comparison

Auto Added by WPeMatico

Pentax vs Contax: Testing 2 Go-To Film Cameras for Wedding Photography

Today we’re going to take a look at two film cameras. We have the Pentax 645nII and the Contax 645: two of the last medium format film bodies ever produced, and the two most popular go-to cameras for wedding shooters who are still shooting film.

There are lots of reasons why this is the case. While they don’t have fancy Eye autofocus or anything like that, they both have pretty good autofocus systems, very fast motor drives, and all the basic bells and whistles you could get on a film camera in the early 2000s.

They also have a meter inside. So if you want to use a meter you can shoot quickly, you can actually rely on the meter, especially if you’re shooting negative film where you have a lot of latitude.

There is a price difference between these two. The Contax is way more expensive at around $3,000, whereas you can get the Pentax for around $1,000. However, the Contax has a Swiss Planar T Zeiss lens, which is an incredibly sharp lens, and you also have a removable back on the Contax.

Pentax does not have a removable back, which means you have to fire through your whole roll of film before you can change. That’s a huge bonus for the Contax.

Another thing worth mentioning is that the lens on the Pentax only goes down to f/2.8, whereas the lens on the Contax goes down to a f/2. A lot of people think that that’s kind of a big deal. They like the idea of shooting at f/2, especially with the medium format. So I’m interested to take some portraits and compare the two to see how much of a difference it makes.

But aperture aside, the Pentax lens also doesn’t feel quite as nice as the Contax, and it shows up in the image quality. Speaking of which…

Image Quality Comparison

When comparing images from these two cameras, right off the bat, you do see there is a little bit more warmth in the Pentax. This image is a little out of focus with the Pentax unfortunately, but you do see that warmth there. And a little more contrast with the Contax Zeiss Lens.

Of course, you are scanning these things after processing the film, so whoever is making the decisions on these scans can push them around a lot. It’s just like being in Lightroom or Camera Raw. I would guess you could make these things match pretty close if you want to, but right out of the gate, with no additional work, the Contax has a richer, heavier color, heavier blacks, and heavier contrast than the Pentax.

Here we have our shot of the Capitol Records building. It is out of focus again on the Pentax.

We were at a distance here, and at this distance with the background that far away, the way the two lenses fall out of focus feels very similar. They’re both set to f/2.8, and they feel very similar, although there is a difference in the character of the bokeh.

If you look at the bokeh on the Pentax on the right it’s more oblong, whereas the bokeh is more round on the Contax.

When we were out working, the Contax was a little slower on the autofocus, even in bright sunlight. When we came back indoors and tried pointing both cameras at things here inside, that’s where the Pentax really pulled ahead. Both cameras would hunt, but the Pentax could actually find a subject and lock on somewhere with pretty much every frame. The Contax never stopped hunting, and because it can’t autofocus it can’t take an image.

I would say autofocus in a low light situation with the Contax is unusable. The Pentax seems a lot more usable in lower light and the viewfinder is brighter in the Pentax. So the ease-of-use in low light is better with the Pentax. That’s quite an advantage…

But back to the image comparison.

In this image, you can really see the difference between f/2.8 on the Pentax and f/2.0 on the Contax—it really falls out of focus much quicker, rendering a much prettier background with beautiful bokeh. I also love the way it resolves the lines around the trees.

It’s not all due to the aperture though, as you can see in the image below:

These are both shot at f/2.8 with the same distance between the subject and the background and the same distance between the subject and the camera. For some reason, the Contax falls out of focus faster letting the subject really pop out of the frame.

In contrast, this is probably the image where they look the most similar, although you definitely see a little more detail in her skin with the Contax. You get those nice lines in her face, and just a little more definition. With the Pentax, you get a much softer look—it feels almost like you’ve softened her face in post.

And here’s our final image. Unfortunately the Pentax was (again) not quite focused right here—it’s not focused on her right eye, it’s focused on her left eye.

But again, you see a little bit more warmth with the Pentax through the leaves, and more fall-off and contrast from the Contax with the Zeiss lens. It’s really a very pretty image, especially from the Contax with the way it falls out of focus.

Conclusion

If I was a wedding photographer and I was doing this seriously, I would definitely get a Contax. Yes, even with the autofocus being unusable in low light, I would just manual focus the Contax. The image quality is worth it.

I just think the Contax looks beautiful. The images have a very pretty look.

That’s not to say the Pentax is a bad option. At 1/3 the price, and with better AF, I feel like the Pentax is a great option, especially if you’re new to this. If you’re trying to get into film and get into the medium format world, the Pentax it’s super affordable, and it’s one of the newest medium format film cameras you can buy. I’ve seen them as low as $650 or $700 for the body.

The truth is, you can’t go wrong either way when it comes to these two cameras, it just depends on what your needs are. What do you want to spend it on? If image quality is king, consider spending the extra money on the Contax; if ease-of-use is more important, and budget it limited, the Pentax is a great option.


About the author: Jay P. Morgan is a commercial photographer with over two decades of experience in the industry. He teaches photography through his company, The Slanted Lens, which runs a popular YouTube channel. This review was also published here.

Canon EOS R5 vs EOS R: Image Quality Comparison

Wedding and portrait photographer Alex Barrera recently conducted a side-by-side comparison between the new 45MP Canon EOS R5 and the two-year-old 30MP Canon EOS R. Setting aside video performance, he wanted to see how these two camera stack up when it comes to pure image quality.

Barrera wanted to test three main things: overall image quality straight out of camera, dynamic range performance in a challenging backlit situation, and low light performance. For each of these broad scenarios, he enlisted his wife to help and they went out to capture portraits in a variety of locations and lighting conditions using the same lens on both cameras (the RF 50mm f/1.2L USM) and trying their best to capture the exact same frame.

Rather than steal Alex’ thunder or impose our opinion on the process, we’re just going to include the files below for you to peruse for yourself: first, SOOC JPEGs converted straight from RAW files; and second, edited versions that show how these files respond to manipulation in Lightroom.

SOOC JPEG Comparison

Here are all the side-by-side (or, rather, top-by-bottom) comparison of straight out of camera (SOOC) JPEGs that were exported as-is from the original raw files at 100% quality. For each duo the Canon EOS R is on top followed by the same photo captured with the same settings on the EOS R5 (click for full resolution):

Edited Photo Comparison

Once edited and exported at a lower resolution, it’s much harder to tell the difference. That may seem like an obvious point, but it’s important to note. If the R files fall apart under editing, or the R5 files handle adjustment better, this is where you should be able to see that.

As above, each set is Canon EOS R first followed by the same photo captured with the EOS R5 (click for full resolution):

There are lots of interesting things to point out about these images: from the surprising difference in white balance and exposure when shooting with the same exact settings, to the low-light performance and how far you can safely push the new R5 files in post. Then there’s the fact that the EOS R still holds up SO well, even when you put it up against the latest and greatest from Canon.

But don’t take our word for it. Check out the full comparison up top or download and explore the Raw files for yourself at this link.

There are lots of reasons why you might want to pony up the extra cash for the R5, even if you don’t shoot video. There’s the dual card slots, faster continuous shooting, extra resolution, and IBIS… just to name a few. But when it comes to pure image quality, the gap between these two cameras may not be as wide as you might imagine.

(via Fstoppers)


Image credits: All photos by Alex Barrera and used with permission.

Super35 or Full Frame? // C300mkIII & C500mkII Comparison

Recently I was given the opportunity to review the C300mkIII, Canon’s newest cinema camera. The C300 comes in hot on the heels of the C500mkII, and shares the same body and most of the same features. The highlight feature of the new camera is a S35 “DGO” sensor, which behaves similarly to the ALEXA in-that it takes two readouts of the image to output a picture with generally less noise in the shadow areas. It also touts an oft-talked about 120fps in 4K reading the full width of the sensor, in raw or compressed formats.

After using the camera for a decent amount of time, I found that I was basically repeating what I had to say about the C500mkII: I love the ergonomics and shape of the camera body, the internal NDs and XLRs, the customizability, the modularity, and of course the image. Seeing as the only real difference between the C500 and this new C300 is the image, I decided this article would be better served as a comparison between the two, highlighting why you might choose one over the other, and going more in-depth with certain concepts that can apply to any camera. This is going to be a long one. A donut hole inside a donut’s hole, if you will.

Super35 or Full Frame? // C300mkIII & C500mkII Comparison 2

SIMILARITIES

As I said, the two cameras are incredibly similar. It’d take longer to go over everything they have in common than to highlight the few things that are different, but I’ll go over the things I like about this new generation of Canon Cine Cameras as a primer:

I like the size. It’s right around ALEXA Mini/RED territory, which is likely the new standard for cine cameras going forward. When the original ALEXA came out, some people thought making a bigger camera was better/preferable. This can be the case if you need the room for additional hardware features or cooling, but once the DSMC2/Mini really took over, we’ve seen that smaller is clearly preferable. It’s funny, people rig up their DSLRs to look more impressive, where at the higher-end people are trying to strip the camera down as much as possible. The nice thing about these new Canons is that you get the benefit of a “Mini-style” camera body with all the benefits of a larger camera such as internal NDs, SDI/HDMI, and full-sized XLR jacks. Between those three camera bodies, the Canon is easily the most useful, plus the ALEXA Mini doesn’t have the modularity of the DSMC2/Canon. The RED Ranger series is more akin to the new C3/500 in feature set in my opinion (and hopefully I’ll be able to review one here soon!)

They boot up in like 3 seconds, switch to S&F modes instantly, and black shade in ~30 seconds. I also like how I can load in custom LUTs and even adjust the picture profiles as I see fit, allowing for a decent amount of customization of the image in-camera.

Both of these cameras have 1.3x and 2x anamorphic modes, but the C500 gets you a 20.1mm tall sensor, whereas with the C300 you’re at 13.8mm. In this case you absolutely would want the additional real estate to project that squeezed image onto because your horizontal resolution is going to be greatly reduced.

Feature-wise, you get a lot of bang for your buck. The aforementioned NDs/XLRs can easily sell me on a camera; I want a camera that needs as few extra add-ons as possible and those two things alone are huge. The left side of the new Canons is home to a series of back-lit buttons that are all customizable. Most of the buttons are pre-set to your most used functions (ISO, WB, S&F, LUT, Magnification, etc) but I personally do change “Zebra” to False Color because I use that more. I don’t think I’ve ever really changed the shutter off of 180° so that’s a free button too. In the menus you’ve got a set of “favorites” pages where I’ve got my most-used menu items so I’m only a couple clicks away from changing (for instance) Resolution, Recording Format, Monitoring LUT, audio stuff, or doing things like formatting cards. I basically never have to go digging through the menus with those pages populated, and I really like that.

There’s also a ton of monitoring options, in regards to information and frame guides. You can set what info you want to see, the guides can be multiple colors/sizes/opacities/ratios, including custom ones you input manually, as well as whether or not the external area of the guide is opaque or transparent. There’s even three “stages” of monitor, which you can toggle through hitting the DISP button, allowing you to have, say, a clean version, a info-heavy version, and one with elaborate guides or whatever. The monitor is your oyster. You’ve got swappable EF/PL mounts as well, that is easy to change by yourself in the field (although I wouldn’t suggest doing it in a dirty area). For a minute there, only the C300 could record proxies while recording XF-AVC where the C500 would only do so when recording RAW, but after a firmware update they now both share this feature, thankfully.

120 FPS

So now the differences. I know a ton of you may be reading this just to learn more about the 120fps 4K. The S&F Mode on these cameras is achieved by pushing the appropriate button on the side and it switches over instantaneously. The screen doesn’t even blink. Where the C500mkII does 120fps in a S16 crop (at 2K DCI), the C300mkIII does it in 4K reading out the entire S35 sensor, however without the DGO processing. It can also do 2K 180fps in a S16 crop.

High speed is an interesting beast. There’s a ton of chatter about it online, some people making it seem like they film literally everything at 120fps or higher and DEMAND it be in every camera from cellphone to cine-cam. So the question then, is why would you need it?

Traditionally high speed was used for slowing down action sequences or giving you a “dreamier” look, but these days it’s more often seen in the realm of advertising/web content. Lots of “b-roll” online (which by the way, if it’s the main thing in your video it’s not b-roll) is shot at high frame rates, and the marketing team always loves to see their products in slow mo, but for traditional narrative work you likely won’t shoot anything faster than 24fps. Product stuff, you likely will.

As I mentioned, the C500 actually can do 120fps, it’s just going to be in 2K DCI and cropped down to S16. Is this ideal? No. Does it still work? Yes. Mostly. Again, it depends on the situation. If your entire commercial is going to be in slow motion, then of course you want as much sensor as you can get, but if it’s going to be an insert shot in your film you can likely get away with the S16 crop version with a little noise reduction (I found the default Resolve noise reduction set to 3 Frame, Chroma at 100 and Luma from 0-10 handled it fine). If it’s well-lit you won’t even need the noise reduction. Blown up to 4K it still looks okay, but obviously a bit soft and lacking some detail. For most b-roll shots I’m fine shooting 48 or 60fps in 6K or 4K or whatever. If I did need 120fps on my C500, pairing the S16 crop with my Sigma 18-35mm pegged at 18mm gives me a wide-enough frame that I’m not concerned about needing a ton of room, and the lens is so sharp/clean it holds detail rather well.

So how does the C300’s 120 look then? I mean, it looks great. What did you think was going to happen? You don’t get the DGO so the shadows are a little less pretty than the normal speed, but you can still shoot in RAW and like I said, a bit of NR goes a long way. In the video below I did shoot the C500 footage on a different day, and I was sunburned, so they look quite different. My apologies. Annoyingly (for me) Autofocus does work on the C300 in high speed, but not on the C500. Why?

 


SENSOR / FF vs S35


I will say, I was a bit taken aback when they released the C300mkIII so soon after the C500mkII. It appeared to me, as it did to many, that they sold everyone on the C500’s capabilities and then immediately released a camera with a lower noise floor and essentially the same features for $5,000 less. I was a little ticked off, as a C500 owner, to say the least. I started to question whether or not Full Frame 6K was worth the price difference, when I was perfectly happy with S35 4K, and the new DGO processing looked like a big deal (120fps would be fun too but I’m less concerned about that, as I mentioned above). I endeavored to get to the bottom of this apparent transgression.

Firstly, what’s the difference between Super35 and Full Frame in general? Technically one can match the “look” of one to the other rather easily by making a few adjustments to the shooting stop and focal length, so there’s no appreciable difference there. Really it’s the simple fact that a Full Frame sensor gives you more physical real estate to record on, giving you a cleaner/more detailed image (assuming the best from the sensor). Now, this is somewhat theoretical because you can still make a large, crappy sensor, but thankfully that’s not the case here.

Now you may say “shooting on a longer lens but getting a wider frame means there’s less distortion” but this is a red herring because if your wide lens is corrected for distortion, as many high-end optics are, that effect is negated. Also, it’s the distance to the subject that changes perspective, not the lens, so that’s out too.

You may also think that a larger sensor lets you get a shallower depth of field, but that is also incorrect; Depth of Field is affected by the focal length, distance to subject, and aperture, and these things can be corrected for. A 50mm lens set to f2.8 has the same depth of field -meaning the area that’s acceptably in focus- on a S35 sensor as it does with Large Format, it’s just that you get a wider Field of View on the larger sensor and therefore would need a longer lens to match the FoV you’re used to on S35, if you wanted to. All that being said, a larger sensor does (theoretically) have more physical resolution and fidelity, giving you a better data set for which to manipulate in post (which we’ll get to here in a bit), but it doesn’t give you an inherent “look”. That all make sense? Focal length simply changes the magnification of the image, where sensor size determines how much of that image you record.

So let’s apply this knowledge practically: Assume we’re lighting our set to an overall f5.6 and our lens is set to f2.8 with a .6ND on (two stops reduction). We’ll say our key is at an f8, one stop over. This is how I might shoot a nice portrait or interview or something (and did for the tests in the video), keeping everything light and giving the sensor plenty of data to work with while keeping my ratios consistent with the high-key image I want to achieve. Let’s also say we like the frame we get on the C300 with a 35mm lens. Well, with the camera in the same spot, to get the same frame, we’ll need a 50mm lens on our C500. This in turn will create a blurrier background because, while we’re still at f2.8, we’re now on a longer lens. The camera remains in the same spot. Again, the focal length does not affect distortion, so all we’re really changing here is our Depth of Field (due to the longer lens not the sensor), everything else remaining the same. This means that in smaller spaces, practically, a Full Frame camera does tend to allow you to get a bit more separation in the shot than you would have on a S35 or smaller sensor.

In our camera tests I used the same Nikkor 35mm for both cameras (to make sure there was no color shift or anything between lenses), moving the C500 in to match the frame of the C300, and we do see the perspective on Dani’s face change (some distortion) but that’s because we’ve moved the camera closer to the subject, not because of something inherent in the lens or sensor.

Now, in regards to lenses and how they apply to different formats, there are reasons to go Full Frame over Super35. As I’ve said, you’ve got more physical resolution on a larger sensor and on top of having “less” noise, this can allow for softer lenses to render as perceptually sharper, less contrast is needed to render an overall sharper image, the aforementioned lower DoF vs FoV (when lighting to a given stop), and in general you get to record more of a lens’ potential character; That Nikkor 35mm on the C300 had less going on in the corners because it was reading the center of the lens (the sharpest part), but is a little more fun on the C500 because we’re getting the majority of the imaging circle and thus more of the lens’ “look”. Now, as distortion can be corrected, so too can “character”, but as with anything it’s a choice not a mandate. As an example, the Sigma 18-35mm I love so much is only designed for S35; you can’t use it on Full Frame unless it’s pegged at 35mm and even so you end up with a bit of pin-cushioning. In the same way, something like the Arri Signature Primes are made for Full Frame, but if you put them on a S35 sensor you won’t get to see the full personality of that design because you’re only reading out a smaller, center-punched area.

I encourage you to read Art Adams’ post “Quantifying the Large Format Look” and the ASC article linked therein, and Steve Yedlin ASC’s pieces on Large Format Optics as well as, honestly, the rest of the articles on his site. The man knows what he’s talking about and studying those write-ups until you understand them intrinsically will make you a better author of your image. Speaking of Mr. Yedlin…

RESOLUTION

Can you see individual pixels in any image at 100%? No? Then K-Count is kind of irrelevant. Hot take, but future proofing isn’t really “a thing” anymore. Let’s say the future holds 90″ 8K televisions: kicking out your 6K or 4K image for that future 8K master isn’t going to appreciably change your image the way it would have back in the old MiniDV days upscaling to HD. I’ve rendered out my 1080p C100mkII footage to 4K plenty of times and no one has ever suggested my original capture medium was any different because the original image was clean and reasonably sharp. Depending on the viewing platform it may be blurrier than if I shot native 4K, but how would you know? I’m not going to display both images side-by-side, and furthermore the compression on the web  (where literally all but two of the videos I’ve ever made ended up) averages out the difference even more aggressively than the scaling.

Going from Standard Def to HD was a huge leap, but nowadays the resolution wars are more of a forensic endeavor than some drastic quality change. 2K is right around the start of the size where we can’t really tell a difference with an increase in K-count, from a practical perspective. Your audience can’t zoom in beyond 100% while they’re watching, so why whinge about how many pixels are visible way in there? Zooming in 10x on the C300 or C500 4K footage still doesn’t show individual pixels, it’s just blurrier.

From Yedlin (roughly quoted):

We do have to make camera and monitor pixels small enough that they can’t be seen, but once we achieve that goal, making them smaller no longer has any appreciable perceptual effect on the audience, and other factors become the chief or only players in the appearance of crispness or image quality.

A camera’s spatial fidelity, how precisely it records photometric data in a tiny area, can not be properly quantified solely by counting camera pixels because there are other attributes that play just as crucial a role such as noise, grain, halation, and optical aberration.

For the purposes of cinema, fussing about this sort of spatial fidelity is misguided anyway, because spatial fidelity only needs to meet a minimum threshold that most modern cameras already meet. Once that threshold is surpassed, for the viewer’s perceptual experience true resolution no longer plays the leading role so often attributed to it.

With that threshold surpassed, other attributes are the frontier to improving the viewer’s experience of image quality. Since some of these aspects are not dependent on camera fidelity, they can be sculpted and crafted in post and are not enforced by the camera type.

The selection of camera should concentrate more on better photosites over more of them, and we should focus more on our post production pipeline for image authorship. Post production tools today allow for much more granular control over our images vs strict camera choice, and almost any camera can be matched to another with the due diligence in the grade.

The more photosites you cram on to a sensor, the noisier the image will be. Which means that the ideal number of photosites for the best final image, is NOT the maximum possible number, but the number that strikes the best balance between noise and pixel count. Some of the cameras with the highest pixel count today are also some of the most compressed.

An increase in noise means the higher resolution is useless because you haven’t recorded a clean image (who wants super crisp noise?). As such, we want cleaner Ks, not necessarily more of them. The CHARACTER of noise also has an affect on your perception of clarity/character, as demonstrated in the C500’s smaller/tighter noise pattern to the C300’s. Also, the difference seen between compressed and uncompressed images is often more drastic than the difference between, say, HD and 4K. A Bluray (1080p) will likely look leagues better than a 4K YouTube video, due to compression/bitrate. When you scale an image, all the original pixels are discarded and new pixels are “invented” anyway, so the method by which you scale your image up or down also can have a drastic effect on your image more so than the strict resolution of the original “negative”. 

If you properly scale up from too-small a source, the image just looks blurry, not pixelated. If you watch that same blown-up image on a cell phone, for instance, then you may not even notice because the display is so small. So with all this in mind, we see that the C300 is giving us those aforementioned “cleaner” images SOOC than the C500. We’ll get to how that can change here in a few paragraphs.

Now, this is all well and good but sometimes you’re not the eventual author of your image. Sometimes, you’re hired to deliver a product, raw footage or what have you, and it’s in someone else’s hands to finish. In these instances, oftentimes, the requirements are completely arbitrary. Marketing departments love to be “on trend” and near the cutting edge, so they’ll just throw out a number and demand you hit it, also somehow framing for 16:9, 5:4, 1:1, and 9:16 simultaneously in the process (no that is not an exaggeration, that happens a lot these days). In that case, you just light it the best you can, apply your knowledge, hope the production designer and costume department did a good job (if applicable) and collect your paycheck. If they demand 4K RAW, both the C300 and C500 can do that. If they want 6K, only the C500 does that. You can try to explain to them the nuances of imaging until you’re blue in the face but that’s not your job in these circumstances. You’re a hired gun, and you’re there to shoot and deliver what they asked of you, even if you know it can be done a different way. In the same way, if they ask for 4K 120fps, you’ll only get that with the C300. Then again, sometimes you get lucky and they just say “HD is fine, we just want to cut it and put it up as fast as possible” in which case you can make your own picture profile or design a LUT to hand off with the LOG footage, if they’re prepared for that sort of thing, and that’ll hopefully give you more authorship over your picture.

THE IMAGE

This is the important stuff. This is where the rubber meets the road. At the end of the day, what is the image we’re getting from each of these cameras? How does the new DGO processing affect the image, and what does that vs the larger sensor give us in the grade? How far can we push the image?

Overall, I found both cameras to be more than adequate in any given situation when properly lit, appearing nearly identical. The C500’s image seems to have slightly lower contrast than the C300, and skews more magenta to the C300’s greenish tint, both of which can generally be corrected to look like the other with a few points of “tint” in resolve or on-camera.

When shot massively under or over exposed, both cameras preferred to have more light, recovering rather gracefully in the grade. No surprise there. I did note that, upon recovery, the C500 did seem to respond better and have more image fidelity than the C300 when graded from the extreme ends of things, but only after noise reduction in post. From over-exposure, clipped areas don’t look horrible when brought back but obviously there’s no data there. It’s a decently soft rolloff to clipping from my experience but it’s not “Arri good” (what is?).

After a ton of time in Resolve, I can say confidently that the image coming out of the C500 carries more information, and is a little easier to pull keys off or what have you. There just seems to be more nuance in tones, whereas the C300’s image is slightly more “mushed” together. I don’t know how to better describe it but the C500 seems to have more information/definition to play with. It’s subtle, but side-by-side it’s pretty easy to spot, especially when you start grading. As with the resolution, your audience won’t see one compared to the other so it may not make a difference at the point at which people see your finished product, but if you have a choice obviously one would tend to lean more towards having more information than less to start off with but you may need some additional post production tools to get the most out of it.

I also found that the C500, likely due to the larger sensor, had a slightly better time resolving fine detail. In my test shoots you could see the individual strands of fabric in Dani’s top as opposed to the C300 where it was more like one solid “unit” of fabric. The overall apparent sharpness of the image was pretty even, but I did note that difference in the fine detail.

In terms of the maximum Dynamic Range of either camera, I don’t know. I can’t accurately measure that with what I have in my apartment, but I can say that it’s high enough on both cameras that I’m not sitting here wishing for more. It seems to be “a lot”. You can capture pretty much everything in front of you with ease. I’ve read that the C500mkII tests at 13.1 real stops, just under the Alexa’s 14.

It seems that, in general, the C300 has a lower noise floor and the C500 has slightly more “info”. At least that’s what I’m going with at the time of this writing.

 



SHOOTING AT DIFFERENT ISOs

Another thing I’ve learned from Canon is that ISO 800 isn’t these cameras’ base-ISO, 800 is just the best balance of stops allocated to the highlights and shadows. Technically 160 is the base and as-such, at least with the C500, I recommend you shoot as low of an ISO as possible for the cleanest shadows. At ISO 200 I’m seeing functionally zero noise at all, and normal lighting situations are still absolutely tenable. The C300 has clean shadows at ISO 800, and that’s where that DGO processing really shines.

Counterintuitively, a darker scene is generally better shot at lower ISOs (when possible) and a brighter scene is generally better off shot at higher ISOs (when possible) where the dynamic range present in the scene is relatively low. This makes sense if you think about it though, because that’s what changing the ISO does: it allocates more stops to the lower or higher end of the range. In that case, if everything in your scene is bright, say outdoors in the middle of the day, it’s potentially better to shoot at a higher ISO and throw a bunch of ND on there as there’s likely to be no dark areas in the scene but tons of varying bright areas. If it’s dark… well at a certain point the answer is to add more light, but a properly-lit scene does seem to be better served by the lower ISOs. Granted it depends on the scene/shooting situation, but it’s something to think about. ISO 200 on the C500mkII (or the non-DGO footage from the C300) looks very clean.

Obviously all of this only applies to the XF-AVC formats, as with RAW you’ll be able to choose which ISO you want to go with in post.

NOISE

As-advertised, the C300 has a lower noise floor than the C500. However it’s not as simple as “less noise good, more noise bad” it would seem.

In my tests I found that on the C300 you had a max ISO of about 8-10,000 before the image was no good, maxing out around 20,000 to 25,600 with Noise Reduction in post before the image became entirely unacceptable.

The C500 maxed out around 5-6400 ISO on its own, with the “noise-reduced” limit being 12,800 to 16,000 before it truly fell apart.

Now, that’s all in Clog2. In Clog3 I found the noise level changed DRAMATICALLY in the higher ISOs, giving you a limit of 16-25,000 on the C300 and 32-40,000 on the C500. To further highlight that ISO numbers are largely meaningless between cameras, the images at 25,600 ISO and 40,000 ISO, respectively, looked nearly the same in both acceptable noise (to me) and image quality. To be clear, I’m saying the “look” and exposures seemed similar. At that point, does it matter what “number” each camera is set at? They don’t record light in the same way.

I found that, at higher ISOs, the C300’s noise renders rather unusually. It starts to look like there are little white grains of rice everywhere; impossible to grade out. On the flipside, the C500 does have rather noticeable fixed pattern noise at times, usually manifesting in vertical lines, but the noise overall is more tightly packed and “pleasant” when reduced in post. Given the proper tools, when bringing the image back from a bad recording, the C500 seemed to have more fidelity. In that case, I find the C500 is actually better in low-light, but the C300 has that lower noise floor which helps when you’re shooting a (properly lit) scene with some deep shadows.

I also noticed on the C500 that the noise would go up when raising the ISO, and then would pop down every couple steps. It would also get noisy when changing ISOs and then even out after a half second, as if the camera was readjusting to each new setting. I do have the stock NR on, without frame correlation, so maybe it’s that? From 800 it’d “pop” at 2000, 3200, 6400, and I think further on. I also seem to notice that the vertical noise is more prevalent when previewing the footage in my NLE but not necessarily when rendered out, so don’t count your eggs before you eat the basket or whatever. In any case, I’ve got my camera set to changing ISOs in whole stops now to avoid any artificial gain.

Practically, I can say that you shouldn’t really need to go much higher than, say, 5000 ISO on either of these cameras. Even at ISO 800, the cameras render the scene in the same way my eyes see it. If the scene is darker, the C300 will give you a lower noise floor, but for most practical shooting situations either camera will perform perfectly. The interesting thing here is that, while the C300 has less shadow noise, at a certain point it becomes really ugly whereas with the C500 it’s more recoverable/finer grained. The C500 has magenta noise, which can easily be corrected out with Resolve’s “Chroma Noise” reduction without having to affect the clarity/sharpness too much, whereas the C300’s noise seems to be more “burnt in” when it gets out of hand but is tinted more green. 

 



FIXING NOISE

The easiest thing would likely be to just shoot Clog3, which drops the noise floor dramatically. Theoretically, if it’s a dark scene (or even just a very low-contrast one) you don’t have very much Dynamic Range present, so going from “16 stops” to “14 stops” available in camera doesn’t really matter when there’s only 6 stops difference between the highs and the lows in the scene anyway. It would almost be better to shoot closer to a 709 curve and make it look exactly as you need in-camera, therefore allocating more data to those (lets say) 5-7 stops of range in the scene more effectively than if you were to shoot Clog2.

Since log is trying to shove 15+ stops of information into a given codec/bitrate/file, each “stop” gets less of the data pie. If the difference between the darkest and brightest thing in the scene is 5 stops, wouldn’t you rather those stops be given more of that pie? Testing revealed that I prefer the look of the image when letting the camera do the work in these scenarios, instead of shooting log compulsively, so I’m passing that info along to you. Do with it as you see fit. Clog3 may be a better option than 709, and RAW might be a better option over-all, but it depends on the project, your preferences, and how much time you’ll have in post. Be sure to test on your own to find what works best for you!

As is the case with ISO, this only applies to XF-AVC as with RAW you’ll be able to choose which gamma curve you want in post.

It would also appear, anecdotally, that the C500mkII (and therefore likely the C300mkIII) beats the RED Gemini in the noise department at higher ISOs, which is pretty shocking. I didn’t run those tests nor do I own a Gemini, so I can’t comment any further than what that video shows, but it’s pretty crazy to see how far imaging technology continues to push in only a few short years.


IN POST


I was able to match a well-exposed image from both cameras by adjusting the shadows with the log wheel and tinting the image by single digits. Aside from the aforementioned level of detail in the C500 image, the two pictures are functionally identical. To get the noise response of the C300 I found that the C500 should be shot at ISO 200. For scenes that have controlled lighting, this hasn’t been an issue in regards to the highlights.

My i7-7700K/GTX1070 system with 64GB of RAM handles all the XF-AVC stuff from either camera just fine, and the 4K RAW from the C300 handles a little better than the 6K RAW which does chug even without a grade, but mostly in Premiere as Resolve is using “more” of my system for playback. The XF-AVC footage from both camera grades and plays back at full speed just fine, unless I’ve got a Grain OFX or something like that over the top, then it’s a bit more than half-speed. Editing with a LUT over the footage you should be just fine, but obviously the proxies are there for you if you need them. If you need to rename the files, I recommend Advanced Renamer (Windows only). There’s probably a billion ways to do it, that’s just the program I use.

A little tip: renumber the “CLIPS001” folder before importing your footage into Premiere. For some reason Premiere displays all the clip names as “Reel001” if you leave it that way, but will display their correct file names if you name the folder to “Reel002” or “Reel123” or whatever. Just start at 002 and go up with each new card in a project. Resolve doesn’t seem to have this problem.


DRAWBACKS

– CFexpress
They’re expensive and they get really hot, but they are fast as hell and dump incredibly quickly (given you have the computer/connections that can facilitate that). It’ll be nice when the price drops on those.

– SDI/HDMI/Monitor
Can’t use them at the same time! SDI or HDMI, you don’t get both. You can split an SDI signal just fine, but (for instance) I was recently using the HDMI connection for streaming purposes which meant I couldn’t use the SDI for the wireless monitoring system. The screen on the camera still worked obviously, but apparently that’s just too much processing power for the camera to send signals all “willy-nilly” like that. Bummer.

– Rolling Shutter
It’s there. Moreso on the C500 than the C300, but if you use the S35 crop it’s about the same. I would have loved to see a global shutter on the C500 at least but it’s not the end of the world. The rolling shutter on the C500 at 6K RAW (5952 × 3140, 17: 9) is tested at 15.8 ms.

– WFM (Always LOG?)
Stick to false color. Your Waveform seems to only give you the LOG levels, even if you’re using a LUT. It would be nice to turn that on/off, or toggle it as appropriate (on with LUT, off without it). That being said, it is displaying whatever gamma you’re “shooting” in, so if you’re on a 709 profile it’ll reflect that, meaning if you shoot raw but “set” to 709 you’ll get the appropriate WFM feedback and you can adjust as-needed in post since it’s not “burning in” that 709 look. A clumsy workaround, but it’s there.

– Price
For most people, the C300 is the easy answer here price/performance-wise. And by most people, I mean working filmmakers/cinematographers. The everyday shooter. For those choosing a camera for a feature film or something like that, the C500 becomes more attractive. Overall I stand by my statement that these cameras are likely the best bang-for-buck on the market, all things considered.

– High Speed AF
For some reason the C500 doesn’t have autofocus in S&F mode, but the C300 does. The workaround would be to just shoot at 60fps normal speed, but that’s kind of lame. Also no audio in S&F mode on either camera, so watch out.


COOL TRICKS

– Magnification
Magnification works in such a way that, whatever the focus box is over is where it punches-in. So if you’ve got face-detect on, and it’s got a face, hitting the “Magn.” button will punch in on that face. If it’s just a floating box, it’ll punch-in there, but as you’ve got the touch screen you can easily move that around

– Face Tracking
The face tracking is really good. Handles multiple people, switching between them with the touchscreen or joystick.

– Proxies-as-Camera Negative
If you pull out the CFexpress card, the camera will still record the proxies to the SD card. It may sound silly, but that might save your life someday (or at least make certain gigs easier to wrangle). The 35mb/s XF-AVC proxies are more than capable of working for you, depending on how demanding the project is, far outpacing the quality of my C100mkII which shoots to a similar file size/format. If, for instance, you suddenly need to record for 10 hours but only have one 128GB SD card, you’re still in shape. It’ll be 2K or HD, but hey, maybe that’s all that’s required of you. In this case I’d likely shoot a 709 profile or at least Clog3, since you’re dealing with 4:2:0/8-bit footage, but it’s a cleaner image than my C100mkII so it’s beyond “proxy” quality.

– Live Streaming
Check out my article about how I used the C500mkII to do a high-quality live stream (with recorded backups) using minimal external tools and a lot of the camera’s internal ones to min/max efficiency and quality.

– LUTS
It’s not really a trick, but use the ability to load custom LUTs to your advantage! Design some yourself, get exactly the look you want.

– Reliability
Also not a trick, but I thought it’d be worth mentioning that reliability is a huge factor on set. You can not have your camera go down for any reason (who hasn’t been on a set where their RED went down for a half hour to black shade?). The Canons seem to be bulletproof, and these two are no different. Odd choices here and there, but they’ve never failed out on me once.


WRAP IT UP


Like I said, these cameras are nearly identical. For most people, as the cameras can easily create nearly the same image when exposed well, the monetary savings of the C300 coupled with the no-crop 120fps and low noise floor will seal the deal. Most of us are used to S35 and the associated lenses anyway. For those of you who want that bump in Image Quality, enjoy shooting Full Frame, or want the added option of 6K, the C500 is for you. Both cameras will serve owner/operators well, and as rentals they’re a no-brainer. For all practical purposes you could pick either camera and get any job done and have it come out looking fantastic with minimal work in post. And hey, if both of these are out of your price range, the C70 is coming out and there’s always the C200, which creates amazing images.

Now I’m sure the question that has been begged (and explicitly asked of me a handful of times) is “if the C300mkIII and C500mkII came out at the same time, would you still have bought the C500?”

Well, no. Probably not. It’s $5,000 cheaper and the image is nearly the same. You can get a sick lighting kit for $5,000. Maybe some extra lenses. The Bright Tangerine cage I have for my C500 works on the new C300, that’s like $1,200 right there. Some extra cinema batteries and CFexpress cards would be nice too.

At the same time, I don’t feel compelled to sell the C500 and pick up a C300; I do prefer having the “better” image of the C500 as well as the larger sensor, which expands my shooting options, and if I shoot at ISO 200 there is zero noise in the shadows and it doesn’t seem like I need much more light than normal to work at that sensitivity. Plus, it’s not like my camera is suddenly worse, it’s just that there’s an option right behind it that’s nearly as good. In regards to low light noise, I find a small touch of chroma NR in Resolve evens the playing field without hitting my PC hard enough to be annoying so I don’t find it to be a problem. If you’ve got an expensive camera but no lights, that’s a bigger problem.

With both of these cameras, the answer to any question is usually “Yes, we can do that” and that’s the hallmark of a well-made tool.

A Comparison of Variable ND Filters

Is there a good variable ND filter out there? Today, we’re going to take a look at variable ND filters. We compare Peter McKinnon’s Polar Pro, B&W, Syrp, and Tiffen variable ND Filters. Let’s see if those pricey Peter McKinnon filters are worth it compared to some of the less expensive options.

So I just dropped $450 on Peter McKinnon’s variable ND Filters from Polar Pro. I needed some quick for a job. They were trusted. I love a couple of features. They have stop markings and hard stops and they seem pretty robust. So I dropped all that money on them because I needed a good set. And I want to see if it was worth it compared to some of the cheaper options.

I was on a shoot one day and I needed a variable ND Filters for three cameras. So I bought the Tiffin $175 inexpensive variable ND filters. Let’s compare these. We’re also going to look at the Syrp ND filters and the B&W ND Filters. What do we need to look for? What’s the concern?

Let me just say this, I don’t love variable ND Filters. One of the biggest reasons for that is color shift. Most of them have some sort of shift that’s not good at all. And then you’re also dealing with stuff like vignetting and cross-hatching because the variable ND is just two polarizers that are stuck back to back. So it can do some weird stuff to the light that’s entering the lens. And sometimes it’ll kill the contrast. It’s really interesting how it will affect the contrast. So it’s really important to look at those three things and to see what’s a good variable ND that is going to work for you. I think you’re going to be very shocked at just how little you have to pay. Let’s check it out.

We want you to see the different filters and exactly how they respond to color contrast and cross-hatching and things like that. We’re going to go quickly through these. Most of the filters were within 400 degrees Kelvin off. I think our B&W filters are like 150 degrees off. We white-balanced all these using an eyedropper in Photoshop back to a clean color and then we saw how much they had changed from the original. How much the color setting had changed.

Here we have the no filter image first. There is a tiny bit of vignetting in the corners. We eye dropped close here to the corner of the building. It’s like 5100 degrees Kelvin plus 11 on the camera.

Now we’re going to pull up our first filter here. It is the B&W 1-5 stop variable ND. The image on the left is uncorrected with the filter and then the image on the right we’ve eye dropped that same spot again to adjust it. This is 1.3 stops, which is the minimum on the B&W filter. So it’s pretty accurate. It corrected back really nicely. It’s a beautiful clean image. The blue is wonderful. It didn’t lose the contrast too much. So it actually looks great. But how does it look now if we go to three stops?

At three stops in the uncorrected image, you see the color start to wash out a little and there’s some weird shading going on. You’re starting to see that cross-hatching develop. But once you correct it, it actually doesn’t look bad. It’s a little shadowy and a little bit desaturated. It’s not amazing. It’s okay.

Then at five stops it actually looks better in my opinion at five stops then at three stops. Even though we did have to correct it a bit more. We see a little darkness in the lower right corner and lightness light in the sky. It shifted 450 degrees K and plus 10 magenta. It really looks better at 1.3 stops or up to five stops.

So let’s go to the Polar Pro. This is the one I’m already really invested in. So I was hoping that it would do great. We will let you be the judge. The minimum is 1.3 stops. It looks great corrected. It actually looks almost identical to the original. It shifted 300 degrees Kelvin and plus 10 magenta. But it easily corrected and looks good.

But then after that it turns kind of ugly. It gets muddy looking in the middle, and this is at three stops here.

At five stops it seems like the colors just got desaturated. So even though we correct it with the white balance, but the color is gone. It was that really. So the Polar Pro at 5 stops shifted minus 300 K plus 9 magenta. But that’s not the issue. The issue is that it just became very desaturated and muddy looking.

The 6 stop mark on this filter is actually delivering 5.6 stops of filtration and then the nine stop mark is delivering 7.3 stops with filtration. So it doesn’t actually filter out as much light as it advertises, which is kind of a disappointment. The color shifting at the minimum setting is still kind of bad. The 6 stop mark is not as bad as the last one was. It’s minus 400 degrees K, 6 magenta. It’s a little desaturated. It’s not quite as bad but feels kind of muddy again to me.

But then at the 9-stop mark or 7.3 stops it clears up. And actually, it looks pretty good. So if you’re using this combo of lenses, it works really well at the minimum, like the one-stop and the seven-stop end.

But in between really messes around with the color. The disappointment here for me is it’s not filtering nine stops of light.

Okay, so the Syrp 1-9 filter. This is the original image. This has way more vignetting. It kind of collapses on the side and gets dark on the sides. The color is actually pretty consistent. We lose about 450 K plus 2 magenta. Not much magenta shift but about 450 K. On its max it is 150 K.

It doesn’t desaturate. It just gets these dark areas on the right and left. You can see where the cross-hatching is starting. Maybe with some subject matters this wouldn’t show but if you have any kind of a high key area in your image like the blue sky like we have here, it is super noticeable, especially for video. If you are moving around, it’s going to be really noticeable. So that Syrp 1-9 is really more like a 1-6. With vignetting all the way through.

Let’s look at the newer Syrp 5-10. They call this the super dark filter. The range is about 5.3 to 9.6. Again, the color looks pretty good, but the vignetting is super heavy. It’s okay at the 5 stop end but it gets really bad at the 9. It looks almost as if you’ve isolated or masked out the building and you made the entire sky black. It is shifting from 300 K to 250 K plus 10 magenta. But the cross-hatching is a problem. I will say the color is pretty good all the way through. It doesn’t give you that muddy look like we got with the Polar Pro.

All right, here’s the Tiffin 1-9. It actually is not bad for being the cheapest option. I think at the 1 stop it’s really about 1.6 stops. At that end, it looks super clean with hardly any effect to the image.

Then we go to 3 stops and it is pretty good, still really clean.

We shot this to 7.3. I think this is probably as far as I felt comfortable pushing it. I felt like this is the max. It doesn’t go to nine because nine doesn’t work.

I would back off of this. I would probably just keep it at 6 stops.

It has the cleanest color this shift. It shifted minus 350 or 250 K. So it didn’t shift that much. But you just start getting that vignetting so early.

Let’s look at all these filters at their minimum setting at 1-2 stops. At this end, they all do pretty well, except the Syrp is already vignetting. But it’s hard to say a clear winner here. And now we have them all at three stops. The Syrp has the vignetting and the Polar Pro and the B&W are both a little desaturated. I’d pick the Tiffin at this point.

This is at 3 stops.

So here’s 6 or 7 stops on all the filters that can go that far. Polar Pro looks pretty good at the dark end. This is 7.3 stops Polar Pro’s colors cleaned up and doesn’t have that muddy look. I’d actually say it’s the best of the bunch at 7 stop end. The Syrp is still vignetting. The Syrp 5-10 vignetting is even worse, which is surprising because you think the super dark filter would be the one that was better at being super dark. The Tiffin does again have vignetting. It’s really more of a 1-5 filter but I thought I’d pull it up again just to see how it compares.

We tested a B&W straight filter and not a variable ND. This is a 3 stop and a 6 stop. So you can stack these to give yourself a 9 stop ND filter. It is startling how much cleaner these are color-wise and you don’t get any of the vignetting. The difference between the corrected and uncorrected images is almost negligible. It’s clean all the way across.

Those B&W filters are super high quality.

We threw in these Amazon filters. I had some filters that my son-in-law wanted. I told my wife they were terrible filters. I think they were around $50 for the set of 3. I’d be shocked if they were any good at all. They are not as good as the B&W. You get a little more of this shading going on and the color was a little less accurate, but it still beats out the variable ND filters.

So here’s my takeaway on this, if you’re a still photographer, and you need ND filters to be able to give yourself blurring water in the daytime or doing time-lapse where you want to have the people moving using a long shutter, buy a straight-up 3 stop and 6 stop set like the B&W. I think that B&W 3 and 6 will give you up to nine stops of ND. It’s going to give you really clean color all the way through but allow you to get that blurry water.

Now if you’re a video shooter and screwing and unscrewing filters isn’t practical in the run and gun situation, you want the variable ND because it’s fast. You have a range from 1-5 stops with just a twist of a thing, it doesn’t have to come off the camera. I think if I had to make a choice, it’d be really tough.

I’m leaning towards the Tiffin just because it was so consistent, but it only gives you up to 5, maybe 6 stops of filtration and often that’s not enough. If I want to shoot at a 2.8 outside in the daylight you really need 8 stops or 10 stops of ND.

I think if I had to choose a runner up, I might go with a B&W quality glass. It was 1-5 stops. There was a little bit of washed-out color. I think the worst offenders were the Polar Pro because the color just went crazy in the middle. It was good on the ends, but in the middle range it is terrible.

But there was no vignetting so it has that going for it. And the Syrp just had crazy vignetting yet the color was okay. This whole thing was a huge surprise. The cheapest option was maybe the best one which never happens in photography.

So kudos to Tiffen, who has been making filters for a long time and they have an excellent product.

So there you have it, a look at variable ND filters and a look at some standard ND filters. Hope this helps you make a decision as a photographer or as a videographer on what’s the best one for you to buy and to use on your camera.


About the author: Jay P. Morgan is a commercial photographer with over two decades of experience in the industry. He teaches photography through his company, The Slanted Lens, which runs a popular YouTube channel. This article was also published here.

Canon R5 and R6: Comparing the File Formats (RAW, CRAW, JPEG, and HEIF)

In my last article, I shot images with both the Canon R5 and Canon R6 to compare the ISO performance of both cameras. As always, this lead to a lot of comments, emails, and DMs asking me if I could also compare the different file formats of these cameras. While I still have both these cameras on loan from Canon, I decided that now would be a good time to tests these parameters for all of you (and me too).

Editor’s note: This article was originally published on photographer Jeff Cable’s blog and has been syndicated with permission.

I have always preferred to shoot in RAW mode only. You can see my latest video explaining why this is the case.

But these new mirrorless cameras offer new formats, such as the new CRAW and HEIF formats. Would either of these be tempting to me? Actually, after doing this test – maybe yes!

Just like my test of the cameras ISO performance, I created labels to put into each image so that there was no mistake on which format I was looking at. Each camera was set to Manual mode, ISO 160, f/8 at 1.6 seconds and using the new Canon RF 24-70mm f/2.8 L IS lens.

With each camera, I took one image in each format (RAW, CRAW, JPEG and HEIF) to compare image quality and file size. Here is the data:

Canon R5

RAW: 47.5MB
CRAW: 22.3MB
JPEG: 12.4MB
HEIF: 13.1MB

Canon R6

RAW: 22.5MB
CRAW: 11.1MB
JPEG: 6.5MB
HEIF: 7.3MB

If you want to see these images in detail, you can download them from this Dropbox folder.

Before I commenced the testing, I already knew where I would focus my attention. I know that the new HEIF format is a newer and better file format as compared to JPEG, but that both those formats pale in comparison to RAW and CRAW. Why? Because both of the RAW formats include a lot more data and make it easier for me to make adjustments to the final images (protecting the details in the shadows and highlights).

So… did I open the JPEG files to compare them to the RAW? Yes, and as I expected, the quality was subpar. Did I open the HEIF files? No, because at this time, Adobe Lightroom and Photoshop (ACR) will not open these files. But once again, I already know that I will shoot one of the two RAW formats to retain the highest quality possible.

But here is where things get interesting!

As I mentioned, for the past 10 years I have opted to shoot in full RAW format for most of my photography, only choosing to shoot in the older MRAW (medium RAW) format when photographing receptions and parties, in situations where clients are not likely to be ordering very large prints. I typically opted to shoot full RAW files because I wanted the best quality images for myself and my clients.

But when comparing the RAW files of the Canon R5 and Canon R6 to the new CRAW (Compact RAW) files [which use a lossy compression algorithm], I could not see a difference in the image quality. even when zooming at 500%! All I saw was equivalent image quality but half the file size to download and store. And keep in mind that both the RAW files and CRAW files are the same resolution (8192×5464 on the R5 and 5472×3648 on the R6).

I even called a friend at Canon to ask if I was missing something here, and he confirmed that it is very difficult to see the difference between a RAW and a CRAW image. So I asked him “Can you give me one reason why I would not default to shooting CRAW only instead of RAW?” and he could not.

Canon R5 CRAW file at 500%
Canon R5 RAW file at 500%

So my conclusion is this: I think I may be defaulting to shooting CRAW from now on. It feels strange to write this since I have always been a proponent of shooting full RAW files. But I just can’t see any reason not to default to CRAW moving forward.


About the author: Jeff Cable is a photographer based in the San Francisco Bay Area. You can find more of his work on his website, Facebook, and Instagram. This article was originally published on Cable’s blog.

Can Pixel Shift Compete with a Bigger, Higher Resolution Sensor?

In a pseudo Part III to their sensor size comparison series, Jay P Morgan and Kenneth Merrill over at The Slanted Lens decided to answer a question that popped up several times in the comments of a previous video: can the pixel shifting on a Micro Four Thirds or full-frame camera match the output of a 100MP medium format sensor?

The three cameras being tested here are the Olympus OM-D E-M1 Mark III, the Sony a7R IV, and the Fujifilm GFX 100, and the point was simple: limitations of pixel shift photography aside, can the output from the Olympus and Sony’s respective ‘pixel shift’ modes out-perform the standard, single-shot performance of the 100MP medium format GFX 100?

The side-by-side test involved capturing photos of The Broad art museum, since its architectural style and location was ideally suited for highlighting issues like aliasing and any problems with dynamic range.

First, they compared the full-frame Sony to the medium format GFX. When using the pixel shift mode with the Sony, you have two options: a 4-image stack that spits out a sharper version at the same 61MP resolution as a single shot; or a 16-image stack that produces a whopping 240MP file.

You can see side-by-side comparisons of the output from the Fuji and the Sony (in both modes) below:

Next, they compared the Olympus output to both the Sony and the Fuji GFX.

Interestingly, the quality of the Olympus file came out much sharper if you process the files in Olympus’ own Workspace software, as compared to using Photoshop. You can see a side-by-side below and there’s really no comparison: if you’re using the pixel shift mode on your Olympus camera, definitely use Workspace to stack the images.

The rest of the comparisons were done using the Workspace version (obviously) in order to keep quality as high as possible.

Here’s the Olympus 4-image stack against the Sony with no image shift, and then the against the Sony’s 4-image stack mode:

Finally, you can see the Olympus pixel shift image followed by the GFX all by itself below. For some reason, they didn’t show these side-by-side in the video:

An interesting aside: Jay makes a good point at the very beginning of the video before jumping in. It’s worth reprinting in full just to keep these kinds of technical comparisons in perspective:

The reality of photography is: if it’s creative, and it communicates, and people will pay you for it, that’s commercial art. It doesn’t matter if you shot it with a shoebox, with a piece of tin, with a hole pocked in the front of it. It doesn’t matter.

So these kinds of conversations are interesting. But the reality is that they are for tech heads.

Of course, there’s nothing wrong with being a “tech head” and enjoying the occasional pixel-peeping comparison, but it’s worth keeping Jay’s words in mind since sensor size comparisons tend to get people’s blood pressure into the danger zone.

Anyway, to see this intriguing comparison for yourself and compare more side-by-side images of pixel shift vs medium format vs non pixel shift, check out the full video up top or head over to The Slanted Lens website to read their full write-up.


Image credits: All photos by The Slanted Lens and used with permission.