rough test comparing HD frames of Red, EX3, and Canon DSLR
Posted 26 March 2010 - 11:31 AM
Here are some cropped portions of a chart I shot at 1600 ASA to test for noise in low-light. The EX3 was shot with a +6db gain boost to match the Red and Canon set to 1600 ASA. I didn't shoot the charts exactly the same size hence the mismath in the cropped size. But I didn't resize the 1920 x 1080 frames, I just cropped them down. They are compressed JPEGS though.
Cropped 1600 ASA Red frame
Cropped +6db EX3 frame
Cropped 1600 ASA Canon frame
Posted 26 March 2010 - 11:56 AM
The Red was converted from 4K HD to 1080P with no sharpening. The EX3 was shot with Detail OFF. The Canon had the Sharpness set one notch above O.
I removed the color from the JPEG just because I found the differences in the color tone of the white backgrounds distracting, but there is a faint amount of color aliasing in the Canon test if you see it in color.
Cropped Red frame
Cropped EX3 frame
Cropped Canon frame
What all of this tells me is that the EX3 is more intercuttable with the Red for an HD show than the Canon 550D T2i is. The only difference is that the Canon would have the same DoF as the Red. But sharpness-wise, not to mention having all the controls of a pro HD camera, the EX3 is a better overall camera than the Canon in video mode. Size & weight & cost, of course, the Canon is the best. I did these tests to tell me what I can get away with when shooting an action scene in terms of slipping in some shots made with these other smaller cameras.
Posted 26 March 2010 - 12:29 PM
Nice job on the off the cuff tests, I'm not really surprised by the results. Speaking of, have you seen that Zacuto test of DSLRS -v- 35mm?
Just a side thought, perhaps another nice side test would be to see how much effort it'd take to get the colors to match on the cameras.. . Take a stock scene, and shoot each camera balanced for it with a typical skin tone, see how they reproduce and then buddy up with a colorist and see how much finagling you need to do? Just a thought.
Posted 26 March 2010 - 01:38 PM
That being said, the Red and Sony do appear to be pretty much a wash for aliasing on the Marconi chart. It's the "C" blocks that get them both. The Canon aliases far more severely. It would be interesting to see those in color, because we should expect that the Bayer sensor of the Red might show us the two different Nyquist limits, green vs. red and blue.
Even more dangerous is to read much into the steps -- but the Red appears not to give us much room if any between the bottom two. The other cameras look like there's plenty to work with for the whole scale.
Posted 26 March 2010 - 01:41 PM
Posted 26 March 2010 - 02:51 PM
Perhaps the Canon DSLRs wouldn't match the sharpness of RED, but it could maybe match 3 perf 35 mm (in terms of sharpness only). Some people who are routinely using the 5D and 7D in professional settings, most notably Shane Hulbut, ASC, hint as one of the main reasons they use them is not necessarily their sharpness --but the softer more film like quality to the image, digital artifacts, picture limitations and tight post options notwithstanding.
Posted 01 April 2010 - 03:57 AM
I did a few tests with wide shots (Canon 15-85mm lens) in deep focus and the aliasing is unacceptable. Tile roofs and brick walls go bonkers - trees are OK. I can reduce it a little in post but not enough.
It's a great camera for stills and time-lapse. If I only use it for stills I'm happy, plus when an improved version comes out I still will be able to use all my Canon and Pentax lenses.
I love the fact that I can download the video directly from the camera with a USB cable - and it is very fast.
Posted 06 April 2010 - 05:47 AM
Seeing that you considered the Canon 550D T2i for sneaking in an action shot or two in a 1080P project, have you considered testing the Panasonic Lumix GH1 as well?