13May Canon 5d mark III clean hdmi output testing with difference matte
Here’s the sample clip from yesterdays post. As I mentioned it’s hard to tell the difference between the footage from the Ninja 2 and the Canon 5d mark III. At first glance the footage doesn’t look much different. However the 200% zoom in side by side comparison I posted yesterday doesn’t quite tell the whole story. As many of you mentioned the image from the Ninja 2 most likely contains more color information then the footage from the camera’s internal codec.
I was trying to think of the best way to expose this extra information in the timeline since it’s not easy to see when comparing clips. After doing a little bit of testing in PP CS6 it seemed like a difference matte was the easiest solution. For the test I placed the clips from both the Ninja 2 and the 5d mark III in the timeline and applied the difference matte to the 5d mark III layer. Then set the View option to “Matte only” and the matching tolerance and matching softness 0%.
The result is basically a map of the missing data in the Canon 5d mark III’s internal codec. When viewed in this manner it’s much easier to see what’s really going on. Sure at first glance the footage from the Ninja 2 doesn’t look like a huge step up, but when you take a look at it from this perspective the amount of missing information is pretty amazing. I was not expecting there to be this much missing color information.
If you don’t do a lot of grading in post, you probably won’t be blown away by the difference in image quality the Ninja 2 provides. But when you peel back the layers and take a closer look at the data, there is a definite advantage to the Ninja 2 when paired with the 5d mark III.
Will all this extra color information translate to better image sharpening in post? At the very least, it can’t hurt. If I get a chance, I might spend some time trying to push the footage around a bit in speedgrade and see how well it handles. I’ll post some more info once I’ve had a chance to do some more testing.
I don’t normally have to get this forensic with footage. If anyone knows a better way to analyze the color information let me know and I’ll give it a try.
May 14th, 2013 at 4:23 am
ML RAW footage, brutal
May 14th, 2013 at 12:13 pm
I agree, it’s interesting. But as of right now, I wouldn’t consider it practical. Currently there aren’t enough tests find out what those kinds of data speeds do to heating inside the camera. When Q mode was first introduced a number of people I know burned up the main board on their t2i’s while trying to capture very high data speeds.
If you decide to give this a try, be very careful about heating. It sucks when you burn up a $500 camera, but it hurts a lot more when you fry a $3000 camera.
May 14th, 2013 at 12:57 pm
It does seem too good to be true, but I look forward to seeing it ironed out. If it works, that would be amazing. Too bad you can’t stream it to an external recorder. I guess that would be asking too much from a hack.
May 14th, 2013 at 2:04 pm
The $400 dollar x1000 cards you’ll need to keep up with the data rage might also be a downside.