DragonColour VS RedColour3

[Hans von Sonntag]

Hi,

in another thread we had a discussion about Red's colour sciences, film, green casts, Alexa etc....

Ok. Since I was not too overwhelmed by the noise floor and sensitivity when I got my Epic back dragonised I checked the new colour science. I was positively surprised and must say that Red got a lot right. With the extra DR, 5KHD being 3-perf S-35 with a meaningful 2.5 factor to 1080p and the new colour science the upgrade is a big leap forward. A no-brainer for Epic owners.

Ok. To share my findings I shot a picture that shows the porcelain skin tones of a 10 year old blond kid and the tanned skin of a grumpy old filmmaking daddy. The lens is a 16-42 Optimo @ T8. Setting was DragonColour and RedColour3, Gamma RedLogFilm. For viewing reasons (we don't talk DR here) I added a friendly S-Curve and adjusted the saturation to my personal liking. The exposure was more on the thin side, camera set to ISO 800, so was RedCineX. The light source was a cloudy sky, no sunlight or blue sky involved. The surrounding environment is a white backyard, no WB applied, Kelvin set to 5600.

The second picture shows the same content but with an exaggerated saturation to make the differences more obvious. The nuances are subtle and hardly visible on a normal laptop. However, in a postproduction suite they are obvious, to say the least.


My findings:

The overall colour rendition of DragonColour is subjectively more pleasing to all the folks in my company I showed the pictures, including my self. As one can see, the saturation appears slightly less compared to RedColour3 but mainly in the red/orange/yellow realm, green seems to be a tad warmer, albeit keeping just about the saturation level. The skin tones in Dragon Colour are more subtle, less "carrot-ish" leaving more room to all the colour nuances mankind has in hers/his skin tones.

Lowering in RedColour3 the saturation about 1 or 2 percent does not lead to the same appearance of what DragonColour is doing. There is much more to it than just saturation level or an over-all decrease of warmth.

The whole thing is a gourmet topic. One can state that talking about such nuances is crazy stuff for crazy people. But then the differences are there and they matter, at least to me.

Hans



http://www.reduser.net/forum/attachment.php?attachmentid=52708&stc=1&d=1400690312

http://www.reduser.net/forum/attachment.php?attachmentid=52709&stc=1&d=1400690365

 

[Phil Holland]

I have a few thoughts on the matter and most have been seeing a lot of John Marchant's and my own very "charty" tests.

Dragon is a lot of things. Pretty much the first couple of things that seasoned Mysterium-X shooters notice are:

- Much wider dynamic range
- A very different and more pleasing highlight roll-off
- how vastly improved the overall color and tonal placement is

I've spoken to Graeme time to time regarding Dragon's color science, but perhaps the most interesting discussion came from NAB 2014. DRAGONcolor was in development for a long time. There were of course obstacles to releasing what is now known as the "standard" release. Different OLPF materials and sensor calibration mainly. Once those things were hammered out more effectively, and in my mind more properly, it was more or less time to put the icing on the cake with the what we now know as DRAGONcolor.

Now we're getting a nice and thick image and I can't stress enough just how good the color is right now under various conditions. And you can do a lot with that color too. All good things.

Much like anything many are going to say things and see things based on 3rd party tests, but more or less once testing it out for yourself it's much clearer to see how much improvement Dragon has over Mysterium-X in those features listed above. It was there from the get go with the earlier tech, but it's far more pronounced now that we have the newer tech and color science.

 

[Marcos Montenegro]

Phil-

What's your take on how the histogram between the "goal posts" behaves with Dragon vs MX. I did a recent shoot with a couple of Dragons and I was surprised how different it behaved.

 

[Phil Holland]

What's your take on how the histogram between the "goal posts" behaves with Dragon vs MX. I did a recent shoot with a couple of Dragons and I was surprised how different it behaved.

This is more or less due to the additional dynamic range in comparision to Mysterium-X. The noise floor and overall dynamic range is different and literally sitting next to each other the histogram might look differently as Dragon is "seeing more".

Goal posts and stop lights are working as intended. As are the various IRE checking tools. Middle Gray is dead on.

 

[Eric Haase]

Goal posts and stop lights are working as intended. As are the various IRE checking tools. Middle Gray is dead on.

Whatever happened to the old RedOne IRE spot meter? Man I loved that feature. You could make that box tiny and move it around the screen. Great for pegging green screen, white cyc, or matching skin tone exposures in narrative work...

 

[Phil Holland]

Whatever happened to the old RedOne IRE spot meter? Man I loved that feature. You could make that box tiny and move it around the screen. Great for pegging green screen, white cyc, or matching skin tone exposures in narrative work...

If you use the 1:1 Magnification the Exposure Tools, Histogram, Goal Posts, and Stop Lights all "see" what that sees.

 

[Matthew Scott]

Nice findings Hans, thanks for sharing

 

[Mark Wilkinson]

Thanks, Hans.

 

[Hans von Sonntag]

I have a few thoughts on the matter and most have been seeing a lot of John Marchant's and my own very "charty" tests.

Dragon is a lot of things. Pretty much the first couple of things that seasoned Mysterium-X shooters notice are:

- Much wider dynamic range
- A very different and more pleasing highlight roll-off
- how vastly improved the overall color and tonal placement is

Yes, I agree.

The visibly better roll-off is there but connects naturally to the wider dynamic range. In a high-contrast environment such as a sunny park one can see the improved DR very well. But to think that the better DR boost sensitivity AND a better roll-off is not working. IMHO, Dragon has a similar "native" sensitivity like the MX sensor but due to the extend DR you can lower the exposure by a stop, or push by a stop or so more than is was possible with the MX-sensor.

To speak about "vastly improved colour" may be a slight exaggeration since the colour of the MX sensor wasn't too bad either. But I agree there has been done a lot of good work.

---

What I find interesting ist that Dragon colour makes an over-saturation possible without transferring faces to burring carrots. Try that with Sony F5, or worse with Canon DSLRs. Alexa has always been good at that. What also strikes me are the cleaner, more neutral shadows when DragonColour is applied. With RedColour3 shadows in the skin-tones tend to have a slight green tint. This tint is gone in Dragon colour. But Green as a strong colour remains but is now warmer interpreted, probably closer to the perception of the human eye.


Hans

 

[Phil Holland]

To speak about "vastly improved colour" may be a slight exaggeration since the colour of the MX sensor wasn't too bad either. But I agree there has been done a lot of good work.

---

What I find interesting ist that Dragon colour makes an over-saturation possible without transferring faces to burring carrots. Try that with Sony F5, or worse with Canon DSLRs. Alexa has always been good at that. What also strikes me are the cleaner, more neutral shadows when DragonColour is applied. With RedColour3 shadows in the skin-tones tend to have a slight green tint. This tint is gone in Dragon colour. But Green as a strong colour remains but is now warmer interpreted, probably closer to the perception of the human eye.

Heh. You said it right there really

Seriously much of my opinion of the color really comes down to how good it is straight out of camera. I have shot many-a-chart, but beyond that it responds well in a variety of conditions. I am also including the fact that IR contamination is no longer an issue. Which for me means under standard shooting conditions using plain old NDs we're getting something good right away. On top of that I like that DRX has been "mapped in or out" depending on how you look at it. It's just no longer needed nor a concern. That's been one of the biggest tells for me in the past of not so properly graded Mysterium-X material.

Now we have something nice right from the start with minimal additional concern. It's subtle to some, but to me it's a bit more important.

 

[Hans von Sonntag]

Ok. Now lets talk DR.

Since my Epic has gone through a metamorphose I cannot show a comparison between Dragon and MX. But what I always do when I got hold of a new camera I visit the park near my house when the sun shines and take a picture of the shadows and bright sun-lit grass and sky.

For comparison reasons I took the same file and transcoded it to RedLogFilm and RedGamma4.



My findings:

- Since green is everywhere skin tone is not the subject to talk about. But nonetheless I integrated a young colleague as reference to show where middle gray approx. is set to.

- RedGamma4 is not a particular heavy Rec709 gamma curve. It sets black to 0 and white to 100, something RedGamma3 is not doing out of the box. Obviously the scene holds a lot of contrast and hence the gamma curve seems a bit heavy.

- Because the shooting environment is so contrasty we can see that the environment challenges the Dragon sensor to its limits. Very good to see when looking at the log encoded RedLogFilm alternative. I only brought down the blacks to 0 and the clipping whites a bit up to 100 to match the full gamut. The picture looks fine, no much further tweaking necessary. This happens rarely when picture are log encoded and only when the scene contains a lot of stops.



Summary: Dragon is not a magic-DR sensor. But is catches incredibly much detail and delivers enough DR to shoot in such a challenging environment. Compression also does not seem to be an issue which has been with MX. We all remember the tree of death in old Build 16 times which broke the codec and I had issues with early Epic MX builds. This park has plenty of such trees and bushes. No blotches, nothing. The compression was set to 1:5 @ 5KHD.

Also nice is the light green of the overexposed lawn in the BG. It contains all the detail, but does not show funky video clipping colours, nor is it artificially de-satuarted for roll-off reasons. Well done!

Recently I purchased a BM pocket camera. Nice little device. But no chance in this environment. Unfortunately the pictures are lost and the camera is somewhere shooting art-house stuff.

Hans


PS: Shots are best viewed with gamma 2.4 (rec709), most laptops have their gamma set to 2.2.

http://www.reduser.net/forum/attachment.php?attachmentid=52742&stc=1&d=1400862780

http://www.reduser.net/forum/attachment.php?attachmentid=52743&stc=1&d=1400862799

 

[Eric Haase]

Very interesting Hans. Noticing how the cloud in the upper right disappears in RG4 but is present in RedlogFilm. A good case for power windows in a final grade... I have also noticed in RG3 shooting a white cyc that I cannot get white to even approach 100- seems to be limited to 93 or 94. Sounds like RG4 may be different and roll off right up to 99 or 100 IRE? Should be getting my Dragon very soon so I will test this too.

 

[Hans von Sonntag]

Very interesting Hans. Noticing how the cloud in the upper right disappears in RG4 but is present in RedlogFilm. A good case for power windows in a final grade... I have also noticed in RG3 shooting a white cyc that I cannot get white to even approach 100- seems to be limited to 93 or 94. Sounds like RG4 may be different and roll off right up to 99 or 100 IRE? Should be getting my Dragon very soon so I will test this too.

All valid points and important arguments to avoid RedGamma for finishing, regardless which iteration. Dailies are a different story, of course. And here the new RedGamma4 seems to be a good development since black is set to 0 and white to 100.

Hans

 

[Hrvoje Simic]

Noticing how the cloud in the upper right disappears in RG4 but is present in RedlogFilm. A good case for power windows in a final grade... .

No need for pulling down highs in the end if one has control to pull them up from log in the start.





Any single gamma will differently affect differing contrast distribution scenes, nicely fitting some while unsuitable to others so RG4 has less to do with this. This example does not mean that RG4 is less suitable in general, it just means that it is less suitable for this scene.

 

[Bob Gundu]

Good point Hrvoje. There is no magic bullet for gamma curves. Thats the beauty of shooting R3D's with a non-destructive approach. Generally the new Redgamma4 is very nice for adding contrast and cleaning up the mid tones. But a high contrast scene is better left going back to REDLogFilm and starting from there.