 1:1 pixel resolution, which basically negates the 6K vs. 5K advantage. To truly evaluate these cameras--including noise--everything should be finished at 4K.
Click to expand...

Fair point.

Michael Tiemann said:
I was also confused about the initial exposure test of the Sekonic chart. Were these truly shot at REDLogFilm? Or were they shot at RG4? The S-curve tells me RG4. He said they were all log, but I didn't see the evidence that he got that right.
Click to expand...

RG4 is certainly not the same as Red Log Film, and would certainly skew results. Not sure if there's a simple way to quickly see which was used though...
 
You must log in or register to reply here.
Back
Top