F3 vs 5D MkII

As part of our Red Centre Podcast coverage we want to share this clip with you that compares the Canon 5D Mk II and the new Sony F3. In this test, both cameras were attached to the same tripod locked together – and then panned over the same technical background. The results show the F3 recording a considerably better rolling shutter and the benefit of the F3 not line skipping. The 5D – while having a very high resolution sensor – reduces this image to 1920×1080 in a very abrupt way. So while the larger sensor gives the cinematographer a shallower depth of field, – the poorer technical down res-ing produces some really unwelcome artifacts.

Of course, the 5D does manage to produce good shots every day around the world. Our intend here is is to highlight the technical issues and allow the cinematographer to shoot knowing the issues and the risks. Click here for a 100meg HD mov sample

For more on the F3 listen to fxguide’s Red Centre podcast #076. There is also considerable more testing in this week’s fxphd BKD course, at www.fxphd.com. Special thanks for the footage to Jason Wingrove (co-host of Red Centre and well known International TVC Director)

30 thoughts on “F3 vs 5D MkII”

    1. cameras in the same price range…would have the same results or thereabouts.
      these tests reveal one of the many reasons WHY the F3 is more than money than a DSLR.
      jas

      1. what i mean is these tests reveal one of the many reasons WHY the F3 is more money than a DSLR. Most DSLR’s have plenty of rolling shutter regardless of their processor, manufacturer or price tag. theyre all pretty bad. When a DSLR comes out that improves on the norm we’ll try to test or mention it
        cheers
        jas

    2. Not so weird considering the fact some guys (like those on EOSHD) claim their DSLR package is just like F3, but way less expensive.

      1. As mike says, we love our 5D’s, this is not meant to show one camera is another cameras ‘killer’ etc. were not into that. its horses for courses.. just offering up as much info as we can get our hands on so others can choose their horse for their particular course

  1. It’s worth comparing because the DSLR that can handle video have captured a lot of attention and are being used widely on high profile projects, often by people for whom cost is not the deciding factor.

    When you see Shane Hurlbut attaching a $20k lens to a Canon 5D for example, he’s all about the image sensor (especially it’s size), not the cost of the camera.

    But it shouldn’t stop here, the next test to see is vs. something from Red?

    1. Steve i ran out of time.. pity as a R1 was right next to the F3. But to honest i can tell you the red will be about the same if not slightly better in the rolling shutter, a LOT sharper and virtually no moire. But to get a similarly equipped R1 would be lots more money too. No really.

    2. I feel like I need to be pedantic here and point out that the difference between something like the F3 and the 5D MkII in the context of video is not necessarily in the quality of the image sensor itself, but in the way that the image is processed, or converted to video. The fundamental reason that the F3’s video quality is superior to the 5D MkII’s is that the F3 is taking its full 12 megapixel frames and down-res-ing them to 1080p, whereas the 5D MkII essentially throws away the data of a large number of its pixels when shooting video, because its current DIGIC processor isn’t powerful enough to down-res 21 megapixel frames to 1080p at 24fps.

      Someone please correct me if I’m wrong, I just felt compelled to be pedantic about this; if the 5D MkII’s sensor was used to its full potential in shooting video, it would rival anything from RED, ARRI, etc. The limiting factor is in fact the image processor, not the sensor itself.

      1. Not true. F3 sensor isn’t 12 MP. It’s only around 3.5 MP. That means each pixel has much larger surface area than 5D Mark II pixels.

  2. Hey guys, thanks for posting this. I think it’s valuable, but I think your blurb above should at least make note of the fact that the F3 implements an optical low pass filter to the 1080p resolution spec. The helps considerably with the moire of the 5D as well.

    I know you guys know this, but I just thought it should be noted here as well.

    So really, it’s an issue of hasty/rough downrezzing/line skipping, but also one of optical low pass filtering.

    Cheers,
    Carey Dissmore

    1. hey Carey Im sure its helping but obviously the F3 is from the ground up designed to get to 1080P without line skipping and pixel binning which im sure is whats responsible for 90% of the improvement here. Regardless of specs or under the hood tech.. what counts is how the final image looks and how it compares. was just interesting to see the two side by side regardless. Yes ones a stills camera and ones a video camera but if they’re both being asked to do the same job
      jas

  3. The F3 looks outstanding, the sample footage wasn’t lying, its a superb full 1080p image. The only disappointment I could see was the rolling shutter, looks to be quite slow compared to what the AF100 seems to be. That would be the next camera to test “as I still don’t believe the AF100 is a real demosaiced and scaled down 1080p image from a 12.1MP sensor”. Of course a RED MX would look even sharper and with even less rolling shutter but this camera will find a lot of fans very soon since it delivers almost what the F35 does for a much more attractive price. Of course until Scarlet ships!

    Beautiful image with a lot of character from the F3, it looks to me way more filmic than the 5D II image for some reason, might be just the detail aside from the Sony color palette.

    Thanks for the test.

    1. Thanks james. yes im sure the Red One would be better again, but the F3 is certainly an improvement in this dept. These tests were also with both cameras on 50th sec / 180 deg shutters and with near matching field of view. Both cameras were bolted together and rolled together so the movement (shown here one after another) is actually the same exactly repeated move. just fyi 🙂

    2. Scarlet S35 is dead. You need to get a crippled Epic now for S35 if you want that sensor size from Red.

  4. I think comparing a camera designed for stills with one designed for motion picture is useless. The F3 was always going to be better because thats what it’s designed for.

    How about comparing the 5D stills with the F3 stills?? Even more useless.

    JW.

    1. In case you missed it. people are using DSLR’s to do video. interviews, TVC’s, shorts, tv series, features every day around the world. Beacuase they want to and often because theres no alternative for cheap, light, portable 35mm imagery. One of the two main issues people have with them is line skipping creating moire and rolling shutter. Since the start of the DSLR revolution people have been begging, nay screaming out for a camera that can give the imagery of say a RED or F35 but with the affordability and portability of a DSLR but without the pain that comes with them. One of these cameras is here now and as we had access to it its our duty to our listeners and viewers to show the improvements these new cameras in the market are bringing and to show that theres another alternative out there. We’re not paid by sony to do or plug these tests and Sony are not a sponsor. Sorry Jell but with all respect its a completely valid test. Not at all useless.

      And the F3 doesn’t do stills.

  5. Great demo. The 5/7d revolution is just about over. It was a great stopgap and the new batch of cameras coming now would never have, had it not been for the 5dII.

  6. Well, this is the same story all over, from day on footage posted from Canon 5D. Yet as Jason said – there is a lot of footage comming from DSLR. Why? Because it has cinematic look, DOF and color. Yet when You search for those – especially shallow DOF it’s easy to pass the line skipping to be so obviously blown into Your face as in this (and other – RED have done similar test long ago) studio test scenes. Yes it’s ugly and it hurts bad 5D. Yet there is only two small steps for Canon in their next DSLR – interpolate the whole frame (that will give quite a sharp image) and eliminate the rolling shutter – they showed a parallel read on their test sensors this summer, so I guess it’s not too far. Yet, how SONY and other would react to 5D Mark II that shoots a crystal clear HD with no rolling shutter for 3-4K? Would Canon dare to oppose all these, and even canabalize it’s own market for video equipment? Who knows. But be happy to shoot with what You have and don’t stop! Seen some Nokia C6 footage recently and found several really good ones 🙂 Best to all and REALLY thanks for the F3 coverage!!!

  7. thanks stas,
    whats clear is stuff is changing and fast. and theres more options every day. options are great but people need to just chill a little, get less fan boyish about gear and just embrace change as as the f3 shows change is good and theres more to come 🙂

  8. Pingback: First F3 footage - Page 2 at DVinfo.net

  9. A Red (or any other 2k or 4k camera) should not be sharper at 1080 than an F3, assuming the F3 is at Nyquist which I’m pretty sure it is. Sure at 4K Red should look sharper than a F3 at 1920×1080, But if you try to keep the 4K sharpness and resolution when you down rez to 1920×1080 you will get aliasing and moire. A decent down rez from 4K to HD should include low pass filtering to bring the resolution down to the appropriate level. Considering that for video production most display and distribution methods are based around and maximum of 1920×1080 then in many cases there is little advantage to 4k. Clearly if you are going to 4k distribution or filmout, 4k may be better. Don’t forget as well that the huge pixel size of the F3 compared to a DSLR brings significant noise, low light and dynamic range advantages.

    Very few people complained about the quality of the live action in Avatar, all shot using Sony HDC1500 and F900R 1920×1080 HD cameras. Resolution is only one part of the image quality equation.

    1. The low pass filtering you use when you downsample digitally is very different from the low pass effect of an optical low pass filter. You can design such a filter so that you get a very sharp image, with negligible aliasing, unlike an optical filter where you can have a sharp image and aliasing, or a softer image and negligible aliasing. That is why there’s a tremendous advantage to capturing a much higher resolution image than you need and employing a strong (some would say optimum) OLPF and then either projecting as-is or downsampling.

      If you only have as many photosites as the target resolution, you have little “room” for the OLPF to work in and hence generally have to have a aliases in the image if you want an MTF at maximum detail that actually allows you to see some detail there. Consequently, the MTF at 2k from a 4k RED image is vastly superior to that of a 1080p imager at 1080p, and has less moire and aliasing, and if you were to downsample to 1080p, you’d still have better MTF and less aliasing and less moire.

  10. Just to repeat a couple of points we have made before:
    1. we love the 5D – we own multiple 5D’s and we do great work with them, but there are some traps and an informed cinematographer is good thing.

    2. We dont think the Sony is perfect, but it is in the particular area we have highlighted – Sony has some great new technical improvements when shooting certain types of video.

    3. the 5D is cheaper and does stills – both excellent reasons for buying them, but there are some difficulties – such as rolling shutter and moire. These are not unknown – so we looked to see how Sony dealt with these two areas.

    4. fxguide and Red Centre are constantly trying to provide technical and creative explanations so people can be better informed. It is true that we worked with Sony and got early access (although no $ changed hands) it is also true we work closely with a lot of other camera companies and post companies. We have offered to work more closely with Canon and yet at our last booked appointment/interview with them (Tim Smith Pro Prod Marketing Manager – he simply never turned up – nor answered any of our emails as to why). That being said we totally love Canon and would welcome any chance to work more closely with them. As I said we use 5D and 7D’s here daily. We have no axe to grind nor do we believe there needs to be one camera to rule them all…. the world is better for competition.

    Mike

  11. I have a dream… a dream of a world Just like on my piano keyboard where black and white live together in perfect harmony.. where all cameras exist happily and equally together and no camera is the natural killer of another.

    Jas

  12. Hi,
    Can you please tell me what lens was on the F3? Was it one of the new Sony primes or something $$$?
    I happened to pause on the chart and it looked really good.
    Thanks
    Chuck

  13. The F3 has not yet been field tested in the real world as yet. We have to
    have filmmakers of all kinds go out and push this camera to it’s limits; give
    it at least six months before we can really determine what this camera can
    or cannot do. The specs are great, but I’ll wait and see. The 5D is a great
    camera, but though it can do video it is not in the same league as
    the F3. But if you feel comfortable with the final results of any camera then
    that is where you probably should be at.

  14. Pingback: F3 vs 5D Mk2 at DVinfo.net

Comments are closed.