noirist

How can the sensor output 6k video (or even 4k video) when it only has 4.44M active pixels? That's more like 1080p-class resolution. The only way I can see it outputting 4k video is if it upscales its 2166 vertical lines to 3180, which can't help but look nasty.

You mean a 2011 camera whose 1080p video footage has lower measured resolution than a 2009 mirrorless camera (Panasonic GH1). And a substantially lower DxO score as well (54 for the V1 vs 64 for the GH1).

Let's be honest here, the Nikon V1 video quality is poor. I've seen footage myself, and it is nasty. The only measurement I've seen is from digitalcamerainfo in 2012, which measured the V1 to have 600 lines of resolution in bright light, and 450 lines in low light. There is no way the V1 is reading the entire sensor for video and downsampling in a reasonable manner.

The Panasonic DMC-LX7 from two years ago does 720p120. Many action cameras have 720p120, including the Panasonic HX-A100, the GoPro Hero 3/3+ Black Edition, JVC XA2, and Sony HDR-As100V.

Matt, in your previous post you said 12800 was "quite usable" and 6400 was "very good". You didn't qualify your claims. When I downloaded the 6400 footage, I was expecting very good footage. But what I got was nasty. I'm glad to see you essentially agree, namely, that the 6400 footage is not usable, just possibly better than other unusable 6400 footage available from similarly priced cameras.

The Gordon Laing review says the D5300 can't change aperture while shooting a movie, even in full manual mode. All you can change is the shutter speed (which you would never want to change) and ISO. Is this true?

Andrew, Did you shoot your 60 fps footage with SteadyShot? From the slashcam.de review and others, it looks like SteadyShot Active Mode uses crop out for roll correction, which is responsible for moire and loss of resolution. But the SteadyShot standard mode is entirely optical, with no moire or loss of resolution.

Thanks for the well-considered review, Andrew, and the great footage. I really enjoyed your use of "found" motion to create a variety interesting shots on the ferris wheel and merry go round.

Has anyone measured the resolution of the GH3 footage, for example, using a test chart? How does the measured resolution of the GH3 footage compare to the GH2 and GH1?

Tzedekh, you are correct. Using the speedbooster changes the lens. It reduces the lens image circle, decreases the lens f-stop, decreases the lens focal length, and increases its field of view. A lens with the speedbooster is optically very different than the same lens without the speedbooster. BurnetRhoades is confused on this and many other points. For example, he seems to (incorrectly) believe that depth-of-field depends on "the size of the opening created by the shutter blades".  When in fact DOF depends only on the aperture of the lens, the lens focal length, sensor size, and the distance from the focal plane to the subject. THE SHUTTER DOES NOT AFFECT DOF IN ANY WAY. Also he does not understand what is an f-stop and how it differs from a t-stop. At any given f-stop setting, adding the speedbooster increases the actual f-stop of the lens at that setting. Yes, after you add the speedbooster, all the f-stop settings on your lens are now incorrectly labeled!

Some of the faces in the videos look very distorted, like the old man with the bushy eyebrows sitting on the right at 1:30, the woman sitting in front of him at 1:36, and the second man walking up the stairs at 2:47.  Is that an out-of-focus bokeh-destroying artifact of the speedbooster or something else? Whatever it is, it's nasty.

Strictly speaking, you are getting a wider and faster lens, that produces an image on a smaller sensor that is equivalent to the image produced by that lens on a larger sensor. If the metabones adapter were optically perfect, the two images would be mathematically indistinguishable. Creating a lens that is both wider and faster is a big deal. Video readout from the smaller sensors is currently more complete (higher resolution, less aliasing) than the video readout from larger sensors, which means the ultimate image coming from the smaller speedboosted sensor will be superior to the ultimate image coming from the larger sensor.

RossF, your reasoning is correct. The only tricky part is that if you use the speedbooster on an aps-c sensor, then the aperture setting shown on a full frame lens will be the full-frame equivalent aperture setting, which is one stop more than how the aperture setting appears to the aps-c sensor.  So if you set a full frame lens to f1.4 without the speedbooster, it looks like f1.4 to the aps-c sensor, and is equivalent to setting the same lens to f2.0 on a full frame sensor. If you set a full frame lens to f1.4 with the speedbooster, it looks like f1.0 to the aps-c sensor, and is equivalent to setting the same lens to f1.4 on a full frame sensor. So when Andrew is comparing pictures taken on the speedboosted FS100 versus 5D Mark III, he needs to set the lens to the same aperture settting to have equivalent pictures.   Andrew's review incorrectly states "It is perhaps fairer to compare the same lens wide open on a full frame camera and stopped down 1 stop on the Speed Booster to match the effective F-stop. For example compare optical performance shooting wide open at F1.4 on the 5D Mark III to shooting stopped down to F2.0 on the FS100. In that situation sharpness is always better in the centre with Speed Booster and almost evenly matched in the corners with the right glass."  The fair comparison is to compare the same lens at the same marked aperture setting on a full frame camera versus a speedboosted aps-c camera.

Andrew,  When you get a chance, take a look at Falk Lumo's discussion of camera equivalence. A 25mm f1.4 lens on a micro 43rds sensor is equivalent to (ie., indistinguishable from) a 33mm f2.0 lens on an APS-C sensor or a 50mm f2.8 lens on a full frame sensor. The nice thing about the speed booster is that it lets you use a full frame lens on an aps-c sensor in a way that makes it equivalent to the same lens on a full frame sensor, but this is only true if you shoot it at the equivalent aperture.  So if you are shooting a 24mm at f1.4 on a full frame sensor, you need to shoot the 24mm lens with speed booster at f1.0 on an aps-c sensor to achieve equivalence.

I agree, Red's notorious production delays are further evidence of the difficulty of producing a digital cinema camera. I didn't explicitly include them in my list because they're no longer considered a "small indie".

These small indies haven't "exposed the true cost of making such a product" until they can actually make such a product and ship it, which they haven't.  Given all the production problems with the BMCC, Digital Bolex, and Kineraw etc cameras have experienced, I think all we can conclude is that it's much more difficult and costly to make a digital cinema camera than they anticipated.

Contrary to your statement, the GoPro Hero3 Black Edition uses line skipping and cropping extensively. Just film a resolution chart at the various settings and you will see how severe the cropping and line skipping is. I'd be surprised if the measured resolution exceeded 500 lines in ANY setting, including the alledged 4k mode.

Many stills photographers have dismissed the Nex7 because there are so few high quality autofocus-capable lenses available for it and none of them are compact. Arguably, there's only one high quality NEX lens, the Zeiss 24mm f1.8, and even that is optically compromised, not that fast, and surprisingly large. The micro four thirds lenses are much better in every way (size, price, speed, optical quality) but the sensors aren't up to speed. Which leaves a lot of people hoping that Fuji can deliver a camera with a great sensor and some great lenses.