May 2, 2012

The Canon 5D Mark III, Canon C300, and RED SCARLET Shoot a Nighttime Bike Race Side-by-Side

Remember when video cameras had sensors that were one third of an inch? It was damn hard to shoot anything at night. But now there are a million cameras (well, not a million, but a bunch) with Super35-size sensors. So while you can actually get a decent exposure while shooting a bike race at night now, not all sensors are created equal. Here's the Canon 5D Mark III, Canon C300, and RED SCARLET filming the same bike race. The first pass of the video has no key (so you don't know which is which), and then they show you the second time around:

The biggest difference is just how much lighter the C300 looks than the SCARLET -- as Robin recently noted, the M-X chips in current-generation RED cameras just don't reach as well into the dark as the offerings from Canon and Sony (the M-X chip is a few years old, after all). To me, the RED is more ideal for features and the C300 better for docs, but in either situation, if you're going to be doing nighttime shooting without the ability to add light, the C300 is a far better choice. We'll see how RED's Dragon chip fairs when it finally comes out... but keep in mind, from a purely mathematical standpoint, the RED's smaller pixels (because of 6K resolution) will theoretically never be a "best case scenario" for low-light shooting. Sensor technology has advanced greatly in the last couple of years, but one does wonder if anyone in the filmmaking world is really asking for an extra K (going from 5 to 6) as opposed to more dynamic range and better low-light performance. The fashion industry may want the extra resolution... but at what price to filmmakers?

Note: I'm not entirely clear on whether RED is just adding extra resolution around the edges (to a correspondingly larger sensor), which would negate this worry. Anyone?

This video was shot by Hello World Communications, a great NYC rental house for video equipment. In fact, to date my RED SCARLET has lived there -- if you've rented a PL-mount SCARLET (they have their own EF mount versions), you probably rented mine! Sub-renting can be a good option to generate a return on your investment, and since I have been knee-deep in Man-child script revisions (and therefore not actively shooting anything), my camera hasn't lived at home in quite some time. Cameras are not toys, and if you are not using yours, someone else should!

Any thoughts from watching the above video? How'd you think the 5D Mark III faired?

Your Comment

58 Comments

I think the 5d kept up pretty well. Nice to see ungraded test footage. All of the cameras performed superbly.

May 2, 2012

0
Reply
G-MAN

Hell yeah Red Hook Critirium!! Dan Chabanov steals it for the second year!

Ha ha, I love when my two passions overlap!

May 2, 2012

-1
Reply
innokenty

Same here man!!

May 3, 2012

-1
Reply
Travis

From the first edit, without the answer key given, I pretty much deduced the cameras. The 5dmkIII looks like a video camera compared to cinema cameras. It could be because they have that blasted light pole for most of the shots, the highlights are a bit harsh.

The RED is darkest (obviously) it also has more magenta in the tint.

To me, shot 2:47 looked the most 'filmic' and that was the C300. It is also a good composed shot.

Both the RED and C300 are smoother and have more dynamic range. The sky had compression artifacts on the 5dmkIII.

May 2, 2012

0
Reply

Canon 5D MkIII - unusably soft, Scarlet unusably dark and noisy, C300 the clear winner and only choice for this type of production. So bummed the MkIII is so soft - like cutting 16mm into 35mm show.

May 2, 2012

0
Reply
Lance Bachelder

Is it just me, or does most of the Scarlet footage look like it was overcranked?

May 2, 2012

-2
Reply
Andrew Swan

By which I mean the shots of the race in progress, not the post-race/interview footage.

May 2, 2012

-1
Reply
Andrew Swan

I was thinking the same thing. Like 48fps. Either that or the cyclists were going way slower in the portion of the race.

May 3, 2012

0
Reply
Drew

First, sad to say there is no info on how each camera was set in terms of look file, or in terms of the RED, the gamma curve applied. There black performance is up in the air for me at that point.

I'll agree the mkIII appears a little soft but I wonder what the C300 was set to as well in terms of sharpening. I haven't gotten my hands on mkIII footage yet (my friend does own one) but I bet it can be "tightened up" in post with sharpening applied between gamma corrections (correct gamma to linear, apply sharpening, and convert back to non-linear).

Ambiguous camera tests like these only prove one thing: we have better digital video cameras now then we did 5 or 10 years ago.

May 3, 2012

0
Reply
Angelo

http://prolost.com/blog/2012/4/13/canon-cinema-eos-1d-c.html is worth a read considering the mkIII's "lack of sharpness". I tend to trust the guy who use to work at ILM.

"Everyone says the 5D Mark III’s video is even softer than the Mark II’s. Maybe just a little in-camera sharpening?
---
No. The Mark III’s softness is simply the lack of artificial sharpness that came from the aliasing that plagued the 5D Mark II. This means that the footage takes sharpening in post even better than 5D Mark II footage, because there are fewer inherent artifacts to bring out.

It would be nice if the 5D Mark III resolved more detail than it does (there is plenty of room for improvement there), but adding in-camera sharpening won’t make that dream a reality. It only adds permanent, ugly artifacts to your image."

May 3, 2012

1
Reply
Angelo

May 3, 2012

1
Reply
Angelo

I think the scarlet had the best shots, but mostly because the Canons just didn't seem to ever be white balanced as well, and they never had the favorably lit shots. The comparison is too unscientific and poorly documented to take seriously, but thanks for sharing.

May 3, 2012

0
Reply
Peter

Personally, I think this is an unfair comparison, because you don't show the Scarlet at it's native 4k resolution!

(just kidding)

May 3, 2012

0
Reply

I really laughed when i read this! (as long as it refurs to the great argue at reduser.com)

May 3, 2012

0
Reply
Jendrik

It sure does... That about sums up Jannard's argument...

May 3, 2012

0
Reply

hahha, you just made my day

May 3, 2012

-1
Reply
Galo

As a student of NFS, what does "key" mean in the context in which you've chosen to use it?

May 3, 2012

1
Reply
MRGABE

The lower third (C300, 5DMIII, RED Scarlet).

May 3, 2012

1
Reply
Chris

This is pretty unfair based on lack of info given. granted i personally believe the c300 fairs better in natural low light, the scarlet provides no in camera noise reduction or sharpening like the other two. I just hope they applied at least a little of both to the scarlet, otherwise this test is useless.

May 3, 2012

0
Reply
domenic barbero

o0o0o0o a Vimeo video. Let's judge a vimeo video.

May 3, 2012

0
Reply
Chet

On the one hand, fair point. On the other hand, what, is the Scarlet going to somehow magically be lighter in the original footage?

May 3, 2012

0
Reply

I've heard the 5D MK2 isn't really 1080P but, rather, something up-res'd from a smaller recorded size like 1280 or 1440.
Is the MK3 the same? Not true 1080P?

May 3, 2012

0
Reply
dixter

More like the 5dmkII downrezed by lineskipping (getting from 21MP to 2MP), therefore, with all the extra lines of resolution skipped, even though the frame size is 1080 lines high, those lines are partially made from aliasing. Thus, when filming a resolution chart examining resolving power, the 5dmkII is under 700 lines.

The 5dmkIII does not lineskip per se (according to Canon) but the image is pretty soft from a low pass filter in front. Do not do in camera sharpening, but sharpen in post and it will respond good.

However, resolving power still seems to be around 700 lines or less and that is to be expected from a 22 MP sensor designed primarily for still resolution, thus having a low pass filter 'tuned' for that (though the guy who took his off found both video and still benefited).. Neither are 'uprezing' images though.

May 3, 2012

0
Reply

The C300 is definitely the winner in this video. The images coming from this camera is much cleaner and has a more film look. We can spot 5DMkIII among the two others because of the compression. the RED is visibly darker than the others. I don't know if it's the C300 or is a bit too green or the RED that is too magenta..? But yeah as you all said, without knowing which settings they use for each camera, the comparison is unfair.

May 3, 2012

-1
Reply

I disagree. I don't think that the C300's image was any cleaner than the Red's. In the dark shots especially the one of the biker being interviewed, the RED's image looks much cleaner and less noisy than each of the other cameras. That's to be expected because it's most likely being exposed at a much lower ISO.

What the C300 does win in this video for is it's great ability and range to work and expose with available light. Obviously it goes deeper into the shadows than most any camera on the market, but that's not really something we haven't seen or known based on data and information posted here in the past.

That being said, I don't think any DP would try and push the RED into a situation like this without using he appropriate lights. In these situations, low available streetlight, the C300 is a beast and does a great job. I don't think it wins if the script was flipped and we were in a well lit, more film like situation.

Nice to see the footage none the less.

May 3, 2012

-1
Reply

"That’s to be expected because it’s most likely being exposed at a much lower ISO. "

I think Scarlet's been exposed at a higher ISO and that is why it's underexposed. There actually is no ISO on Scarlet, it's just metadata, what hits the sensor is being recorded, so only shutter speed, aperture and NDs or filters will control exposure. Exposing for a higher ISO (you still set ISO as a starting point for post production) in low light will often result in underexposed images, because you see a brighter image on your monitor than what the sensor actually records. Exposing for low light isn't that easy with Scarlet and hasn't been done too well here.

May 3, 2012

0
Reply
mike35

I was referring to ISO settings on the RED relative to the insanely high numbers that the C300 is able to reach and create pleasing images. I've read and seen test footage up to 20,000 ISO that shows a nice image without distracting noise.

Thank you for the information regarding how the RED exposes it's image. As strong as the metadata features are with REDCODE you still have to be able to shoot for the conditions that best suit the camera. Extreme low light like this, just how you said, isn't easy to capture on that camera.

That being said I can't wait to get my hands on one and test it's ability to deliver quality images.

May 3, 2012

-1
Reply

Man, I've been watching a ton of Redbox movies, and many DP's are exposing the RED totally wrong. Noise is all over the place in DVD let alone blu-ray quality. just finished "The Divide" last night, shot on the Genesis. In one scene, character A is exposed perfectly, character B sitting across in some shadows has noise, I mean sparkly, shiny, dancing noise EVERYWHERE. I'm thinking, the footage was pushed 4 stops and looks horrible compared to the noise of Canon DSLR's, the the noise was sharper and smaller. Like salt grains everywhere.

"Mercenaries" had the same problem with two characters. This one shot on the Red One I could immediatly tell the horrible grid banding in the Red.

Both of these are more than visible to 'the average' person watching. Shane H. and his production team that did Act of Valor is SO much better than Mercenaries. Shows what a 5dmkII can do in the hands of a good DP and what a better story can do than a poorly lit 4k Red One.

Acting on both is bad, but Act of Valor had much better bad guy actors.

My point in all of this: DP's are not lighting the RED and other semi fixed iso cameras properly on decent budget films.

May 3, 2012

-1
Reply

bad guys with beards!

May 3, 2012

1
Reply

Yeah you have to light to your equipment. Obviously the C300 is a monster in low, available light and performs really well there. Other cameras, like the Scarlett don't do that. If I were a Scarlett owner, or working on a shoot with the Scarlett, I wouldn't expect to put it in this environment and get strong results. I'd be making sure to light the scene appropriately and go from there.

May 3, 2012

0
Reply

c300 wins this one and the mark 3 held up really well with its older brother .. but the red was seriously dim the entire time ... smototh yes , you could see the dynamic range it had but it really didn't look that great specifically compared to the c300

May 3, 2012

0
Reply
Ali

I have to say - this speaks to every pointless test I've seen between DSLR, C300 & Scarlet.
To compare, the cameras LITERALLY need to be put side-by-side. Exact same location so they all have the same lighting environment. REDRAW is a digital negative, so really an investigation would be needed into the postproduction workflow, grading, sharpening / noise reduction etc etc.
5D III looks bad, C300 is better in low light (but then, that is known), but looks pretty washed out, Scarlet doesnt seemed to have been put anywhere near the sort of light sources of the other two cameras.

May 3, 2012

1
Reply
GWhistle

We have Zacuto for their AWESOME scientific tests - and the Red camp is not happy. Then we have "real world" tests and the Red camp is not happy. Then I watch movies shot on Red, and I am not happy :)

Seriously, as mentioned in my other post above, I've been watching a ton of decently budgeted Lionsgate movies and DP's are not using the Red correctly or maybe it is underperformed. Noise everywhere and not just noise, but GRID-like banding.

Watching Contagion close up on the big screen, I thought at first for sure it was a 5dmkII. Sure it was sharper, but the highlights blew out, especially in the bar with Gweneth P. Skin tones look similar. All I knew is, this is digital not film. I go back to IMDB, it shows Red.

May 3, 2012

1
Reply

In addition to the processing questions already raised, what were the exposure settings on these cameras? I'm no fan of the RED, but it looks a good stop-and-a-half darker. I find it hard to believe that's just a sensor issue.

May 3, 2012

1
Reply
Alex

Shooting low light/low contrast on the Scarlet, ISO needs to be LOWERED, not raised. It records RAW, so it's only metadata, but if you expose for higher ISO in such environment, you will underexpose the sensor if you just go for the image on your LCD. That's what happened here, it's underexposed and that results in more noise. On the C300 or 5D, what you see is what you get, making it much easier to expose properly in low light.
Plus you'll have to blackshade the Scarlet before shooting low light, which reduces noise. I don't think this test proves anything considering Scarlet as most shots have been cropped slomo also, which increases noise size further opposed to real time footage of the other cams.
Plus, as mentioned somewhere above, RAW footage never has any noise reduction applied, opposed to the compressed Canon footage, it has to be done in post, I don't know if the guys did.

C300 should still perform a bit better in low light but a properly exposed Scarlet with proper post handling should come much closer than in this video.

May 3, 2012

0
Reply
mike35

Don´t know why everybody is mentioning the Scarlett being darker while it was clearly recording slow motion at a higher frame rate.
Given a shutter of 180° and (presumebly) the same aperture on all the cameras lenses, it´s just logical that the Scarlett has a darker image because of the shorter exposure time you get with higher frame rates while staying at the mentioned 180° shutter. I guess.

May 3, 2012

-1
Reply
Jendrik

It was not overcranked for all of the shots, especially the interview at the end. Scarlet: dim, noisy but sharp.

May 3, 2012

0
Reply

I belief that for the cost and what you get in low light no camera comes close to the 5DMIII. amazing this video, thanks a lot for sharing it.

I love the C300 but the price (my personal opinion) is way too much for the camera. I want to see what the Sony FS700 can do in low light situations like this one because at $8,000 Dollars and the option to upgrade latter to 4K that will be a killer camera.

For now we keep our 5DMII and 7D and the 60D, once we can we will get a 5DMIII as A-Camera and the 5dMII as a second one.

May 3, 2012

0
Reply

I'll try to be nice but how dumb is a night time "test" without the ISO information!!!! Correct, worthless ...

May 3, 2012

1
Reply
Mako

Is it just me, or is most of the Scarlet footage seem "squished" ... it seems to be taller than it is wide, like back in the non-square pixel days. That combined with the fact that it seemed to have been under-exposed (perhaps because of the aforementioned slow-mo) and the blacks pushed way up kinda killed that footage for me. I've seen the Scarlet in other usage and it didn't have these issues.

Also, I hear many complaining about the Canons being "too soft" and yet the standard seems to be to turn in-camera sharpening completely off. I can understand some of the benefit of turning off the sharpening in-camera, but are they sharpening it in post or just leaving it unsharpened? Personally, I leave the sharpening on 3 and still do some sharpening in post...then again, I have T3i.

--Ray

May 3, 2012

-1
Reply

there is a new outstanding camera coming soon named green cinema.it will be a way better than these cameras!

May 3, 2012

0
Reply
josh peterson

but the real question is what output from the RED did they use? Did they output REDRAW and then work the footage to bring the shadow detail out? The RAW output is a strong feature for RED....really.

It's like comparing the quality of high end DSLRs...using only in-camera JPEG generation.

I'm just saying, having RAW as an option give the user far more latitude to play with than the others.

I liked the C300 footage best in this demo.

May 3, 2012

-1
Reply
seanmcr6

Nice, The C300 is the clear winner here....."Cameras are not toys, and if you are not using yours, someone else should" with that said I just sold my HVX200 to someone who need a hd cam..

May 3, 2012

0
Reply

I don't really understand these sort of comparison tests. When you test cameras to learn how best to utilize them, there needs to be references in the footage not random footage. Comparing regular footage doesn't matter half as much as knowing where different cameras breaks in the highlights and where the noise floor is in the darks, how many stops can you under or overexpose. I personally would benefit more from a write up telling me the average foot candles those spot lights exposed in the road, and at what iso's where each camera exposed.

For those who might be interested in actually comparisons between c300 and scarlet x, you can overexpose middle grey by around 4 stops with a scarlet before it clips at 800iso, on the c300 is right around the same, as far as the darks are concerned you can underexpose 3 stops with the scarlet before noise is to prevelent and around 4 stops with the c300(noise tolerance is really a personal preference so don't take these numbers as facts). And for those interested in the 10k price market, the f3 performs better than both cameras by at least a stop in both directions. It has the best dynamic range distribution by a considerable amount, the c300 has the best low light, and the scarlet has the most punch and vibrancy. Best all around camera in the price range=sony f3, Best Doc camera=c300, best commercial/narrative filmmaking camera=scarlet, at least my some what substantiated opinion

May 3, 2012

2
Reply
Ryan

I'll change the topic a bit :). Why on earth they used track bycilcles on road? They are not faster than road bikes, and don't have breaks and are limitted to one gear :).

May 3, 2012

1
Reply
Antonios

why are the over-cranked scarlet shots being compared to normal speed c300 and 5D mkIII footage? Over cranking requires a higher shutter speed which in turn would allow less light to be captured. Also the 3 cameras had obviously different white balance selections which also makes a difference to the captured image and the amount of noise. This test is irrelevant.

May 3, 2012

1
Reply
mike

I was asking the same questions about the white balance. At the very least they could have set it manually.

July 5, 2012

-1
Reply
Joe Smokie

The 5D MkIII looks horribly blurry - that can't be right?

The differences in white balance are normal for different types of cameras. It is hard to exactly white balance even the same camera models when they're not studio cameras which can be manually adjusted with a vectorscope. Balancing totally different cameras from different manufacturers to exactly the same look is kind of impossible - and I guess these shots weren't graded much as not to make this test totally irrelevant

May 3, 2012

-1
Reply
Heiko

I don't agree that it is hard to white balance different camera. There is manual white balancing available on each of these cameras. Now color correcting for different cameras may be difficult but that is a different story.

July 5, 2012

0
Reply
Joe Smokie

mark III is soft, but what lens, and what f-stop was used for each??? That would give me much more info then just the camera body. If mark III had ANY zoom lens on, and C300 & Scarlet had ANY movie lens, then it suks for the little guy. any DSLR with FF will be soft without top PRIME lens ( Leica summilux, Zeiss CP or top Nikon glass ). Do we know the answer to this?

May 3, 2012

1
Reply

This doc is not fairly done. It does;t say anything about M3 which is price wise affordable most of the small film makers. Palisady's comment and his questions were very appropriate. Hope the team who made this not that effective comparison video could shed more light about it.

May 3, 2012

0
Reply
nyuntshwe

If you've already identified that the colour balance isnt the same, and that you could fix that. Then take the test for what it is. I thought it was a good example, how many ppl chose the cuts the first time round?

Good blind test.

May 3, 2012

0
Reply
Ian Murphy

http://www.facebook.com/helloworldny

Friends:
Thanks for all the responses to our video. We started out wanting to test the low-light capability of the Canon EOS C300 and the 5D MarkIII. Since we also own a Scarlet we decided to add it to the mix. Low-light sensitivity is obviously not a strength of the Scarlet, and yes it was shot over cranked. However, we concluded that with some tweaking all cameras could be cut together.
We would love to do a more comprehensive test and welcome your suggestions. Please submit your wish list to the MESSAGE box on our Facebook page.
We will get started on the comparative test ASAP!

Sincerely,

Ron Yoshida and the Hello World Communications Team

May 4, 2012

0
Reply

Very simply, give us enough information to reproduce the results entirely if we were there with your equipment at that time. Anything that doesn't affect the results you can leave out (e.g. what brand of CF card etc.). Anything that did affect the results you should make clear enough to do again (e.g. exact picture style/gamma settings). This would mean not hand-tweaking things in post of course, because that is testing your colorist not your camera, but automated, reproducible actions in post can be done if listed to the point of reproducibility.

And try to use the same lens on each and ideally have all the cameras in line taking the same shot at once (meaning you're not pulling focus or zooming or, unless you have a triple-header tripod, panning). I think custom white balance should be used via gray card in the field. And use the best settings for each camera that you can, rather than trying to match them camera-to-camera (e.g. expose properly, don't just set everything to "ISO 800" and I put that in scare quotes for a good reason).

This is very hard to do right, it's work, but it only has real value if done right.

May 4, 2012

0
Reply
Peter

I don't see anywhere in this article that it's stated that this was a specific "low light test". Everyone calm down and turn down the gear head nerd factor by about 10-20 notches.

With that said, a true test would be great but I still am a little partial to the C300. I liked more of the shots from that one than any others.

Cool vid by the way.

May 4, 2012

0
Reply

A true test of what??? Each camera has its strengths. I don´t loose any jobs by using a 5D or C300 or Scarlet or even Alexa!!! I choose the camera I want to use according to the needs. And not because someone wants me to use this or that camera. I specially use cameras that MAKE money...I leave off my work and my cameras. So why trying to show which camera is the camera?? By the way I loved this test because ot its simplicity...this is a real life situation. Without seeing this footage and if I was given these shootings conditions I probably would choose the 5D. But this is my personal opinion...Meanwhile I shoot with what I have. My camera judgement never had any complaints from my clients... ;-)

May 6, 2012

0
Reply
Augusto Alves d...

Agreed! It's not a "test"...the situation of shooting like that--a live (non-repeatable) event doesnt really lend itself to the the kind of control required to make it a "test". It's good for a quick impression.

I wish the F3 could've been included b/c it works great in low light, but I haven't seen it next to C-300 footage yet.

May 7, 2012

0
Reply
Daniel Mimura

I think the 5D Mark 111 look's great for the price. It is hard to pay the amount of the C-300. after seeing how they both look in the dark.

December 19, 2012

0
Reply
charles