Question: What is in this reel that shows me what an amazing camera the Cion is? Answer: nothing. That doesn’t mean it isn’t amazing, but if it actually is amazing their marketing department is doing its damndest to keep it a secret…
I hate to admit it, but bad marketing drives me nuts — especially when it’s by a company that has never made a camera before and wants to convince me that I should buy, and use, their first attempt. Wouldn’t you go out of your way to shoot awesome footage that shows how uniquely this camera solves the everyday problems that vex us all?
I sure would. AJA apparently wouldn’t.
If you haven’t seen the footage in question, take a look. Then we’ll tawk.
If the embedded video doesn’t work, click here.
When AJA says this is “ungraded” my assumption is that it was created entirely in camera: the crew white balanced the camera and exposed it properly, but nothing else was done. Some sort of built-in color matrix or LUT was applied as the material was recorded and it wasn’t touched in post.
I kinda wish they had touched it in post. I really want to like new cameras, but the footage has to be promising, nay, spectacular, to really get my attention. Nothing I see in this reel meets that standard.
I can’t escape the feeling that this was shot by someone who is primarily a still photographer. Most of these setups feel like still fashion frames with movement in them. There’s nothing wrong with that, I suppose, but it doesn’t feel as if the material is targeted at someone like me: a working cinematographer.
There’s something else going on here that I can’t put my finger on. From the highlight in her eyes the source looks reasonably large, and from the color I’m guessing it’s an HMI that’s either bounced or pushed through a frame of thick diffusion. The model’s face looks magenta, which is not a good look. There’s a hard fill light from the left that’s so far around that it’s casting a hair shadow on her face, but as best I can tell it’s not affecting her flesh tone much.
The Vimeo description says “The studio shots were captured using three point lighting” which is not exactly something you’d boast about to anyone who has been in the business for a while.
It’s hard to make an LED light with just the right amount of green in it (excluding remote phosphor designs), so rather than take a chance on a green spike some manufacturers will overcompensate and pull a lot of green out. Too little green in a light makes it appear magenta. If the light in her eyes was smaller I’d assume she was lit with a tungsten LED panel, as those are most likely to show green or magenta color shifts. My best guess, though, based on the coolness of the background, is that this is lit by an HMI key. (I feel okay using the word “key” in this context because, you know, three-point lighting.)
In my article on Panasonic’s latest demo reel I used a color picker to examine the RGB values in images to confirm that the hues I saw were actually present and not a trick of my monitor or my eyes. When I sample this model’s forehead it comes up as RGB 154, 121, 131, and that’s way too much blue for flesh tone. Either the light is blue or the white balance is off, or maybe the camera matrix overcompensates for the lack of red in daylight sources by punching red hard.
The background looks slightly cyan to me, which is confirmed by my color picker (111, 118, 122–not crazy cyan but enough to look cool and not neutral) so there’s definitely something wrong with either the color of the light or the camera’s response to it. This is not something one wants to wonder about while viewing a new camera’s demo reel… and this is only the first shot!
If the warm colors in the background are dull and the neutral tones seem cyan–both of which imply either a poor white balance or an old HMI bulb–why is her face bright magenta? As best I can tell it’s an interaction between the blue light and warm makeup.
The colored background and the model’s dark skin make it difficult to see what’s going on here, but the specular highlights in her cheek and forehead feel cool to me. My color picker says that, like the previous model, her flesh tone has more blue in it than green (RGB 97,79,85) and that’s simply wrong for flesh tone. She appears to be lit by the same light as was used for the previous shot. She looks a lot better, though, as dark skin can be lit with a wide variety of colors and still look amazing. Caucasian skin is not that forgiving.
Why is it that, when she turns to her left, her near check turns rosey red? That’s just not natural. Either she’s wearing a lot of makeup, or that “fill light” from the far left is actually a tungsten light. To find out I zoomed in to the Vimeo HD version of the video as far as I could:
The fill light is a brighter point source, but it appears to be the same color. (Isn’t the fill light traditionally bigger and softer than the key?) As best I can tell, she’s just wearing a lot of makeup. That’s not really a good idea for a camera test: you don’t want people to wonder if what they’re seeing is the result of the camera’s colorimetry or dyes adhering to the talent’s face. A little makeup is okay, but this doesn’t look like a little makeup to me.
The lack of a background in this shot helps us see more clearly what’s going on with this model’s flesh tones. She’s nearly purple. Her forehead is RGB 163, 118, 134, which confirms what my monitor seems to be showing me: there’s a lot more blue than green in her skin. (I’m using an NEC display that was calibrated at the factory, so it’s not perfectly Rec 709 but it’s pretty close.) Flesh tone is primarily red and green; there should be very little blue in it. My color picker says that her hair light is very slightly blue, so this may be a white balance issue.
Canon takes great pains to make flesh tone look great in every circumstance. Arri and Sony do this as well, although not to quite the same degree. (Canon sacrifices some color accuracy in reds to ensure flesh tone looks good no matter what. A Canon sales rep once told me, “We get flesh tone right at all costs. All other colors are optional.”) It would be nice if AJA gave this a try. I’m not sure why this shot made it onto a demo reel: the 4K detail is impressive, but the color of her skin is just wrong, with way too much blue in it (or not enough green), and it exaggerates imperfections in her skin. I feel as if I’m looking at some sort of medical exam documentation: “This woman was not harmed in any way during the making of this film.”
Here’s our model again, and it’s hard to tell how she’s lit. The light source appears to be mostly overhead skylight with some sunlight bounced off natural surroundings in the distance. I confirmed this by looking deep into her eyes:
I see some blue sky in there, but also something that looks like sunlight bouncing off trees or maybe a building. I guess the conditions of the test were too rigorous to consider using a neutral white bounce card. That might make her flesh tones look… well, normal. That doesn’t seem to be the goal here.
It’s hard to blame this look on an old HMI bulb as I don’t see a highlight in her eyes that tells me there’s a bright light nearby. I’m starting to wonder if the camera has some sort of white balancing or matrix issue that prevents it from rendering reds properly under blueish light. (That makes one wonder how it does under tungsten light. Maybe we’ll see some later on in the reel? SPOILER: The answer is NO.)
There’s not a lot of subtlety or depth to the colors i’m seeing here. Her face is almost uniformly magenta–her cheek on the far side of her face just under her right eye is actually rouge instead of magenta–and her scarf is a non-subtle orange. Parts of her hair look a little cool, as if it’s reflecting skylight.
If this was indeed shot primarily under 8,000K-10,000K skylight, I have to wonder: how was this supposed to make her look great while showing off a new camera?
The shutters in the wall actually match the hue of her jeans perfectly. Blue light tends to make everything look a bit dull except for things that are already blue, so I’m guessing that this really was shot under high color temp blue sky. Still… what are the odds that the color of her jeans and those shutters would match so perfectly? I’m starting to wonder if this camera has a hard time discriminating between blues and greens, and as a result mixes a bit of green into anything that’s blue. I can’t otherwise explain why those two hues match so perfectly.
In my Varicam demo reel critique I noted that Panasonic colorimetry is incredibly adept at resolving subtle nuances of hue. I don’t see that here. The grass is all one color with only some variation of tone; all the blue hues in the shot are exactly the same. I can’t tell if the camera sucks at resolving color or if there’s just no color in the shot to resolve.
Once again, these are not questions one should be asking while watching a new camera’s marketing demo reel.
Out of desperation, I zoomed in on the dark-skinned model’s teeth to check white balance. As her flesh tones look beautifully warm, and the caucasian model’s skin looks like she’s been in the oven a bit too long, and neither of their shirts appear to be a neutral color, I needed something to tell me whether the camera’s white balance was accurate or not. My color picker says her teeth are perfectly white, so it appears that the camera is reasonably white balanced. The skewed colors appear to be either a LUT or matrixing issue.
Apparently someone managed to round up a bounce card for this shot and stuck it out in some sunlight. The left side of the model’s face is a little warmer and there’s clearly some shape to the light, indicating that it’s not simply ambient bounce off some distant trees. Still, though, her flesh tone has more blue in it than green: I can see that by eye, and the color picker says so as well (RGB 172, 119, 125). There’s less blue present in her flesh tone than in previous shots, at least on the warm side of her face, but there’s still too much.
The cool highlight in her hair implies open sky overhead. And, for open sky, it’s not reading blue so much as cyan. The dark part of the wall at the top of frame has equal parts blue and and green in it (RGB 39, 50, 54) and that’s odd for open sky. Some camera manufacturers like to add a lot of green to blues (or they can’t manage to remove enough green from blues) to the point that skies become a beautiful and dazzling cyan that’s also utterly unreal. I’m wondering if something similar is happening here. I’m not seeing bright blue hues in this footage without seeing a lot of green mixed in at the same time.
This model’s face is beautifully warmed by the bounced sunlight, although the color picker says that blue and green are at equal levels instead of green leading blue the way it should. My gut says the colors in her shirt should stand out more: the red around her neck is really saturated, but it doesn’t grab me for some reason. It’s certainly red, but I don’t sense any subtlety or delicacy in its reproduction. (It’s also a blueish red, which may be the actual hue of the fabric or contamination from the sky. Again, not a question you want to ask yourself while watching a camera’s marketing demo reel…)
Look how cool the background is, and notice the cyan reflections in her forehead. There’s plenty of skylight contaminating the shot from overhead… and it’s not reading as blue but as cyan.
She also has the darkest hair ever. I see no detail in it at all, even in the parts that should be lit. This could be due to web compression, but still… I don’t see a single instance in this video where her hair has any detail in it unless she’s standing in front of a bright background and the strands are silhouetted.
I can’t help but wonder why every shot so far has taken place in open shade. This model’s hair has almost no detail in it–honestly, the blacks look crushed–and there are no bright highlights to be found. What is the actual dynamic range of this camera? Surely 12 stops of dynamic range is enough to show us some sunlit footage…
We don’t know what color any of these things really are, other than the leaves, so… hmmm.
I’d normally assume that the window frame was white, but it’s on the greenish side of cyan. This seems to play into my theory that this camera renders saturated blues as greenish, as I’d normally assume that a white window frame would appear bluish under skylight. Maybe it’s actually a greenish white…?
There’s a patch of sky reflected in the top of the window, and when I zoom in…
That’s sky, and it’s definitely not blue. In fact, my color picker says that both the blue and green channels are clipped. There’s no subtle roll-off into overexposure here, and no gentle desaturation of highlights as they approach clip. Instead, something that’s supposed to be a pale blue color–or at least a neutral clipped white–is positively electric cyan. That’s just wrong.
This is the nicest flesh tone so far. When I went in with the color picker to find out why I discovered that, for only the second time in this video (and the first time for a full face), the amount of blue in her flesh tones didn’t exceed or match the amount of green (RGB 133,93,83). Her flesh tone looks fairly nice.
The Vimeo description says that this is the only exterior shot that employed a light. There does seem to be some sort of light source in play here:
This could be a gentle kiss from a daylight LED. Whatever it is, it’s not significant enough to impact flesh tones compared to a similar shot (see below).
Here’s another still photograph in motion… literally. It’s interesting how her flesh tone looks pretty good in these two images, and they’re also the only images where it’s obvious that the sun is present. It seems that this camera can resolve flesh tone reasonably well in the presence of neutral daylight… so why was everything else shot under open sky? What is the true dynamic range of this camera?
It looks like there’s something bouncing a little sunlight into her face in this shot. I don’t see the pinpoint of light in her eyes that I saw in the previous shot, which was noted as the only exterior shot lit with an actual light, but there’s no effective difference between flesh tones in this shot and the previous one. There does seem to be something lighting her, but it’s impossible to determine whether it’s a sunlit bounce or a small light through diffusion.
Note that there are some clipped highlights in the leaves behind her…
This camera doesn’t seem to handle clipping gracefully. I don’t see any exposure or saturation roll off: there’s dark green, light green, a ring of yellow, and then white. Blech.
Some of this ringing could be due to 8-bit web compression, but if so I’d have worked a bit harder to fix it or use another shot. And 8-bit compression would likely have eliminated some of the delicate transitions between tones, but that yellow ring doesn’t look like a compression artifact to me. That’s indicative of a color channel clipping in the camera.
Last time I saw something like this was in footage shot by a Panasonic AF-100.
I’m not sure I understand the point of this shot.
Okay: tell me if you see a difference between this waterfall and the next…
One side of her face looks natural, as it does in the previous two sunlit shots, so I’m guessing that there’s some bounced sunlight coming in from frame right. (It can’t be a light… AJA says only one light was used in shooting an exterior, and we’ve already covered that shot.) But… her hair is green and the water in the background is cyan, so my guess is they white balanced to the bounced sunlight and let the rest go. This seems to show that the camera has a hard time separated greens from blues, as the water in the background is definitely cyan even though it is lit by blue skylight.
It’s clearly the same waterfall in both images. One was apparently white balanced for skylight and the other for bounced sunlight, but wow… she looks like she’s hanging out in a fluorescent-lit laundromat in the woods.
Her scarf is (dull) blue, and her shirt seems to be (dull) blue, but her hair and the water behind her are an almost electric cyan. Is this really the kind of image you want to show in a marketing demo? (Spoiler: NO.)
I suspect this is supposed to be beautiful, but the clipping hurts my eyes. There’s no subtle transitions between clipped and unclipped. It looks very “video-y” to me.
I don’t know that the Cion is an awful camera. I don’t know that it is a good camera, either. This demo has me completely confused. If I were supervising this project I’d throw out all but one or two shots and start over. Most of the flesh tones in this piece look too blue. The things that are actually blue look too green. It appears that most of the footage was shot under extremely blue light that doesn’t have much, if any, red in it to begin with, and I’d think you’d want SOME red in your light source if you’re shooting beautiful models with perfect skin AND you want them to look good.
I do know that the clipping artifacts, pervasive magenta/purple flesh tones, and the fact that there’s not a bright pure blue anywhere in the piece that doesn’t have a lot a green in it, tells me that maybe this camera’s color science isn’t ready for prime time. RED went through similar phases during its development so I know these issues are surmountable, but… this is a marketing demo reel released shortly before a product launch! Come on! (To be fair, RED had already released their original cameras when similar issues popped up.)
What makes me extremely suspicious is what’s not here. I don’t see any tungsten light examples. Tungsten is very hard to get right in video cameras as silicon is naturally most sensitive to red and least sensitive to blue, which means it does better in light where there’s some, but not a ton, of red and lots and lots of blue.
Sound familiar? I just described daylight. Tungsten light is spectrally imbalanced between warm and cool, and some cameras struggle significantly with that lack of balance. Nearly every camera, however, does just fine under daylight, as the ratio of warm spectrum to cool isn’t equal but it’s not nearly so unbalanced as tungsten light. I can’t help but wonder what the problem is when I see a new camera’s demo reel and it appears to employ zero tungsten light sources.
The interiors could have been shot under tungsten lighting, but in that case flesh tone shot under tungsten is just as bad as it is under daylight. I don’t think that’s the case, though: even in Arri, Sony and Canon cameras you’ll see subtle hue saturation differences between daylight and tungsten exposures: warm colors will be a little punchier under tungsten and cool colors will be a little punchier under daylight. (Arri has a great demonstration this here.) The studio and exterior flesh tones, but for a couple of shots, look exactly the same.
I also don’t see any shots that are particularly high in contrast. In the couple shots that do show some background leaves clipped in sunlight I see really garish highlights that I don’t find pleasant at all. The shot of the window shows clipped sky reflections with awful color distortion. The only reason I can think of to shoot this under open sky is to reduce contrast, as it’s much like shooting on an overcast day: there’s no sunlight raking through the shot, just broad illumination without much variation in contrast.
I, personally, would never try to show off flesh tones under light that had almost no red spectrum in it, but for some reason AJA did. There must be a reason, and a burning desire to keep contrast low is all I can come up with.
To sum up:
- We saw AJA tout this footage as really impressive ungraded imagery straight out of the camera.
- We saw beautiful models lit primarily with bright blue light, most of which was skylight but a couple of shots appeared to be lit by HMI light.
- We saw no scenes that appeared to be lit by tungsten light.
- We saw lots of magenta, and almost purple, flesh tone.
- We saw normal caucasian flesh tone in two, or maybe three, shots.
- We saw unpleasant, clipped highlights with color distortions in the few shots containing bright sunlight, or reflections of the sky in glass.
- We saw what appear to be crushed blacks, although that could be due to web compression… but we also didn’t see anything that showed off how the camera can see into deep and subtle shadows.
- We saw cyan everywhere. LOTS of cyan.
- We saw some shots of plants, water and dead leaves under various white balances.
Last but not least, this video appears to be shot by a still photographer, and while I have nothing against still photographers most of them don’t know how to shoot motion well enough to impress cinematographers. And many aren’t particularly good at editing, which is another area in which this project suffers greatly.
Does AJA know how jaded our industry is, and how hard it is to impress seasoned cinematographers? Does AJA actually understand the market they are selling to?
The Sony FS7 is due to be released around the same time as Cion, and at around the same price point… which one would you buy? I’ve seen video from both, and I’d buy the FS7. The FS7 demo video–which is okay, but not great–completely blows this Cion footage out of the water. There’s nothing like the FS7’s delicacy of color rendition and beautiful neutral flesh tones in Cion’s ungraded demo reel.
Compare and contrast:
While we’re at it, take a look at Panasonic’s Varicam 35 demo too:
One thing I do know: with a catch phrase like “Cion: Science of the Beautiful,” someone in AJA’s marketing department demonstrated a hell of a sense of humor by letting this footage out into the world while also boasting that it was “ungraded.”
Disclosure: I have worked as a consultant, both paid and unpaid, to Sony, Canon, Arri and Sound Devices (who competes with AJA in other areas).