First DRAGON clip down, many more to come. Before the DRAGON sensor updates start rolling out in big numbers this fall, we should be getting a lot more footage. DP Evin Grant got a little fire-breathing monster to play around with, and he's posted his results online, along with some stills. Tom Lowe, who you might remember from Timescapes, has also received a DRAGON to test, and he's posted a 6K still so far, with likely much more to come. Check it out below, and be sure to head on over to the comments for what is always a fun time.

Some info on Evin's test, followed by the video. Be sure to head on over to Vimeo to download it in higher quality (emphasis added):


This camera's color science is beta, in fact the version of RCX I used is still using RG3 & RC3 so they are not yet matched to the immense DR and color info of these files.

This camera also had a brand new IR filter installed and no adjustments had been made to the CS or RCX for any impact that might have on color rendering.

All of the above meant I felt in order to be fair to the camera and sensor I needed to do some color correction beyond RCX, to that end I did a single primary pass in Resolve on the Prores 444 output.

The kelvin and tint settings noted in the video are my approximation of the ambient lighting according to the eye-dropper in RCX, I developed at different settings when I though it looked best. I also pushed the color saturation as far as I felt I could before looking garish, this is not actually my preferred taste but I felt it was important for the test.

Last off I am not a colorist, I'm a DP and I'm sure a professional who works in Resolve every day could do an even better job given how much color info is in there.

Here are some of his thoughts:

My feeling is there is more DR visible here than the stated spec, but it's hard to quantify. You'll also notice I was very good about avoiding any hard clips. Even the fire ball shots hold color. Another thing is I'm really in love with the grain-like noise when the sensor is pushed to ISO 3200+ it's really organic and gives the truly high ISO material a beautiful pushed film quality. It also handles mixed and discontinuous lighting like nothing I've seen. The shot of the girl dancing at 6400 ISO is shot under just an LED street lamp and plays much brighter than my eyes were seeing.

As far as skin tone the sensor picks up all the subtle colors and variances in the flesh and make-up, this is the trade off for so much info and detail and if you really want smooth skin rendering you'll need to have a great make-up artist and really understand filtration. The only diffusion filter I used on some shots was a Schneider 1/16 HD classic soft just to take the edge off.

Check out some selected stills (click for larger):

Metrogirls

Uniceecu

Timwalk

Timface

Sophia

Elevator

The elevator shot has some atrocious looking light in the elevator, but the amount of detail being held is impressive. I think it's also important to realize that while there is noise at 4000 ISO, neither the camera nor the post software is doing any noise reduction, but cameras like the Canon C300 (and many others) are doing all kinds of noise processing internally to reach such crazy high ISOs. This is also similar to JPEGs coming out of a DSLR, which have gone through a number of steps before finally reaching the CF/SD card.

Here is a 6K 4000 ISO still from Tom Lowe's shoot (click for larger):

Nvrt

Again, there is noise but when you consider that it's coming from a 6K motion file, it's mind-blowing. We're spoiled, let's face it. We'll never be happy, even with noiseless 16K images at ISOs over 100,000. It still has a distinctly RED look, but from everything we've seen, and everything that has been said from those who have used it, there is a ton of information to work with. If you're lighting a project, it's rare that you need ISOs this high, but for stuff that's more unpredictable, having this capability is fantastic.

It's unfortunate that it's using old color science. Clearly this sensor is better than MX, but by how much is the big question many are currently asking. Is the current color science holding back the performance? Probably enough to notice a difference, which is why it is surprising they are releasing footage so early. I think if RED really wanted to knock the socks off potential buyers, a Peter Jackson-style short like Crossing the Line would have done it. They are taking the real-world approach where a number of people get the camera for a little while and post whatever they want without any input from RED. That's fine if you know what you're looking for, but they need to impress many more people than that. With how much vitriol has been thrown at the company thanks to an inability not to hype its products, it's a little surprising we didn't get an amazing short film to show off what it can do on a controlled set with proper grading (maybe one is coming).

There should be a lot more soon enough, and Tom has also mentioned he'll be shooting a side-by-side test with MX. I'm sure that's the kind of test many of you have been waiting to see. Feel free to discuss in the comments, though staying on topic is appreciated.

Links: