Rolling Shutter compared, 7D vs 5D vs GH1 vs D90

very informative test! Thanks man! To my untrained eyes the 7D and GH1 seem about the same, standing slightly above the 5D. But my god the D90 is a mess!
 
Good comparison, thanks!
To me it seemed like the 7D edged out the 5d and GH1 by a bit.
The D90 was awful as already noted.
The GH1 - I know dynamic range wasn't the purpose of this test, but the 5D and 7D seemed to have greater DR, or perhaps it was an exposure error on the GH1?
 
haha. special thanks to drk3p. thanks guys. the dynamic range is the worst on the GH1. makes sense, it has the smallest sensor. there is rolling shutter in all of them, esp when you do a very fast whip pan, panning at a normal speed 7D, gh1 and 5D are acceptable. I am just staying away from panning altogether, or, do it very slooow.
 
I was thinking a great test would be to somehow mount or stack all the cams on top of eachother on a tripod head, then you would be making the exact same motions with each ...

Or get fancier and use some further scientific method to measure exact read/reset times and then just publish the numbers...

Thanks for the test regardless... it's cool you had all those cameras there at once!
 
thanks Matthew, yes of course. that's how it should be done. too much work to built a rig but i might do it with the 5d2 and 7d since i will be mostly shooting with those two now
 
Matthew Bennett said:
I was thinking a great test would be to somehow mount or stack all the cams on top of eachother on a tripod head, then you would be making the exact same motions with each ...

Or get fancier and use some further scientific method to measure exact read/reset times and then just publish the numbers...

Thanks for the test regardless... it's cool you had all those cameras there at once!
Click to expand...

Mike Kobal said:
thanks Matthew, yes of course. that's how it should be done. too much work to built a rig but i might do it with the 5d2 and 7d since i will be mostly shooting with those two now
Click to expand...

If its difficult to stack the cams and pan an alternative is to keep the cams on separate tripods and point them all at a vertical lines on a fast moving object while at the same focal length. Maybe a long focal length would be more effective. The doors and windows on the side of a moving elevated train for example.
 
Matthew Bennett said:
I was thinking a great test would be to somehow mount or stack all the cams on top of eachother on a tripod head
Click to expand...

Easiest and most accurate thing would be to set them up in row parallel to a street and then drive a car past at different speeds. If you were feeling very creative you could print out a striped patten and tape it to the side of the of the car. That would get you pretty scientific results.
 
No no, the easiest way would be to give them all out to fellow dvxuser members so we can see for our selves, I say a good 10 year test period would get great data.
 
The easiest method would be to use a Vinyl Turntable with some pattern on it, such as radius-like lines. It would also be the most "cientific", because it would be based on a controlled environment. You could try them all at 45rpm and at 33rpm, at the same distance, and you could even try different lenses. It's simpler because you don't need a special rig of any sort, or expect any regularity from uncontrolled objects, and you don't even need all the cameras at the same time.

On a different note, and thinking about the different speeds, I wonder if there is such a thing as different progressions or accelerations in the rolling shutters i.e. maybe say the 7D has less RS than the GH1 at fast speeds, but the GH1 is a little better with smaller speeds, or vice-versa. The turn table would allow for such a comparison. I don't know if there are any grounds for this supposition, but hypothetically there could be some inconsistency. But of course all of this is mere tech-obsession and the important thing is to get out there and film - preferably things that change the status quo (now that's a real discussion).

In any case, thanks to Mike for such a useful test.
 
Last edited:
You must log in or register to reply here.
Back
Top