RedShark News lives in a world where the AJA Cion is the best thing since sliced bread. In this world, however, sliced bread is still king. Read on…
To say I’ve been critical of the AJA Cion is a bit of an understatement. While AJA has shamelessly promoted their first-try camera as one of the most amazing cinema tools ever, their demo footage has me wondering if/when the real camera will show up. I’ve not been impressed with anything I’ve seen, and there’s nothing that makes me think the Cion is competition for any other camera in its price range.
Imagine my surprise when RedShark News gave the Cion an absolutely glowing review while publishing stills that showed all the visual quirks that have convinced me it is nothing more than an average first try. Imagine my further surprise when, after noting that the Cion has no log mode, they then went on to publish a lengthy article about why log mode is almost completely unnecessary.
I feel like I’m starring in my own personal episode of “Sliders.”
My first introduction to the Cion was this video:
CION – Science of the Beautiful; Ungraded Images Reel from AJA Marketing on Vimeo.
I used a color picker tool (Art Directors Toolkit) and discovered the light-skinned model’s face in the opening shots has more blue in her flesh tone than she should: it should have mostly red and green with a touch of blue, but in this case the amount of blue in her skin exceeds the amount of green. That’s not right, and it’s why her skin looks cyan overall and reds are shifted toward magenta.
The dark-skinned model is equally cyan, with zero detail on the fill side of her face.
- Some other things I noticed in this video are:
- Assuming the interior scenes are shot using HMIs or daylight Kinos (which appears to be the case), this entire video was lit by daylight. Tungsten light is quite unbalanced spectrally as it contains lots of red and very little blue, while daylight’s red/blue balance is a bit more even. Cameras tend to perform better color-wise under daylight than under tungsten light, so it’s curious that there are no tungsten light samples in this film.
- Every shot takes place in open shade, presumably to limit contrast to a range that the camera can handle.
- There’s a shot of a window whose reflection shows clipped sky, which turns a most unpleasant electric flat cyan color.
- Shadows and dark tones lose detail quickly. I see any no detail in dark hair.
To my eye, this video shows the camera as unready for prime time. There are a number of clear issues with the image that are subpar compared to nearly all modern video cameras, and it appears that whoever shot this video went out of their way to shoot around those problems. The preferable method of shooting a camera demo is to play to the camera’s strengths, not avoid its weaknesses, but I don’t see any strengths here.
Then I saw this Cinema5D review at NoFilmSchool. This review is fair, although I disagree with their conclusions that this camera’s color rendition is “cinematic.” Their demo footage looks muddy and monochromatic to my eye. I don’t see any strong examples of subtle hue differentiation or strong, saturated yet accurate colors. I don’t see the “high color resolution” that the tester mentions. This could be a side effect of web compression, but I can see these subtleties in Arri Alexa and Amira footage published to the web.
One of my key factors in assessing a camera is to observe how well it can reproduce similar but slightly different hues. The Arri Alexa does this well, which is one of the reasons it appears so filmic, but the master of this may be the Panasonic Varicam 35. I haven’t seen more than a few Varicam demo pieces on the web but I’ve seen a couple of shots where fabric samples that are very close in hue stand out from each other beautifully. I’ve not seen any of that in Cion’s imagery.
I’m also concerned about Cinema5D’s use of the DSC Labs Xyla chart to test dynamic range. I own one of these, and whenever I test cameras I use it as the instructions dictate: expose the first chip at clip and then count the number of steps visible to assess dynamic range. Cinema5D seems to be measuring from the third chip down, and the reason for this methodology is not explained. Perhaps they are measuring from the point where the highlight slope reaches a certain steepness, but if this is the case then this should be noted as this is not how the chart is designed to be used.
I also noted that the first chip on the Alexa test is clipped so hard that it’s flaring, which implies that it is well clipped instead of just clipped as the chart was meant to be used.
Because of this non-standard and unexplained methodology it’s hard to know whether this camera has eight stops of dynamic range or ten. It certainly doesn’t have twelve, which is the absolute minimum we should expect out of a modern camera.
As I saw in the original video, at top, highlights turn cyan when clipped, which is a really unpleasant old-time video look. This isn’t acceptable in a modern camera.
The short film “Waterproof” confirms much of what we saw above:
This is a nice little film. Once again, though, I see that the subject is photographed in open shade or very diffuse low sun, apparently to keep contrast as low as possible. Dark fabrics disappear into blackness even under diffuse light, which is concerning as dyed fabrics are usually no more than three stops below middle gray in brightness. I see awful cyan overexposure in the sky in the second shot in the piece, and a worse example in skylight reflecting off water at 1:26. Reds in the background appear to shift toward orange, and flesh tone still has too much blue in it.
By the way, AJA really hasn’t gone out of their way to shoot gorgeous demo footage with this camera. Here’s Arri’s Amira showreel:
Arri is clearly targeting their camera toward corporate media production, which is a huge industry. AJA is targeting their camera at… who? Based on the demo material they’re showing I can’t figure this out. So far the target audience seems to be blueish fashion photography and home movies.
This brings me to RedSharkNews. They say, “The images coming out of the CION are extremely pleasing and cinematic right out of the gate without the need for extensive processing in post production.” This is after showing a series of stills on page one showing extremely contrasty day exteriors where the shadows seem to plunge off a cliff into inky blackness.
On page two there’s an image of a woman drinking coffee next to what appears to be a day-lit window. The window is so blown out and featureless that it reminds me of a Betacam from the 1990s, and there’s a cyan fringe around the highlights that I find extremely unpleasant.
Directly under this image there appears this line: “Highlight roll off was organic and smooth, and the brightest areas never seemed to burn out in a displeasing way.”
Time for a quick universe check… hmmm, in my universe that statement doesn’t match the example.
The article ends: “Combined with attention to the aesthetics of rich, vibrant color reproduction can really hit the sweet spot in terms of that elusive magic best described as ‘cinematic mojo.’ In the case of the CION, AJA totally nails it.”
Okay, let me put away my writing hat and put on my DP hat for a moment. Here’s my opinion based on what I’ve seen from the Cion so far:
This is the 2015 version of the Panasonic AF-100.
Budgets are shrinking, but we’re being asked to keep the quality of our work high. Everything is shot on digital now and film is a dying animal. Crews have to work faster than ever while not compromising on image quality. This is true not just in feature and episodic TV production but also in commercials and corporate marketing, which is where I make my living. I had a job a while back that required me to shoot 26 shots on seven sets, plus an exterior 40’ dolly move, in 12 hours. I shot it with an Alexa and none of the footage was color corrected before it went live. It looks great.
I could not have done that with a Cion. Nasty cyan clipped highlights are not acceptable, and I would have had to work hard to eliminate those on set. Crunchy shadows would have to be filled in, and while that was easily done in older times because the film stocks of the time required a higher lighting budget, they are a huge impediment now when most exterior shots are photographed so quickly that we often use no more than a 4×4 bounce card if the light is right.
I don’t see Cion’s “amazing cinematic colors,” but that’s a taste issue and someone may consider its look cinematic. I do see blueish and ruddy red flesh tones under daylight. Beyond this test I haven’t seen many examples of this camera under tungsten light, and—strangely—it seems to do better with clipped highlights under tungsten than under daylight. In fact, based on the Geoff Boyle/DSC Labs color saturation chart used in this test the results seem to show Cion has more overexposure headroom in tungsten light than in daylight. It’s curious that we aren’t seeing more tungsten light examples.
But this next article really surprised me. Here, Cion’s inability to record in log format is actually touted as a strength:
“There was a time when everything was shot to look good right out of the camera. This goes for both video and film. There is a real skill and confidence involved in deciding on a look from the outset and creating it with the use of lighting, both natural and artificial, and glass filters.”
Transferring film to video always required a grading process of some sort. There’s no other way to get film imagery onto video. And printed film didn’t look good straight out of the camera unless it was processed according to the DP’s wishes and printed with the proper lights on the correct print stock. There was lots of leeway, and sometimes it was so difficult to get proper dailies prints made that DPs would shoot their own tests and dictate printer lights to the lab.
Even in the film world it is wildly incorrect to say that the DP’s vision was automatically preserved all the way through the post pipeline. If the studio wanted the film to be brighter then it printed the movie brighter.
Also, the reason we had to make the video image perfect when we shot it was because it had very limited dynamic range and the format the images were recorded in did lot allow much in the way of grading. We were working within the limitations of technology. Expectations were different—video wasn’t expected to look as good as film did—and on higher-budget jobs we were provided the tools to paint the image (paintbox, waveform monitor/vectorscope) and the personnel (video engineer) to make it look the best we could make it.
None of that happens anymore. Producers and directors expect video to look as good as film did because film is now gone and HD cameras are awesome in their capabilities compared to what went before. And because of Alexa they expect us to be able to shoot any digital camera as we would a film stock.
Speaking of Alexa… this next quote caused my jaw to drop, hit the floor, bounce, and then drop again:
“The AMIRA and the ALEXA, cameras that AJA hopes that the CION will compliment, are both capable of producing a wonderful looking picture straight out of the box, should they be required to do so.”
The Cion is expected to compliment Alexa and Amira? I see nothing in the Cion look that comes close to touching the Arri’s dynamic range and color.
This video, posted recently on Reddit as an Arri Amira test, is vastly better looking than anything I’ve seen come out of a Cion:
The subtlety of color in that opening shot alone, with the cat’s pale green eyes offset against its slightly pink nose, blows the doors off anything I’ve seen come out of a Cion… and yet Cion will somehow “compliment” Amira.
Nothing that Cion has shown so far beats this casually-shot ungraded Amira footage of cats cleaning themselves, and yet Cion is meant to compliment Alexas and Amiras.
Time for another universe check…
On page two it gets even better: “When it comes to grading, the elephant in the room is that in order to obtain the best results from the enormous dynamic range that modern camcorders are capable of recording, the user needs a good deal of knowledge, time and skill when using the grading software.”
Perhaps you’ve heard of people who call themselves “colorists.” They do amazing things with grading tools because that’s all they do, and they’ve done it for years. They have “the good deal of time, knowledge and skill” that it takes to do a good job. Just as buying a camera doesn’t make one a cinematographer, installing color grading software does not make one a colorist.
The rest of the article is… interesting. It treats grading as a way to decide on a look later in post rather than committing to one at the start of a project, but ignores the fact that in-camera processing is decidedly limiting. Yes, many cameras look great out of the box—I’ve shot many projects with Alexa where the Rec 709 look was baked in and never touched again—but for precision work nothing beats sitting in a grading suite with controlled lighting next to a trained colorist working off a beautiful monitor. The point is not to decide a look in post, but rather to create a better look than one could ever do on location given the time and budget constraints we currently have to work under. The necessary controls for modern commercial and feature looks don’t actually exist inside modern cameras, and a field-quality LCD monitor on a brightly-lit set is not the best place to critically craft a subtle look. We do this frequently, but it’s not ideal and a proper grade by a proper colorist results in vastly better results and in less time than I could do the same thing myself.
The seed for this article is the fact that Cion doesn’t offer a log mode, only a flat mode. “Flat” implies that a low contrast S-curve has been applied that allows for grading but doesn’t require it. That’s nice, but it’s not as efficient at storing tonal information as log is. Most importantly, it still rolls off highlights and shadow details more aggressively than log does, making them harder to recover. One of the huge benefits of log curves is that it gives us access to tonal detail right at the edges of what the camera can see. Linear-encoded raw does this better, but log is still very, very good.
Based on what I’ve seen so far, recovering highlight and shadow detail is more than a minor concern with Cion.
I’ve not read RedSharkNews much. Once in a while someone will send me a link and I’ll take a look, but I’m always confused by what I read. They really do seem to be living in a parallel universe as their experiences with cameras do not match mine at all. Writing that Cion has beautifully rolled-off highlights just below a picture that shows horrible, featureless clipped highlights with cyan fringes makes my cognitive dissonance alarm go off at full volume. Calling images “cinematic” that are so contrasty that details in backlit open shade are lost in deep shadow makes my teeth hurt. This is a positively glowing review of a camera with less dynamic range than a Canon 5D, has horrible highlight handling, can’t see into shadows that a Sony FS7 doesn’t blink at, and whose missing log mode is a “feature.”
Like the AF-100 I have no doubt the Cion will make pretty pictures under perfect conditions in a rigidly controlled environment, but how often does that happen? It’s true—within reason—that a good cinematographer can make pretty pictures with anything, but the length of time it takes do to so will vary considerably based on how good their camera is. As best I can tell, Cion will be a major time suck that brings us back to the days of having to flag every highlight and fill every shadow. Given how economical we all have to be these days on set, I don’t see that being a successful strategy.
I wish RedSharkNews the best of luck with their Cion. In their world it sounds like it makes marvelous pictures. In my universe—based on what I’ve seen so far—I’ll be shooting with just about anything else.
Art Adams
03/31/2015